METHODS AND DEVICES FOR CANDIDATE DERIVATION FOR AFFINE MERGE MODE IN VIDEO CODING

CROSS-REFERENCE TO RELATED APPLICATION

[0001] The present application is filed upon and claims priority to U.S. Provisional Application No. 63/358,511, entitled “Methods and Devices for Candidate Derivation for Affine Merge Mode in Video Coding,” filed on July 5, 2022, the entirety of which is incorporated by reference for all purposes.

FIELD

[0002] The present disclosure relates to video coding and compression, and in particular but not limited to, methods and apparatus on improving the affine merge candidate derivation for affine motion prediction mode in a video encoding or decoding process.

BACKGROUND

[0003] Various video coding techniques may be used to compress video data. Video coding is performed according to one or more video coding standards. For example, nowadays, some well-known video coding standards include Versatile Video Coding (WC), High Efficiency Video Coding (HEVC, also known as H.265 or MPEG-H Part2) and Advanced Video Coding (AVC, also known as H.264 or MPEG-4 Part 10), which are jointly developed by ISO/IEC MPEG and ITU-T VECG. AOMedia Video 1 (AVI) was developed by Alliance for Open Media (AOM) as a successor to its preceding standard VP9. Audio Video Coding (AVS), which refers to digital audio and digital video compression standard, is another video compression standard series developed by the Audio and Video Coding Standard Workgroup of China. Most of the existing video coding standards are built upon the famous hybrid video coding framework i.e., using block-based prediction methods (e.g., inter-prediction, intra-prediction) to reduce redundancy present in video images or sequences and using transform coding to compact the energy of the prediction errors. An important goal of video coding techniques is to compress video data into a form that uses a lower bit rate while avoiding or minimizing degradations to video quality.

[0004] The first generation AVS standard includes Chinese national standard “Information Technology, Advanced Audio Video Coding, Part 2: Video” (known as AVS1) and “Information Technology, Advanced Audio Video Coding Part 16: Radio Television Video” (known as AVS+). It can offer around 50% bit-rate saving at the same perceptual quality compared to MPEG-2 standard. The AVS1 standard video part was promulgated as the Chinese national standard in February 2006. The second generation AVS standard includes the series of Chinese national standard “Information Technology, Efficient Multimedia Coding” (knows as AVS2), which is mainly targeted at the transmission of extra HD TV programs. The coding efficiency of the AVS2 is double of that of the AVS+. In May 2016, the AVS2 was issued as the Chinese national standard. Meanwhile, the AVS2 standard video part was submitted by Institute of Electrical and Electronics Engineers (IEEE) as one international standard for applications. The AVS3 standard is one new generation video coding standard for UHD video application aiming at surpassing the coding efficiency of the latest international standard HE VC. In March 2019, at the 68-th AVS meeting, the AVS3-P2 baseline was finished, which provides approximately 30% bit-rate savings over the HE VC standard. Currently, there is one reference software, called high performance model (HPM), is maintained by the AVS group to demonstrate a reference implementation of the AVS3 standard.

SUMMARY

[0005] The present disclosure provides examples of techniques relating to improving the motion vector candidate derivation for motion prediction mode in a video encoding or decoding process.

[0006] According to a first aspect of the present disclosure, there is provided a method of video decoding. In the method of video decoding, a decoder may obtain, based on a storage granularity, a normalized position for an affine coding unit (CU) at a non-adjacent position obtained by scanning area that is not adjacent to a current CU. Additionally, the decoder may obtain determine whether valid affine model information is stored for the normalized position, where the valid affine model information may include a position of the affine CU. Furthermore, the decoder may derive affine model information for the affine CU based on the valid affine model information that is stored for the normalized position in response to determining that the valid affine model information is stored for the normalized position.

[0007] According to a second aspect of the present disclosure, there is provided a method of video encoding. In the method of video encoding, an encoder may obtain, based on a storage granularity, a normalized position for an affine CU at a non-adjacent position obtained by scanning area that is not adjacent to a current CU. Additionally, the encoder may determine whether valid affine model information is stored for the normalized position, where the valid affine model information may include a position of the affine CU. Furthermore, the encoder may derive affine model information for the affine CU based on the valid affine model information that is stored for the normalized position in response to determining that the valid affine model information is stored for the normalized position.

[0008] According to a third aspect of the present disclosure, there is provided a method for video decoding. In the video decoding, a decoder may scan a neighboring area of a current CU to obtain a neighboring affine CU that is previously coded, where the neighboring affine CU may include a plurality of internal sub-blocks. Additionally, the decoder may derive a plurality of first affine model parameters based on the neighboring affine CU. Furthermore, the decoder may derive a plurality of second affine model parameters based on a second regression model, where an input of the second linear regression model may include motion information of a second set of internal sub-blocks and motion information of a plurality of adjacent sub-blocks of the current CU, the second set of internal sub-blocks may be selected from the plurality of internal sub-blocks, and a number of the second set of internal sub-blocks is smaller than or equal to a number of all the plurality of internal sub-block included in the neighboring affine CU, and where the plurality of adjacent sub-blocks are located in an adjacent area that is adjacent to the current CU. Moreover, the decoder may obtain a plurality of CPMVs based on the plurality of first affine model parameters and the plurality of second affine model parameters.

[0009] According to a fourth aspect of the present disclosure, there is provided a method for video encoding. In the method for video encoding, an encoder may scan a neighboring area of a current CU to obtain a neighboring affine CU that is previously coded, where the neighboring affine CU may include a plurality of internal sub-blocks. Additionally, the encoder may derive a plurality of first affine model parameters based on the neighboring affine CU and derive a plurality of second affine model parameters based on a second regression model, where an input of the second linear regression model includes motion information of a second set of internal sub-blocks and motion information of a plurality of adjacent sub-blocks of the current CU, the second set of internal sub-blocks may be selected from the plurality of internal sub-blocks, and a number of the second set of internal sub-blocks is smaller than or equal to a number of all the plurality of internal sub-block included in the neighboring affine CU, and the plurality of adjacent sub-blocks are located in an adjacent area that is adjacent to the current CU. Furthermore, the encoder may obtain a plurality of CPMVs based on the plurality of first affine model parameters and the plurality of second affine model parameters. [0010] According to a fifth aspect of the present disclosure, there is provided an apparatus for video decoding. The apparatus includes one or more processors and a memory coupled to the one or more processors and configured to store instructions executable by the one or more processors. Furthermore, the one or more processors, upon execution of the instructions, are configured to perform the method according to the first or third aspect above.

[0011] According to a sixth aspect of the present disclosure, there is provided an apparatus for video encoding. The apparatus includes one or more processors and a memory coupled to the one or more processors and configured to store instructions executable by the one or more processors. Furthermore, the one or more processors, upon execution of the instructions, are configured to perform the method according to the second or fourth aspect above. [0012] According to a seventh aspect of the present disclosure, there is provided a non- transitory computer-readable storage medium storing computer-executable instructions that, when executed by one or more computer processors, cause the one or more computer processors to receive a bitstream, and perform the method according to the first or third aspect above.

[0013] According to an eight aspect of the present disclosure, there is provided a non- transitory computer-readable storage medium storing computer-executable instructions that, when executed by one or more computer processors, cause the one or more computer processors to perform the method according to the second or fourth aspect above to encode a current CU into a bitstream, and transmit the bitstream.

[0014] According to a ninth aspect of the present disclosure, there is provided a non-transitory computer-readable storage medium for storing a bitstream to be decoded by the method of any of the first or third aspect above.

[0015] According to a tenth aspect of the present disclosure, there is provided a non-transitory computer-readable storage medium storing a bitstream generated by the method of any of the second or fourth aspect above.

BRIEF DESCRIPTION OF THE DRAWINGS

[0016] A more particular description of the examples of the present disclosure will be rendered by reference to specific examples illustrated in the appended drawings. Given that these drawings depict only some examples and are not therefore considered to be limiting in scope, the examples will be described and explained with additional specificity and details through the use of the accompanying drawings.

[0017] FIG. 1 A is a block diagram illustrating a system for encoding and decoding video blocks in accordance with some examples of the present disclosure.

[0018] FIG. IB is a block diagram of an encoder in accordance with some examples of the present disclosure. [0019] FIGS. 1C-1F are block diagrams illustrating how a frame is recursively partitioned into multiple video blocks of different sizes and shapes in accordance with some examples of the present disclosure.

[0020] FIG. 1G is a block diagram illustrating an exemplary video encoder in accordance with some examples of the present disclosure

[0021] FIG. 2A is a block diagram of a decoder in accordance with some examples of the present disclosure.

[0022] FIG. 2B is a block diagram illustrating an exemplary video decoder in accordance with some examples of the present disclosure

[0023] FIG. 3A is a diagram illustrating block partitions in a multi-type tree structure in accordance with some examples of the present disclosure.

[0024] FIG. 3B is a diagram illustrating block partitions in a multi-type tree structure in accordance with some examples of the present disclosure.

[0025] FIG. 3C is a diagram illustrating block partitions in a multi-type tree structure in accordance with some examples of the present disclosure.

[0026] FIG. 3D is a diagram illustrating block partitions in a multi-type tree structure in accordance with some examples of the present disclosure.

[0027] FIG. 3E is a diagram illustrating block partitions in a multi-type tree structure in accordance with some examples of the present disclosure.

[0028] FIG. 4A illustrates 4-parameter affine model in accordance with some examples of the present disclosure.

[0029] FIG. 4B illustrates 4-parameter affine model in accordance with some examples of the present disclosure.

[0030] FIG. 4C illustrates positions of spatial merge candidates in accordance with some examples of the present disclosure.

[0031] FIG. 4D illustrates candidate pairs that are considered for redundancy check of spatial merge candidates in accordance with some examples of the present disclosure. [0032] FIG. 4E illustrates motion vector scaling for temporal merge candidates in accordance with some examples of the present disclosure.

[0033] FIG. 4F illustrates candidate positions for temporal merge candidates Co and Ci in accordance with some examples of the present disclosure.

[0034] FIG. 5 illustrates 6-parameter affine model in accordance with some examples of the present disclosure.

[0035] FIG. 6 illustrates adjacent neighboring blocks for inherited affine merge candidates in accordance with some examples of the present disclosure.

[0036] FIG. 7 illustrates adjacent neighboring blocks for constructed affine merge candidates in accordance with some examples of the present disclosure.

[0037] FIG. 8 illustrates non-adjacent neighboring blocks for inherited affine merge candidates in accordance with some examples of the present disclosure.

[0038] FIG. 9 illustrates derivation of constructed affine merge candidates using non-adjacent neighboring blocks in accordance with some examples of the present disclosure.

[0039] FIG. 10 illustrates perpendicular scanning of non-adjacent neighboring blocks in accordance with some examples of the present disclosure.

[0040] FIG. 11 illustrates parallel scanning of non-adjacent neighboring blocks in accordance with some examples of the present disclosure.

[0041] FIG. 12 illustrates combined perpendicular and parallel scanning of non-adjacent neighboring blocks in accordance with some examples of the present disclosure.

[0042] FIG. 13A illustrates neighbor blocks with the same size as the current block in accordance with some examples of the present disclosure.

[0043] FIG. 13B illustrates neighbor blocks with a different size than the current block in accordance with some examples of the present disclosure.

[0044] FIG. 14A illustrates an example of the bottom-left or top-right block of the bottommost or rightmost block in a previous distance is used as the bottommost or rightmost block of a current distance in accordance with some examples of the present disclosure. [0045] FIG. 14B illustrates an example of the left or top block of the bottommost or rightmost block in the previous distance is used as the bottommost or rightmost block of the current distance in accordance with some examples of the present disclosure.

[0046] FIG. 15A illustrates scanning positions at bottom-left and top-right positions used for above and left non-adjacent neighboring blocks in accordance with some examples of the present disclosure.

[0047] FIG. 15B illustrates scanning positions at bottom-right positions used for both above and left non-adjacent neighboring blocks in accordance with some examples of the present disclosure.

[0048] FIG. 15C illustrates scanning positions at bottom-left positions used for both above and left non-adjacent neighboring blocks in accordance with some examples of the present disclosure.

[0049] FIG. 15D illustrates scanning positions at top-right positions used for both above and left non-adjacent neighboring blocks in accordance with some examples of the present disclosure.

[0050] FIG. 16 illustrates a simplified scanning process for deriving constructed merge candidates in accordance with some examples of the present disclosure.

[0051] FIG. 17A illustrates spatial neighbors for deriving inherited affine merge candidates in accordance with some examples of the present disclosure.

[0052] FIG. 17B illustrates spatial neighbors for deriving constructed affine merge candidates in accordance with some examples of the present disclosure.

[0053] FIG. 18 illustrates an example of inheritance based derivation method for deriving affine constructed candidates in accordance with some examples of the present disclosure.

[0054] FIG. 19 illustrates template and reference samples of a template in reference list 0 and reference list 1 in accordance with some examples of the present disclosure. [0055] FIG. 20 illustrates template and reference samples of a template for block with subblock motion using the motion information of the subblocks of a current block in accordance with some examples of the present disclosure.

[0056] FIG. 21 illustrates an example where non-adjacent spatial area is restricted to be within half CTU size on the area above and left of the current CTU in accordance with some examples of the present disclosure.

[0057] FIG. 22A illustrates one storage method of directly saving affine motion information about an affined coded block CTU in accordance with some examples of the present disclosure. [0058] FIG. 22B illustrates one storage method of projecting and saving affine motion information at each sub-block in accordance with some examples of the present disclosure.

[0059] FIG. 23 illustrates an example of using center point to derive regular/translational motion at each 4x4 regular block in accordance with some examples of the present disclosure. [0060] FIG. 24 is a diagram illustrating a computing environment coupled with a user interface in accordance with some examples of the present disclosure.

[0061] FIG. 25 illustrates an example of storing motion information of an affine-coded block at a granularity greater than the minimum affine block size in accordance with some examples of the present disclosure.

[0062] FIG. 26 illustrates merge mode with MVD (MMVD) search points respectively in L0 reference and LI reference in accordance with some examples of the present disclosure.

[0063] FIG. 27A illustrates an example of motion storage for non-adjacent spatial neighbors including affine neighbor CUs and non-affine neighbor CUs when allowable non-adjacent spatial area beyond a current coding tree unit (CTU) in accordance with some examples of the present disclosure.

[0064] FIG. 27B illustrates an example of motion storage for non-adjacent spatial neighbors including affine neighbor CUs and non-affine neighbor CUs motion storage in line buffer in accordance with some examples of the present disclosure. [0065] FIG. 28 illustrates an example of projected or clipped non-adjacent neighbor positions when a scanned non-adjacent neighbor position is beyond the allowable spatial area in accordance with some examples of the present disclosure.

[0066] FIG. 29 illustrates another example of projected/clipped non-adjacent neighbor position if the scanned non-adjacent neighbor position is beyond the allowed spatial area, e.g., beyond current CTU and available line buffer, in accordance with some examples of the present disclosure.

[0067] FIGS. 30A-30C illustrate an access of non-adjacent affine motion information from existing storage of translational motion information in accordance with some examples of the present disclosure.

[0068] FIG. 31A-31C illustrate an access of non-adjacent affine motion information from existing storage of translational motion information by using fewer stored information than FIGS. 30A-30C in accordance with some examples of the present disclosure.

[0069] FIG. 32 illustrate neighboring 4 x 4 subblocks that are used for Regression based Motion Vector Field (RMVF) parameter derivation in accordance with some examples of the present disclosure.

[0070] FIG. 33 is a flow chart illustrating a method for video decoding in accordance with some examples of the present disclosure.

[0071] FIG. 34 is a flow chart illustrating a method for video encoding corresponding to the method for video decoding as shown in FIG. 33 in accordance with some examples of the present disclosure.

[0072] FIG. 35 is a flow chart illustrating a method for video decoding in accordance with some examples of the present disclosure.

[0073] FIG. 36 is a flow chart illustrating a method for video encoding corresponding to the method for video decoding as shown in FIG. 35 in accordance with some examples of the present disclosure.

DETAILED DESCRIPTION [0074] Reference will now be made in detail to specific implementations, examples of which are illustrated in the accompanying drawings. In the following detailed description, numerous non-limiting specific details are set forth in order to assist in understanding the subject matter presented herein. But it will be apparent to one of ordinary skill in the art that various alternatives may be used. For example, it will be apparent to one of ordinary skill in the art that the subject matter presented herein can be implemented on many types of electronic devices with digital video capabilities.

[0075] Terms used in the disclosure are only adopted for the purpose of describing specific embodiments and not intended to limit the disclosure. “A/an,” “said,” and “the” in a singular form in the disclosure and the appended claims are also intended to include a plural form, unless other meanings are clearly denoted throughout the disclosure. It is also to be understood that term “and/or” used in the disclosure refers to and includes one or any or all possible combinations of multiple associated items that are listed.

[0076] Reference throughout this specification to “one embodiment,” “an embodiment,” “an example,” “some embodiments,” “some examples,” or similar language means that a particular feature, structure, or characteristic described is included in at least one embodiment or example. Features, structures, elements, or characteristics described in connection with one or some embodiments are also applicable to other embodiments, unless expressly specified otherwise.

[0077] Throughout the disclosure, the terms “first,” “second,” “third,” etc. are all used as nomenclature only for references to relevant elements, e.g., devices, components, compositions, steps, etc., without implying any spatial or chronological orders, unless expressly specified otherwise. For example, a “first device” and a “second device” may refer to two separately formed devices, or two parts, components, or operational states of a same device, and may be named arbitrarily.

[0078] The terms “module,” “sub-module,” “circuit,” “sub-circuit,” “circuitry,” “sub-circuitry,” “unit,” or “sub-unit” may include memory (shared, dedicated, or group) that stores code or instructions that can be executed by one or more processors. A module may include one or more circuits with or without stored code or instructions. The module or circuit may include one or more components that are directly or indirectly connected. These components may or may not be physically attached to, or located adjacent to, one another.

[0079] As used herein, the term “if’ or “when” may be understood to mean “upon” or “in response to” depending on the context. These terms, if appear in a claim, may not indicate that the relevant limitations or features are conditional or optional. For example, a method may comprise steps of: i) when or if condition X is present, function or action X’ is performed, and ii) when or if condition Y is present, function or action Y’ is performed. The method may be implemented with both the capability of performing function or action X’, and the capability of performing function or action Y’. Thus, the functions X’ and Y’ may both be performed, at different times, on multiple executions of the method.

[0080] A unit or module may be implemented purely by software, purely by hardware, or by a combination of hardware and software. In a pure software implementation, for example, the unit or module may include functionally related code blocks or software components, that are directly or indirectly linked together, so as to perform a particular function.

[0081] FIG. 1A is a block diagram illustrating an exemplary system 10 for encoding and decoding video blocks in parallel in accordance with some implementations of the present disclosure. As shown in FIG. 1A, the system 10 includes a source device 12 that generates and encodes video data to be decoded at a later time by a destination device 14. The source device 12 and the destination device 14 may include any of a wide variety of electronic devices, including desktop or laptop computers, tablet computers, smart phones, set-top boxes, digital televisions, cameras, display devices, digital media players, video gaming consoles, video streaming device, or the like. In some implementations, the source device 12 and the destination device 14 are equipped with wireless communication capabilities.

[0082] In some implementations, the destination device 14 may receive the encoded video data to be decoded via a link 16. The link 16 may include any type of communication medium or device capable of moving the encoded video data from the source device 12 to the destination device 14. In one example, the link 16 may include a communication medium to enable the source device 12 to transmit the encoded video data directly to the destination device 14 in real time. The encoded video data may be modulated according to a communication standard, such as a wireless communication protocol, and transmitted to the destination device 14. The communication medium may include any wireless or wired communication medium, such as a Radio Frequency (RF) spectrum or one or more physical transmission lines. The communication medium may form part of a packet-based network, such as a local area network, a wide-area network, or a global network such as the Internet. The communication medium may include routers, switches, base stations, or any other equipment that may be useful to facilitate communication from the source device 12 to the destination device 14.

[0083] In some other implementations, the encoded video data may be transmitted from an output interface 22 to a storage device 32. Subsequently, the encoded video data in the storage device 32 may be accessed by the destination device 14 via an input interface 28. The storage device 32 may include any of a variety of distributed or locally accessed data storage media such as a hard drive, Blu-ray discs, Digital Versatile Disks (DVDs), Compact Disc Read-Only Memories (CD-ROMs), flash memory, volatile or non-volatile memory, or any other suitable digital storage media for storing the encoded video data. In a further example, the storage device 32 may correspond to a file server or another intermediate storage device that may hold the encoded video data generated by the source device 12. The destination device 14 may access the stored video data from the storage device 32 via streaming or downloading. The file server may be any type of computer capable of storing the encoded video data and transmitting the encoded video data to the destination device 14. Exemplary file servers include a web server (e.g., for a website), a File Transfer Protocol (FTP) server, Network Attached Storage (NAS) devices, or a local disk drive. The destination device 14 may access the encoded video data through any standard data connection, including a wireless channel (e.g., a Wireless Fidelity (Wi-Fi) connection), a wired connection (e.g., Digital Subscriber Line (DSL), cable modem, etc.), or a combination of both that is suitable for accessing encoded video data stored on a file server. The transmission of the encoded video data from the storage device 32 may be a streaming transmission, a download transmission, or a combination of both.

[0084] As shown in FIG. 1A, the source device 12 includes a video source 18, a video encoder 20 and the output interface 22. The video source 18 may include a source such as a video capturing device, e.g., a video camera, a video archive containing previously captured video, a video feeding interface to receive video from a video content provider, and/or a computer graphics system for generating computer graphics data as the source video, or a combination of such sources. As one example, if the video source 18 is a video camera of a security surveillance system, the source device 12 and the destination device 14 may form camera phones or video phones. However, the implementations described in the present application may be applicable to video coding in general, and may be applied to wireless and/or wired applications.

[0085] The captured, pre-captured, or computer-generated video may be encoded by the video encoder 20. The encoded video data may be transmitted directly to the destination device 14 via the output interface 22 of the source device 12. The encoded video data may also (or alternatively) be stored onto the storage device 32 for later access by the destination device 14 or other devices, for decoding and/or playback. The output interface 22 may further include a modem and/or a transmitter.

[0086] The destination device 14 includes the input interface 28, a video decoder 30, and a display device 34. The input interface 28 may include a receiver and/or a modem and receive the encoded video data over the link 16. The encoded video data communicated over the link 16, or provided on the storage device 32, may include a variety of syntax elements generated by the video encoder 20 for use by the video decoder 30 in decoding the video data. Such syntax elements may be included within the encoded video data transmitted on a communication medium, stored on a storage medium, or stored on a file server. [0087] In some implementations, the destination device 14 may include the display device 34, which can be an integrated display device and an external display device that is configured to communicate with the destination device 14. The display device 34 displays the decoded video data to a user, and may include any of a variety of display devices such as a Liquid Crystal Display (LCD), a plasma display, an Organic Light Emitting Diode (OLED) display, or another type of display device.

[0088] The video encoder 20 and the video decoder 30 may operate according to proprietary or industry standards, such as WC, HEVC, MPEG-4, Part 10, AVC, or extensions of such standards. It should be understood that the present application is not limited to a specific video encoding/decoding standard and may be applicable to other video encoding/decoding standards. It is generally contemplated that the video encoder 20 of the source device 12 may be configured to encode video data according to any of these current or future standards. Similarly, it is also generally contemplated that the video decoder 30 of the destination device 14 may be configured to decode video data according to any of these current or future standards.

[0089] The video encoder 20 and the video decoder 30 each may be implemented as any of a variety of suitable encoder and/or decoder circuitry, such as one or more microprocessors, Digital Signal Processors (DSPs), Application Specific Integrated Circuits (ASICs), Field Programmable Gate Arrays (FPGAs), discrete logic, software, hardware, firmware or any combinations thereof. When implemented partially in software, an electronic device may store instructions for the software in a suitable, non-transitory computer-readable medium and execute the instructions in hardware using one or more processors to perform the video encoding/decoding operations disclosed in the present disclosure. Each of the video encoder 20 and the video decoder 30 may be included in one or more encoders or decoders, either of which may be integrated as part of a combined encoder/decoder (CODEC) in a respective device. [0090] Like HEVC, WC is built upon the block-based hybrid video coding framework. FIG. IB is a block diagram illustrating a block-based video encoder in accordance with some implementations of the present disclosure. In the encoder 100, the input video signal is processed block by block, called coding units (CUs). The encoder 100 may be the video encoder 20 as shown in FIG. 1A. In VTM-1.0, a CU can be up to 128x128 pixels. However, different from the HEVC which partitions blocks only based on quad-trees, in WC, one coding tree unit (CTU) is split into CUs to adapt to varying local characteristics based on quad/binary/ternary-tree. Additionally, the concept of multiple partition unit type in the HEVC is removed, i.e., the separation of CU, prediction unit (PU) and transform unit (TU) does not exist in the WC anymore; instead, each CU is always used as the basic unit for both prediction and transform without further partitions. In the multi-type tree structure, one CTU is firstly partitioned by a quad-tree structure. Then, each quad-tree leaf node can be further partitioned by a binary and ternary tree structure.

[0091] FIGS. 3A-3E are schematic diagrams illustrating multi-type tree splitting modes in accordance with some implementations of the present disclosure. FIGS. 3A-3E respectively show five splitting types including quaternary partitioning (FIG. 3A), vertical binary partitioning (FIG. 3B), horizontal binary partitioning (FIG. 3C), vertical extended ternary partitioning (FIG. 3D), and horizontal extended ternary partitioning (FIG. 3E).

[0092] For each given video block, spatial prediction and/or temporal prediction may be performed. Spatial prediction (or “intra prediction”) uses pixels from the samples of already coded neighboring blocks (which are called reference samples) in the same video picture/slice to predict the current video block. Spatial prediction reduces spatial redundancy inherent in the video signal. Temporal prediction (also referred to as “inter prediction” or “motion compensated prediction”) uses reconstructed pixels from the already coded video pictures to predict the current video block. Temporal prediction reduces temporal redundancy inherent in the video signal. Temporal prediction signal for a given CU is usually signaled by one or more motion vectors (MVs) which indicate the amount and the direction of motion between the current CU and its temporal reference. Also, if multiple reference pictures are supported, one reference picture index is additionally sent, which is used to identify from which reference picture in the reference picture store the temporal prediction signal comes.

[0093] After spatial and/or temporal prediction, an intra/inter mode decision circuitry 121 in the encoder 100 chooses the best prediction mode, for example based on the rate-distortion optimization method. The block predictor 120 is then subtracted from the current video block; and the resulting prediction residual is de-correlated using the transform circuitry 102 and the quantization circuitry 104. The resulting quantized residual coefficients are inverse quantized by the inverse quantization circuitry 116 and inverse transformed by the inverse transform circuitry 118 to form the reconstructed residual, which is then added back to the prediction block to form the reconstructed signal of the CU. Further, in-loop filtering 115, such as a deblocking filter, a sample adaptive offset (SAO), and/or an adaptive in-loop filter (ALF) may be applied on the reconstructed CU before it is put in the reference picture store of the picture buffer 117 and used to code future video blocks. To form the output video bitstream 114, coding mode (inter or intra), prediction mode information, motion information, and quantized residual coefficients are all sent to the entropy coding unit 106 to be further compressed and packed to form the bit-stream.

[0094] For example, a deblocking filter is available in AVC, HEVC as well as the now-current version of WC. In HEVC, an additional in-loop filter called SAO is defined to further improve coding efficiency. In the now-current version of the WC standard, yet another in-loop filter called ALF is being actively investigated, and it has a good chance of being included in the final standard.

[0095] These in-loop filter operations are optional. Performing these operations helps to improve coding efficiency and visual quality. They may also be turned off as a decision rendered by the encoder 100 to save computational complexity. [0096] It should be noted that intra prediction is usually based on unfiltered reconstructed pixels, while inter prediction is based on filtered reconstructed pixels if these filter options are turned on by the encoder 100.

[0097] FIG. 2A is a block diagram illustrating a block-based video decoder 200 which may be used in conjunction with many video coding standards. This decoder 200 is similar to the reconstruction-related section residing in the encoder 100 of FIG. IB. The block-based video decoder 200 may be the video decoder 30 as shown in FIG. 1 A. In the decoder 200, an incoming video bitstream 201 is first decoded through an Entropy Decoding 202 to derive quantized coefficient levels and prediction-related information. The quantized coefficient levels are then processed through an Inverse Quantization 204 and an Inverse Transform 206 to obtain a reconstructed prediction residual. A block predictor mechanism, implemented in an Intra/inter Mode Selector 212, is configured to perform either an Intra Prediction 208, or a Motion Compensation 210, based on decoded prediction information. A set of unfiltered reconstructed pixels are obtained by summing up the reconstructed prediction residual from the Inverse Transform 206 and a predictive output generated by the block predictor mechanism, using a summer 214.

[0098] The reconstructed block may further go through an In-Loop Filter 209 before it is stored in a Picture Buffer 213 which functions as a reference picture store. The reconstructed video in the Picture Buffer 213 may be sent to drive a display device, as well as used to predict future video blocks. In situations where the In-Loop Filter 209 is turned on, a filtering operation is performed on these reconstructed pixels to derive a final reconstructed Video Output 222.

[0099] FIG. 1G is a block diagram illustrating another exemplary video encoder 20 in accordance with some implementations described in the present application. The video encoder 20 may perform intra and inter predictive coding of video blocks within video frames. Intra predictive coding relies on spatial prediction to reduce or remove spatial redundancy in video data within a given video frame or picture. Inter predictive coding relies on temporal prediction to reduce or remove temporal redundancy in video data within adjacent video frames or pictures of a video sequence. It should be noted that the term “frame” may be used as synonyms for the term “image” or “picture” in the field of video coding.

[00100] As shown in FIG. 1G, the video encoder 20 includes a video data memory 40, a prediction processing unit 41, a Decoded Picture Buffer (DPB) 64, a summer 50, a transform processing unit 52, a quantization unit 54, and an entropy encoding unit 56. The prediction processing unit 41 further includes a motion estimation unit 42, a motion compensation unit 44, a partition unit 45, an intra prediction processing unit 46, and an intra Block Copy (BC) unit 48. In some implementations, the video encoder 20 also includes an inverse quantization unit 58, an inverse transform processing unit 60, and a summer 62 for video block reconstruction. An in-loop filter 63, such as a deblocking filter, may be positioned between the summer 62 and the DPB 64 to filter block boundaries to remove blockiness artifacts from reconstructed video. Another in-loop filter, such as Sample Adaptive Offset (SAO) filter and/or Adaptive in-Loop Filter (ALF), may also be used in addition to the deblocking filter to filter an output of the summer 62. In some examples, the in-loop filters may be omitted, and the decoded video block may be directly provided by the summer 62 to the DPB 64. The video encoder 20 may take the form of a fixed or programmable hardware unit or may be divided among one or more of the illustrated fixed or programmable hardware units.

[00101] The video data memory 40 may store video data to be encoded by the components of the video encoder 20. The video data in the video data memory 40 may be obtained, for example, from the video source 18 as shown in FIG. 1 A. The DPB 64 is a buffer that stores reference video data (for example, reference frames or pictures) for use in encoding video data by the video encoder 20 (e.g., in intra or inter predictive coding modes). The video data memory 40 and the DPB 64 may be formed by any of a variety of memory devices. In various examples, the video data memory 40 may be on-chip with other components of the video encoder 20, or off-chip relative to those components. [00102] As shown in FIG. 1G, after receiving the video data, the partition unit 45 within the prediction processing unit 41 partitions the video data into video blocks. This partitioning may also include partitioning a video frame into slices, tiles (for example, sets of video blocks), or other larger Coding Units (CUs) according to predefined splitting structures such as a Quad- Tree (QT) structure associated with the video data. The video frame is or may be regarded as a two-dimensional array or matrix of samples with sample values. A sample in the array may also be referred to as a pixel or a pel. A number of samples in horizontal and vertical directions (or axes) of the array or picture define a size and/or a resolution of the video frame. The video frame may be divided into multiple video blocks by, for example, using QT partitioning. The video block again is or may be regarded as a two-dimensional array or matrix of samples with sample values, although of smaller dimension than the video frame. A number of samples in horizontal and vertical directions (or axes) of the video block define a size of the video block. The video block may further be partitioned into one or more block partitions or sub-blocks (which may form again blocks) by, for example, iteratively using QT partitioning, Binary-Tree (BT) partitioning or Triple-Tree (TT) partitioning or any combination thereof. It should be noted that the term “block” or “video block” as used herein may be a portion, in particular a rectangular (square or non- square) portion, of a frame or a picture. With reference, for example, to HEVC and WC, the block or video block may be or correspond to a Coding Tree Unit (CTU), a CU, a Prediction Unit (PU) or a Transform Unit (TU) and/or may be or correspond to a corresponding block, e.g., a Coding Tree Block (CTB), a Coding Block (CB), a Prediction Block (PB) or a Transform Block (TB) and/or to a sub-block.

[00103] The prediction processing unit 41 may select one of a plurality of possible predictive coding modes, such as one of a plurality of intra predictive coding modes or one of a plurality of inter predictive coding modes, for the current video block based on error results (e.g., coding rate and the level of distortion). The prediction processing unit 41 may provide the resulting intra or inter prediction coded block to the summer 50 to generate a residual block and to the summer 62 to reconstruct the encoded block for use as part of a reference frame subsequently. The prediction processing unit 41 also provides syntax elements, such as motion vectors, intra-mode indicators, partition information, and other such syntax information, to the entropy encoding unit 56.

[00104] In order to select an appropriate intra predictive coding mode for the current video block, the intra prediction processing unit 46 within the prediction processing unit 41 may perform intra predictive coding of the current video block relative to one or more neighbor blocks in the same frame as the current block to be coded to provide spatial prediction. The motion estimation unit 42 and the motion compensation unit 44 within the prediction processing unit 41 perform inter predictive coding of the current video block relative to one or more predictive blocks in one or more reference frames to provide temporal prediction. The video encoder 20 may perform multiple coding passes, e.g., to select an appropriate coding mode for each block of video data.

[00105] In some implementations, the motion estimation unit 42 determines the inter prediction mode for a current video frame by generating a motion vector, which indicates the displacement of a video block within the current video frame relative to a predictive block within a reference video frame, according to a predetermined pattern within a sequence of video frames. Motion estimation, performed by the motion estimation unit 42, is the process of generating motion vectors, which estimate motion for video blocks. A motion vector, for example, may indicate the displacement of a video block within a current video frame or picture relative to a predictive block within a reference frame relative to the current block being coded within the current frame. The predetermined pattern may designate video frames in the sequence as P frames or B frames. The intra BC unit 48 may determine vectors, e.g., block vectors, for intra BC coding in a manner similar to the determination of motion vectors by the motion estimation unit 42 for inter prediction, or may utilize the motion estimation unit 42 to determine the block vector.

[00106] A predictive block for the video block may be or may correspond to a block or a reference block of a reference frame that is deemed as closely matching the video block to be coded in terms of pixel difference, which may be determined by Sum of Absolute Difference (SAD), Sum of Square Difference (SSD), or other difference metrics. In some implementations, the video encoder 20 may calculate values for sub-integer pixel positions of reference frames stored in the DPB 64. For example, the video encoder 20 may interpolate values of one-quarter pixel positions, one-eighth pixel positions, or other fractional pixel positions of the reference frame. Therefore, the motion estimation unit 42 may perform a motion search relative to the full pixel positions and fractional pixel positions and output a motion vector with fractional pixel precision.

[00107] The motion estimation unit 42 calculates a motion vector for a video block in an inter prediction coded frame by comparing the position of the video block to the position of a predictive block of a reference frame selected from a first reference frame list (List 0) or a second reference frame list (List 1), each of which identifies one or more reference frames stored in the DPB 64. The motion estimation unit 42 sends the calculated motion vector to the motion compensation unit 44 and then to the entropy encoding unit 56.

[00108] Motion compensation, performed by the motion compensation unit 44, may involve fetching or generating the predictive block based on the motion vector determined by the motion estimation unit 42. Upon receiving the motion vector for the current video block, the motion compensation unit 44 may locate a predictive block to which the motion vector points in one of the reference frame lists, retrieve the predictive block from the DPB 64, and forward the predictive block to the summer 50. The summer 50 then forms a residual video block of pixel difference values by subtracting pixel values of the predictive block provided by the motion compensation unit 44 from the pixel values of the current video block being coded. The pixel difference values forming the residual video block may include luma or chroma component differences or both. The motion compensation unit 44 may also generate syntax elements associated with the video blocks of a video frame for use by the video decoder 30 in decoding the video blocks of the video frame. The syntax elements may include, for example, syntax elements defining the motion vector used to identify the predictive block, any flags indicating the prediction mode, or any other syntax information described herein. Note that the motion estimation unit 42 and the motion compensation unit 44 may be highly integrated, but are illustrated separately for conceptual purposes.

[00109] In some implementations, the intra BC unit 48 may generate vectors and fetch predictive blocks in a manner similar to that described above in connection with the motion estimation unit 42 and the motion compensation unit 44, but with the predictive blocks being in the same frame as the current block being coded and with the vectors being referred to as block vectors as opposed to motion vectors. In particular, the intra BC unit 48 may determine an intra-prediction mode to use to encode a current block. In some examples, the intra BC unit 48 may encode a current block using various intra-prediction modes, e.g., during separate encoding passes, and test their performance through rate-distortion analysis. Next, the intra BC unit 48 may select, among the various tested intra-prediction modes, an appropriate intraprediction mode to use and generate an intra-mode indicator accordingly. For example, the intra BC unit 48 may calculate rate-distortion values using a rate-distortion analysis for the various tested intra-prediction modes, and select the intra-prediction mode having the best rate-distortion characteristics among the tested modes as the appropriate intra-prediction mode to use. Rate-distortion analysis generally determines an amount of distortion (or error) between an encoded block and an original, unencoded block that was encoded to produce the encoded block, as well as a bitrate (i.e., a number of bits) used to produce the encoded block. Intra BC unit 48 may calculate ratios from the distortions and rates for the various encoded blocks to determine which intra-prediction mode exhibits the best rate-distortion value for the block.

[00110] In other examples, the intra BC unit 48 may use the motion estimation unit 42 and the motion compensation unit 44, in whole or in part, to perform such functions for Intra BC prediction according to the implementations described herein. In either case, for Intra block copy, a predictive block may be a block that is deemed as closely matching the block to be coded, in terms of pixel difference, which may be determined by SAD, SSD, or other difference metrics, and identification of the predictive block may include calculation of values for sub-integer pixel positions.

[00111] Whether the predictive block is from the same frame according to intra prediction, or a different frame according to inter prediction, the video encoder 20 may form a residual video block by subtracting pixel values of the predictive block from the pixel values of the current video block being coded, forming pixel difference values. The pixel difference values forming the residual video block may include both luma and chroma component differences.

[00112] The intra prediction processing unit 46 may intra-predict a current video block, as an alternative to the inter-prediction performed by the motion estimation unit 42 and the motion compensation unit 44, or the intra block copy prediction performed by the intra BC unit 48, as described above. In particular, the intra prediction processing unit 46 may determine an intra prediction mode to use to encode a current block. To do so, the intra prediction processing unit 46 may encode a current block using various intra prediction modes, e.g., during separate encoding passes, and the intra prediction processing unit 46 (or a mode selection unit, in some examples) may select an appropriate intra prediction mode to use from the tested intra prediction modes. The intra prediction processing unit 46 may provide information indicative of the selected intra-prediction mode for the block to the entropy encoding unit 56. The entropy encoding unit 56 may encode the information indicating the selected intra-prediction mode in the bitstream.

[00113] After the prediction processing unit 41 determines the predictive block for the current video block via either inter prediction or intra prediction, the summer 50 forms a residual video block by subtracting the predictive block from the current video block. The residual video data in the residual block may be included in one or more TUs and is provided to the transform processing unit 52. The transform processing unit 52 transforms the residual video data into residual transform coefficients using a transform, such as a Discrete Cosine Transform (DCT) or a conceptually similar transform. [00114] The transform processing unit 52 may send the resulting transform coefficients to the quantization unit 54. The quantization unit 54 quantizes the transform coefficients to further reduce the bit rate. The quantization process may also reduce the bit depth associated with some or all of the coefficients. The degree of quantization may be modified by adjusting a quantization parameter. In some examples, the quantization unit 54 may then perform a scan of a matrix including the quantized transform coefficients. Alternatively, the entropy encoding unit 56 may perform the scan.

[00115] Following quantization, the entropy encoding unit 56 entropy encodes the quantized transform coefficients into a video bitstream using, e.g., Context Adaptive Variable Length Coding (CAVLC), Context Adaptive Binary Arithmetic Coding (CABAC), Syntaxbased context-adaptive Binary Arithmetic Coding (SB AC), Probability Interval Partitioning Entropy (PIPE) coding or another entropy encoding methodology or technique. The encoded bitstream may then be transmitted to the video decoder 30 as shown in FIG. 1A, or archived in the storage device 32 as shown in FIG. 1 A for later transmission to or retrieval by the video decoder 30. The entropy encoding unit 56 may also entropy encode the motion vectors and the other syntax elements for the current video frame being coded.

[00116] The inverse quantization unit 58 and the inverse transform processing unit 60 apply inverse quantization and inverse transformation, respectively, to reconstruct the residual video block in the pixel domain for generating a reference block for prediction of other video blocks. As noted above, the motion compensation unit 44 may generate a motion compensated predictive block from one or more reference blocks of the frames stored in the DPB 64. The motion compensation unit 44 may also apply one or more interpolation filters to the predictive block to calculate sub-integer pixel values for use in motion estimation.

[00117] The summer 62 adds the reconstructed residual block to the motion compensated predictive block produced by the motion compensation unit 44 to produce a reference block for storage in the DPB 64. The reference block may then be used by the intra

15 BC unit 48, the motion estimation unit 42 and the motion compensation unit 44 as a predictive block to inter predict another video block in a subsequent video frame.

[00118] FIG. 2B is a block diagram illustrating another exemplary video decoder 30 in accordance with some implementations of the present application. The video decoder 30 includes a video data memory 79, an entropy decoding unit 80, a prediction processing unit 81, an inverse quantization unit 86, an inverse transform processing unit 88, a summer 90, and a DPB 92. The prediction processing unit 81 further includes a motion compensation unit 82, an intra prediction unit 84, and an intra BC unit 85. The video decoder 30 may perform a decoding process generally reciprocal to the encoding process described above with respect to the video encoder 20 in connection with FIG. 1G. For example, the motion compensation unit 82 may generate prediction data based on motion vectors received from the entropy decoding unit 80, while the intra-prediction unit 84 may generate prediction data based on intra-prediction mode indicators received from the entropy decoding unit 80.

[00119] In some examples, a unit of the video decoder 30 may be tasked to perform the implementations of the present application. Also, in some examples, the implementations of the present disclosure may be divided among one or more of the units of the video decoder 30. For example, the intra BC unit 85 may perform the implementations of the present application, alone, or in combination with other units of the video decoder 30, such as the motion compensation unit 82, the intra prediction unit 84, and the entropy decoding unit 80. In some examples, the video decoder 30 may not include the intra BC unit 85 and the functionality of intra BC unit 85 may be performed by other components of the prediction processing unit 81, such as the motion compensation unit 82.

[00120] The video data memory 79 may store video data, such as an encoded video bitstream, to be decoded by the other components of the video decoder 30. The video data stored in the video data memory 79 may be obtained, for example, from the storage device 32, from a local video source, such as a camera, via wired or wireless network communication of video data, or by accessing physical data storage media (e.g., a flash drive or hard disk). The video data memory 79 may include a Coded Picture Buffer (CPB) that stores encoded video data from an encoded video bitstream. The DPB 92 of the video decoder 30 stores reference video data for use in decoding video data by the video decoder 30 (e.g., in intra or inter predictive coding modes). The video data memory 79 and the DPB 92 may be formed by any of a variety of memory devices, such as dynamic random access memory (DRAM), including Synchronous DRAM (SDRAM), Magneto-resistive RAM (MRAM), Resistive RAM (RRAM), or other types of memory devices. For illustrative purpose, the video data memory 79 and the DPB 92 are depicted as two distinct components of the video decoder 30 in FIG. 2B. But it will be apparent to one skilled in the art that the video data memory 79 and the DPB 92 may be provided by the same memory device or separate memory devices. In some examples, the video data memory 79 may be on-chip with other components of the video decoder 30, or off-chip relative to those components.

[00121] During the decoding process, the video decoder 30 receives an encoded video bitstream that represents video blocks of an encoded video frame and associated syntax elements. The video decoder 30 may receive the syntax elements at the video frame level and/or the video block level. The entropy decoding unit 80 of the video decoder 30 entropy decodes the bitstream to generate quantized coefficients, motion vectors or intra-prediction mode indicators, and other syntax elements. The entropy decoding unit 80 then forwards the motion vectors or intra-prediction mode indicators and other syntax elements to the prediction processing unit 81.

[00122] When the video frame is coded as an intra predictive coded (I) frame or for intra coded predictive blocks in other types of frames, the intra prediction unit 84 of the prediction processing unit 81 may generate prediction data for a video block of the current video frame based on a signaled intra prediction mode and reference data from previously decoded blocks of the current frame.

[00123] When the video frame is coded as an inter-predictive coded (i.e., B or P) frame, the motion compensation unit 82 of the prediction processing unit 81 produces one or more predictive blocks for a video block of the current video frame based on the motion vectors and other syntax elements received from the entropy decoding unit 80. Each of the predictive blocks may be produced from a reference frame within one of the reference frame lists. The video decoder 30 may construct the reference frame lists, List 0 and List 1, using default construction techniques based on reference frames stored in the DPB 92.

[00124] In some examples, when the video block is coded according to the intra BC mode described herein, the intra BC unit 85 of the prediction processing unit 81 produces predictive blocks for the current video block based on block vectors and other syntax elements received from the entropy decoding unit 80. The predictive blocks may be within a reconstructed region of the same picture as the current video block defined by the video encoder 20.

[00125] The motion compensation unit 82 and/or the intra BC unit 85 determines prediction information for a video block of the current video frame by parsing the motion vectors and other syntax elements, and then uses the prediction information to produce the predictive blocks for the current video block being decoded. For example, the motion compensation unit 82 uses some of the received syntax elements to determine a prediction mode (e.g., intra or inter prediction) used to code video blocks of the video frame, an inter prediction frame type (e.g., B or P), construction information for one or more of the reference frame lists for the frame, motion vectors for each inter predictive encoded video block of the frame, inter prediction status for each inter predictive coded video block of the frame, and other information to decode the video blocks in the current video frame.

[00126] Similarly, the intra BC unit 85 may use some of the received syntax elements, e.g., a flag, to determine that the current video block was predicted using the intra BC mode, construction information of which video blocks of the frame are within the reconstructed region and should be stored in the DPB 92, block vectors for each intra BC predicted video block of the frame, intra BC prediction status for each intra BC predicted video block of the frame, and other information to decode the video blocks in the current video frame. [00127] The motion compensation unit 82 may also perform interpolation using the interpolation filters as used by the video encoder 20 during encoding of the video blocks to calculate interpolated values for sub-integer pixels of reference blocks. In this case, the motion compensation unit 82 may determine the interpolation filters used by the video encoder 20 from the received syntax elements and use the interpolation filters to produce predictive blocks.

[00128] The inverse quantization unit 86 inverse quantizes the quantized transform coefficients provided in the bitstream and entropy decoded by the entropy decoding unit 80 using the same quantization parameter calculated by the video encoder 20 for each video block in the video frame to determine a degree of quantization. The inverse transform processing unit 88 applies an inverse transform, e.g., an inverse DCT, an inverse integer transform, or a conceptually similar inverse transform process, to the transform coefficients in order to reconstruct the residual blocks in the pixel domain.

[00129] After the motion compensation unit 82 or the intra BC unit 85 generates the predictive block for the current video block based on the vectors and other syntax elements, the summer 90 reconstructs decoded video block for the current video block by summing the residual block from the inverse transform processing unit 88 and a corresponding predictive block generated by the motion compensation unit 82 and the intra BC unit 85. An in-loop filter 91 such as deblocking filter, SAO filter and/or ALF may be positioned between the summer 90 and the DPB 92 to further process the decoded video block. In some examples, the in-loop filter 91 may be omitted, and the decoded video block may be directly provided by the summer 90 to the DPB 92. The decoded video blocks in a given frame are then stored in the DPB 92, which stores reference frames used for subsequent motion compensation of next video blocks. The DPB 92, or a memory device separate from the DPB 92, may also store decoded video for later presentation on a display device, such as the display device 34 of FIG. 1A.

[00130] In the current WC and AVS3 standards, motion information of the current coding block is either copied from spatial or temporal neighboring blocks specified by a merge candidate index or obtained by explicit signaling of motion estimation. The focus of the present disclosure is to improve the accuracy of the motion vectors for affine merge mode by improving the derivation methods of affine merge candidates. To facilitate the description of the present disclosure, the existing affine merge mode design in the WC standard is used as an example to illustrate the proposed ideas. Please note that though the existing affine mode design in the WC standard is used as the example throughout the present disclosure, to a person skilled in the art of modern video coding technologies, the proposed technologies can also be applied to a different design of affine motion prediction mode or other coding tools with the same or similar design spirit.

[00131] In a typical video coding process, a video sequence typically includes an ordered set of frames or pictures. Each frame may include three sample arrays, denoted SL, SCb, and SCr. SL is a two-dimensional array of luma samples. SCb is a two-dimensional array of Cb chroma samples. SCr is a two-dimensional array of Cr chroma samples. In other instances, a frame may be monochrome and therefore includes only one two-dimensional array of luma samples.

[00132] As shown in FIG. 1C, the video encoder 20 (or more specifically a partition unit in a prediction processing unit of the video encoder 20) generates an encoded representation of a frame by first partitioning the frame into a set of CTUs. A video frame may include an integer number of CTUs ordered consecutively in a raster scan order from left to right and from top to bottom. Each CTU is a largest logical coding unit and the width and height of the CTU are signaled by the video encoder 20 in a sequence parameter set, such that all the CTUs in a video sequence have the same size being one of 128x128, 64x64, 32x32, and 16x16. But it should be noted that the present application is not necessarily limited to a particular size. As shown in FIG. ID, each CTU may include one CTB of luma samples, two corresponding coding tree blocks of chroma samples, and syntax elements used to code the samples of the coding tree blocks. The syntax elements describe properties of different types of units of a coded block of pixels and how the video sequence can be reconstructed at the video decoder 30, including inter or intra prediction, intra prediction mode, motion vectors, and other parameters. In monochrome pictures or pictures having three separate color planes, a CTU may include a single coding tree block and syntax elements used to code the samples of the coding tree block. A coding tree block may be an NxN block of samples.

[00133] To achieve a better performance, the video encoder 20 may recursively perform tree partitioning such as binary-tree partitioning, ternary-tree partitioning, quad-tree partitioning or a combination thereof on the coding tree blocks of the CTU and divide the CTU into smaller CUs. As depicted in FIG. IE, the 64x64 CTU 400 is first divided into four smaller CUs, each having a block size of 32x32. Among the four smaller CUs, CU 410 and CU 420 are each divided into four CUs of 16x16 by block size. The two 16x16 CUs 430 and 440 are each further divided into four CUs of 8x8 by block size. FIG. IF depicts a quad-tree data structure illustrating the end result of the partition process of the CTU 400 as depicted in FIG. IE, each leaf node of the quad-tree corresponding to one CU of a respective size ranging from 32x32 to 8x8. Eike the CTU depicted in FIG. ID, each CU may include a CB of luma samples and two corresponding coding blocks of chroma samples of a frame of the same size, and syntax elements used to code the samples of the coding blocks. In monochrome pictures or pictures having three separate color planes, a CU may include a single coding block and syntax structures used to code the samples of the coding block. It should be noted that the quad-tree partitioning depicted in FIGS. 1E-1F is only for illustrative purposes and one CTU can be split into CUs to adapt to varying local characteristics based on quad/ternary/binary-tree partitions. In the multi-type tree structure, one CTU is partitioned by a quad-tree structure and each quad-tree leaf CU can be further partitioned by a binary and ternary tree structure. As shown in FIGS. 3A-3E, there are five possible partitioning types of a coding block having a width W and a height H, i.e., quaternary partitioning, horizontal binary partitioning, vertical binary partitioning, horizontal ternary partitioning, and vertical ternary partitioning.

[00134] In some implementations, the video encoder 20 may further partition a coding block of a CU into one or more MxN PBs. A PB is a rectangular (square or non-square) block of samples on which the same prediction, inter or intra, is applied. A PU of a CU may include a PB of luma samples, two corresponding PBs of chroma samples, and syntax elements used to predict the PBs. In monochrome pictures or pictures having three separate color planes, a PU may include a single PB and syntax structures used to predict the PB. The video encoder 20 may generate predictive luma, Cb, and Cr blocks for luma, Cb, and Cr PBs of each PU of the CU.

[00135] The video encoder 20 may use intra prediction or inter prediction to generate the predictive blocks for a PU. If the video encoder 20 uses intra prediction to generate the predictive blocks of a PU, the video encoder 20 may generate the predictive blocks of the PU based on decoded samples of the frame associated with the PU. If the video encoder 20 uses inter prediction to generate the predictive blocks of a PU, the video encoder 20 may generate the predictive blocks of the PU based on decoded samples of one or more frames other than the frame associated with the PU.

[00136] After the video encoder 20 generates predictive luma, Cb, and Cr blocks for one or more PUs of a CU, the video encoder 20 may generate a luma residual block for the CU by subtracting the CU’s predictive luma blocks from its original luma coding block such that each sample in the CU’s luma residual block indicates a difference between a luma sample in one of the CU’s predictive luma blocks and a corresponding sample in the CU’s original luma coding block. Similarly, the video encoder 20 may generate a Cb residual block and a Cr residual block for the CU, respectively, such that each sample in the CU’s Cb residual block indicates a difference between a Cb sample in one of the CU’s predictive Cb blocks and a corresponding sample in the CU’s original Cb coding block and each sample in the CU’s Cr residual block may indicate a difference between a Cr sample in one of the CU's predictive Cr blocks and a corresponding sample in the CU's original Cr coding block.

[00137] Furthermore, as illustrated in FIG. IE, the video encoder 20 may use quad -tree partitioning to decompose the luma, Cb, and Cr residual blocks of a CU into one or more luma, Cb, and Cr transform blocks respectively. A transform block is a rectangular (square or non- square) block of samples on which the same transform is applied. A TU of a CU may include a transform block of luma samples, two corresponding transform blocks of chroma samples, and syntax elements used to transform the transform block samples. Thus, each TU of a CU may be associated with a luma transform block, a Cb transform block, and a Cr transform block. In some examples, the luma transform block associated with the TU may be a sub-block of the CU's luma residual block. The Cb transform block may be a sub-block of the CU's Cb residual block. The Cr transform block may be a sub-block of the CU's Cr residual block. In monochrome pictures or pictures having three separate color planes, a TU may include a single transform block and syntax structures used to transform the samples of the transform block.

[00138] The video encoder 20 may apply one or more transforms to a luma transform block of a TU to generate a luma coefficient block for the TU. A coefficient block may be a two-dimensional array of transform coefficients. A transform coefficient may be a scalar quantity. The video encoder 20 may apply one or more transforms to a Cb transform block of a TU to generate a Cb coefficient block for the TU. The video encoder 20 may apply one or more transforms to a Cr transform block of a TU to generate a Cr coefficient block for the TU. [00139] After generating a coefficient block (e.g., a luma coefficient block, a Cb coefficient block or a Cr coefficient block), the video encoder 20 may quantize the coefficient block. Quantization generally refers to a process in which transform coefficients are quantized to possibly reduce the amount of data used to represent the transform coefficients, providing further compression. After the video encoder 20 quantizes a coefficient block, the video encoder 20 may entropy encode syntax elements indicating the quantized transform coefficients. For example, the video encoder 20 may perform CABAC on the syntax elements indicating the quantized transform coefficients. Finally, the video encoder 20 may output a bitstream that includes a sequence of bits that forms a representation of coded frames and associated data, which is either saved in the storage device 32 or transmitted to the destination device 14. [00140] After receiving a bitstream generated by the video encoder 20, the video decoder 30 may parse the bitstream to obtain syntax elements from the bitstream. The video decoder 30 may reconstruct the frames of the video data based at least in part on the syntax elements obtained from the bitstream. The process of reconstructing the video data is generally reciprocal to the encoding process performed by the video encoder 20. For example, the video decoder 30 may perform inverse transforms on the coefficient blocks associated with TUs of a current CU to reconstruct residual blocks associated with the TUs of the current CU. The video decoder 30 also reconstructs the coding blocks of the current CU by adding the samples of the predictive blocks for PUs of the current CU to corresponding samples of the transform blocks of the TUs of the current CU. After reconstructing the coding blocks for each CU of a frame, video decoder 30 may reconstruct the frame.

[00141] As noted above, video coding achieves video compression using primarily two modes, i.e., intra-frame prediction (or intra-prediction) and inter-frame prediction (or interprediction). It is noted that IBC could be regarded as either intra-frame prediction or a third mode. Between the two modes, inter-frame prediction contributes more to the coding efficiency than intra-frame prediction because of the use of motion vectors for predicting a current video block from a reference video block.

[00142] But with the ever improving video data capturing technology and more refined video block size for preserving details in the video data, the amount of data required for representing motion vectors for a current frame also increases substantially. One way of overcoming this challenge is to benefit from the fact that not only a group of neighboring CUs in both the spatial and temporal domains have similar video data for predicting purpose but the motion vectors between these neighboring CUs are also similar. Therefore, it is possible to use the motion information of spatially neighboring CUs and/or temporally co-located CUs as an approximation of the motion information (e.g., motion vector) of a current CU by exploring their spatial and temporal correlation, which is also referred to as “Motion Vector Predictor (MVP)” of the current CU. [00143] Instead of encoding, into the video bitstream, an actual motion vector of the current CU determined by the motion estimation unit as described above in connection with FIG. IB, the motion vector predictor of the current CU is subtracted from the actual motion vector of the current CU to produce a Motion Vector Difference (MVD) for the current CU. By doing so, there is no need to encode the motion vector determined by the motion estimation unit for each CU of a frame into the video bitstream and the amount of data used for representing motion information in the video bitstream can be significantly decreased.

[00144] Like the process of choosing a predictive block in a reference frame during interframe prediction of a code block, a set of rules need to be adopted by both the video encoder 20 and the video decoder 30 for constructing a motion vector candidate list (also known as a “merge list”) for a current CU using those potential candidate motion vectors associated with spatially neighboring CUs and/or temporally co-located CUs of the current CU and then selecting one member from the motion vector candidate list as a motion vector predictor for the current CU. By doing so, there is no need to transmit the motion vector candidate list itself from the video encoder 20 to the video decoder 30 and an index of the selected motion vector predictor within the motion vector candidate list is sufficient for the video encoder 20 and the video decoder 30 to use the same motion vector predictor within the motion vector candidate list for encoding and decoding the current CU.

[00145] Affine Model

[00146] In HEVC, only translation motion model is applied for motion compensated prediction. While in the real world, there are many kinds of motion, e.g., zoom in/out, rotation, perspective motions and other irregular motions. In the WC and AVS3, affine motion compensated prediction is applied by signaling one flag for each inter coding block to indicate whether the translation motion model or the affine motion model is applied for inter prediction. In the current WC and AVS3 design, two affine modes, including 4-paramter affine mode and 6-parameter affine mode, are supported for one affine coding block. [00147] The 4-parameter affine model has the following parameters: two parameters for translation movement in horizontal and vertical directions respectively, one parameter for zoom motion and one parameter for rotational motion for both directions. In this model, horizontal zoom parameter is equal to vertical zoom parameter, and horizontal rotation parameter is equal to vertical rotation parameter. To achieve a better accommodation of the motion vectors and affine parameter, those affine parameters are to be derived from two MVs (which are also called control point motion vector (CPMV)) located at the top-left corner and top-right corner of a current block. As shown in FIGS. 4A-4B, the affine motion field of the block is described by two CPMVs (Vo, Vi). Based on the control point motion, the motion field (y _x , Vy) of one affine coded block is described as

[00148] The 6-parameter affine mode has the following parameters: two parameters for translation movement in horizontal and vertical directions respectively, two parameters for zoom motion and rotation motion respectively in horizontal direction, another two parameters for zoom motion and rotation motion respectively in vertical direction. The 6-parameter affine motion model is coded with three CPMVs. As shown in FIG. 5, the three control points of one 6-paramter affine block are located at the top-left, top-right and bottom left corner of the block. The motion at top-left control point is related to translation motion, and the motion at topright control point is related to rotation and zoom motion in horizontal direction, and the motion at bottom-left control point is related to rotation and zoom motion in vertical direction. Compared to the 4-parameter affine motion model, the rotation and zoom motion in horizontal direction of the 6-paramter may not be same as those motion in vertical direction. Assuming (Vo, Vi, V2) are the MVs of the top-left, top-right and bottom-left corners of the current block in FIG. 5, the motion vector of each sub-block ( _x , v ) is derived using the three MVs at control points as:

[00149] Affine Merge Mode

[00150] In affine merge mode, the CPMVs for the current block are not explicitly signaled but derived from neighboring blocks. Specifically, in this mode, motion information of spatial neighbor blocks is used to generate CPMVs for the current block. The affine merge mode candidate list has a limited size. For example, in the current WC design, there may be up to five candidates. The encoder may evaluate and choose the best candidate index based on rate-distortion optimization algorithms. The chosen candidate index is then signaled to the decoder side. The affine merge candidates can be decided in three ways. In the first way, the affine merge candidates may be inherited from neighboring affine coded blocks. In the second way, the affine merge candidates maybe constructed from translational MVs from neighboring blocks. In the third way, zero MVs are used as the affine merge candidates.

[00151] For the inherited method, there may be up to two candidates. The candidates are obtained from the neighboring blocks located at the bottom-left of the current block (e.g., scanning order is from A0 to Al as shown in FIG. 6) and from the neighboring blocks located at the top-right of the current block (e.g., scanning order is from B0 to B2 as shown in FIG. 6), if available.

[00152] For the constructed method, the candidates are the combinations of neighbor’s translational MVs, which may be generated by two steps.

[00153] Step 1: obtain four translational MVs including MV1, MV2, MV3 and MV4 from available neighbors. MV1: MV from the one of the three neighboring blocks close to the top-left corner of the current block. As shown in FIG. 7, the scanning order is B2, B3 and A2.

MV2: MV from the one of the one from the two neighboring blocks close to the topright corner of the current block. As shown in FIG. 7, the scanning order is Bland BO.

MV3: MV from the one of the one from the two neighboring blocks close to the bottom-left corner of the current block. As shown in FIG. 7, the scanning order is Aland AO.

MV4: MV from the temporally collocated block of the neighboring block close to the bottom-right corner of current block. As shown in the Fig, the neighboring block is T.

[00154] Step 2: derive combinations based on the four translational MVs from Step 1.

Combination 1: MV1, MV2, MV3;

Combination 2: MV1, MV2, MV4;

Combination 3: MV1, MV3, MV4;

Combination 4: MV2, MV3, MV4;

Combination 5: MV1, MV2;

Combination 6: MV1, MV3.

[00155] When the merge candidate list is not full after filling with inherited and constructed candidates, zero MVs are inserted at the end of the list.

[00156] Affine AMVP Mode

[00157] Affine advanced motion vector prediction (AMVP) mode may be applied for CUs with both width and height larger than or equal to 16. An affine flag in CU level is signaled in the bitstream to indicate whether affine AMVP mode is used and then another flag is signaled to indicate whether 4-parameter affine or 6-parameter affine. In this mode, the difference of the CPMVs of current CU and their CPMV predictors (CPMVPs) is signaled in the bitstream. The affine AVMP candidate list size is 2 and the affine AMVP candidate list is generated by using the following four types of CPMV candidate in order below:

- Inherited affine AMVP candidates that extrapolated from the CPMVs of the neighbour CUs; - Constructed affine AMVP candidates CPMVPs that are derived using the translational MVs of the neighbor CUs;

- Translational MVs from neighboring CUs;

- Temporal MV from collocated CUs; and

- Zero MVs.

[00158] The checking order of inherited affine AMVP candidates is the same to the checking order of inherited affine merge candidates. The only difference is that, for AMVP candidate, only the affine CU that has the same reference picture as in current block is considered. No pruning process is applied when inserting an inherited affine motion predictor into the candidate list.

[00159] Constructed AMVP candidate is derived from the same spatial neighbors as affine merge mode. The same checking order is used as done in affine merge candidate construction. In addition, reference picture index of the neighboring block is also checked. The first block in the checking order that is inter coded and has the same reference picture as in current CUs is used. When the current CU is coded with 4-parameter affine mode, and mv ₀ and mv ₁ are both available, mv ₀ and mv ₁ are added as one candidate in the affine AMVP candidate list. When the current CU is coded with 6-parameter affine mode, and all three CPMVs are available, they are added as one candidate in the affine AMVP candidate list. Otherwise, constructed AMVP candidate is set as unavailable.

[00160] If the number of candidates in the affine AMVP list is still less than 2 after valid inherited affine AMVP candidates and constructed AMVP candidate are inserted, mv ₀ , mv ₁ and mv ₂ will be added, in order, as the translational MVs to predict all control point MVs of the current CU, when available. Finally, zero MVs are used to fill the affine AMVP list if it is still not full.

[00161] Regular Inter Merge Mode

[00162] In some embodiments, the regular inter merge candidate list is constructed by including the following five types of candidates in order: (1) Spatial MVP from spatial neighbor CUs;

(2) Temporal MVP from collocated CUs;

(3) History-based MVP from a first in first out (FIFO) table;

(4) Pairwise average MVP; and

(5) Zero MVs.

[00163] The size of merge list is signaled in sequence parameter set header and the maximum allowed size of merge list is 6. For each CU code in merge mode, an index of best merge candidate is encoded using truncated unary binarization (TU). The first bin of the merge index is coded with context and bypass coding is used for other bins.

[00164] The derivation process of each category of merge candidates is provided above. In some embodiments, parallel derivation of the merging candidate lists may be supported for all CUs within a certain size of area.

[00165] Spatial Candidate Derivation

[00166] The derivation of spatial merge candidates in WC is same to that in HEVC except the positions of first two merge candidates are swapped. A maximum of four merge candidates are selected among candidates located in the positions depicted in FIG. 4C. The order of derivation is B0, A0, Bl, Al and B2. Position B2 is considered only when one or more than one CUs of position B0, A0, Bl, Al are not available (e.g., because it belongs to another slice or tile) or is intra coded. After candidate at position Al is added, the addition of the remaining candidates is subject to a redundancy check which ensures that candidates with same motion information are excluded from the list so that coding efficiency is improved. To reduce computational complexity, not all possible candidate pairs are considered in the mentioned redundancy check. Instead, only the pairs linked with an arrow in FIG. 4D are considered and a candidate is only added to the list if the corresponding candidate used for redundancy check has not the same motion information. FIG. 4D illustrates candidate pairs that are considered for redundancy check of spatial merge candidates.

[00167] Temporal Candidate Derivation [00168] In this step, only one candidate is added to the list. Particularly, in the derivation of this temporal merge candidate, a scaled motion vector is derived based on co-located CU belonging to the collocated reference picture. The reference picture list and the reference index to be used for derivation of the co-located CU is explicitly signaled in the slice header. The scaled motion vector for temporal merge candidate is obtained as illustrated by the dotted line in FIG. 4E, which is scaled from the motion vector of the co-located CU using the POC distances, tb and td, where tb is defined to be the POC difference between the reference picture of the current picture and the current picture and td is defined to be the POC difference between the reference picture of the co-located picture and the co-located picture. The reference picture index of temporal merge candidate is set equal to zero.

[00169] The position for the temporal candidate is selected between candidates CO and Cl, as depicted in FIG. 4F. If CU at position CO is not available, is intra coded, or is outside of the current row of CTUs, position Cl is used. Otherwise, position CO is used in the derivation of the temporal merge candidate.

[00170] History-based Merge Candidate Derivation

[00171] The history-based MVP (HMVP) merge candidates are added to merge list after the spatial MVP and temporal motion vector prediction (TMVP). In this method, the motion information of a previously coded block is stored in a table and used as MVP for the current CU. The table with multiple HMVP candidates is maintained during the encoding/decoding process. The table is reset (emptied) when a new CTU row is encountered. Whenever there is a non-subblock inter-coded CU, the associated motion information is added to the last entry of the table as a new HMVP candidate.

[00172] The HMVP table size S may be set to be 6, which indicates up to 5 History-based MVP (HMVP) candidates may be added to the table. When inserting a new motion candidate to the table, a constrained first-in-first-out (FIFO) rule is utilized wherein redundancy check is firstly applied to find whether there is an identical HMVP in the table. If found, the identical HMVP is removed from the table and all the HMVP candidates afterwards are moved forward, and the identical HMVP is inserted to the last entry of the table.

[00173] HMVP candidates could be used in the merge candidate list construction process. The latest several HMVP candidates in the table are checked in order and inserted to the candidate list after the TMVP candidate. Redundancy check is applied on the HMVP candidates to the spatial or temporal merge candidate.

[00174] To reduce the number of operations for redundancy check, the following simplifications are introduced. First, the last two entries in the table are redundancy checked to Al and Bl spatial candidates, respectively. Second, once the total number of available merge candidates reaches the maximally allowed merge candidates minus 1, the merge candidate list construction process from HMVP is terminated.

[00175] Pair-Wise Average Merge Candidates Derivation

[00176] Pairwise average candidates are generated by averaging predefined pairs of candidates in the existing merge candidate list, using the first two merge candidates. The first merge candidate is defined as pOCand and the second merge candidate can be defined as plCand, respectively. The averaged motion vectors are calculated according to the availability of the motion vector of pOCand and plCand separately for each reference list. If both motion vectors are available in one list, these two motion vectors are averaged even when they point to different reference pictures, and its reference picture is set to the one of pOCand; if only one motion vector is available, use the one directly; if no motion vector is available, keep this list invalid. Also, if the half-pel interpolation filter indices of pOCand and plCand are different, it is set to 0.

[00177] When the merge list is not full after pair-wise average merge candidates are added, the zero MVPs are inserted at the end of the merge list until the maximum merge candidate number is encountered.

[00178] Adaptive Reordering of Merge Candidates with Template Matching (ARMC) [00179] The reordering method, named as ARMC, is applied to regular merge mode, template matching (TM) merge mode, and affine merge mode (excluding the SbTMVP candidate), where the SbTMVP represents the Subblock-based Temporal Motion Vector Prediction candidate. For the TM merge mode, merge candidates are reordered before the refinement process.

[00180] After a merge candidate list is constructed, merge candidates are divided into several subgroups. The subgroup size is set to 5. Merge candidates in each subgroup are reordered ascendingly according to cost values based on template matching. For simplification, merge candidates in the last but not the first subgroup are not reordered.

[00181] The template matching cost is measured by the sum of absolute differences (SAD) between samples of a template of the current block and their corresponding reference samples. The template includes a set of reconstructed samples neighboring to the current block. Reference samples of the template are located by the same motion information of the current block.

[00182] When a merge candidate utilizes bi-directional prediction, the reference samples of the template of the merge candidate are also generated by bi-prediction as shown in FIG. 19.

[00183] Regression based affine candidate derivation

[00184] During the VVC standardization progress, the Regression based Motion Vector Field (RMVF) derivation method was proposed in JVET-L0171 which provides a new variety of subblock-based merge candidate. The motion vectors and center positions from the neighboring subblocks of the current CU, as illustrated in FIG. 32, are used as the input to the linear regression process to derive a set of linear model parameters.

[00185] During the JVET standardization exploration progress after VVC, the regression based affine candidate derivation method was proposed in JVET-Z0125. Tn this method, the subblock motion field from a previous coded affine CU and the motion vectors from the adjacent subblocks of current CU are used as the input for the regression process. Compared to the regression process in JVET -L0171, the difference is that the predicted CPMVs instead of the subblock motion field for current block are derived as output. The derived CPMVs can be added into the subblock merge candidate list or the affine AMVP list.

[00186] Merge mode with MVD (MMVD)

[00187] In addition to merge mode, where the implicitly derived motion information is directly used for prediction samples generation of the current CU, the merge mode with motion vector differences (MMVD) is introduced in WC. A MMVD flag is signaled right after sending a regular merge flag to specify whether MMVD mode is used for a CU.

[00188] In MMVD, after a merge candidate is selected, it is further refined by the signaled MVDs information. The further information includes a merge candidate flag, an index to specify motion magnitude, and an index for indication of motion direction. In MMVD mode, one for the first two candidates in the merge list is selected to be used as MV basis. The MMVD candidate flag is signaled to specify which one is used between the first and second merge candidates.

[00189] Distance index specifies motion magnitude information and indicate the predefined offset from the starting point. As shown in FIG. 26, an offset is added to either horizontal component or vertical component of starting MV. The relation of distance index and pre-defined offset is specified in below Table 6.

Table 6

[00190] Direction index represents the direction of the MVD relative to the starting point. The direction index can represent of the four directions as shown in above table. It’s noted that the meaning of MVD sign could be variant according to the information of starting MVs. When the starting MVs is a uni-prediction MV or bi-prediction MVs with both lists point to the same side of the current picture (i.e. POCs of two references are both larger than the POC of the current picture, or are both smaller than the POC of the current picture), the sign in below Table 7 specifies the sign of MV offset added to the starting MV. When the starting MVs is bi-prediction MVs with the two MVs point to the different sides of the current picture (i.e. the POC of one reference is larger than the POC of the current picture, and the POC of the other reference is smaller than the POC of the current picture), and the difference of POC in list 0 is greater than the one in list 1 , the sign in below table specifies the sign of MV offset added to the listO MV component of starting MV and the sign for the listl MV has opposite value. Otherwise, if the difference of POC in list 1 is greater than list 0, the sign in below table specifies the sign of MV offset added to the listl MV component of starting MV and the sign for the listO MV has opposite value.

[00191] The MVD is scaled according to the difference of POCs in each direction. If the differences of POCs in both lists are the same, no scaling is needed. Otherwise, if the difference of POC in list 0 is larger than the one of list 1, the MVD for list 1 is scaled, by defining the POC difference of L0 as td and POC difference of LI as tb, described in FIG. 23. If the POC difference of LI is greater than L0, the MVD for list 0 is scaled in the same way. If the starting MV is uni-predicted, the MVD is added to the available MV.

Table 7

[00192]

[00193] For subblock-based merge candidates with subblock size equal to Wsub * Hsub, the above template includes several sub-templates with the size of Wsub x 1, and the left template includes several sub-templates with the size of 1 x Hsub. Wsub is the width of the subblock and Hsub is the height of the subblock. As shown in FIG. 20. The motion information of the subblocks in the first row and the first column of current block is used to derive the reference samples of each sub-template. [00194] For the current video standards WC and AVS, only adjacent neighboring blocks are used to derive affine merge candidates for the current block, as shown in FIG. 6 and FIG. 7 for inherited candidates and constructed candidates respectively. To increase the diversity of merge candidates and further explore spatial correlations, it is straightforward to extend the coverage of neighboring blocks from adjacent areas to non-adjacent areas.

[00195] In the current video standards WC and AVS, each affine inherited candidate is derived from one neighboring block with affine motion information. On the other hand, each affine constructed candidate is derived from two or three neighboring blocks with translational motion information. To further explore spatial correlations, a new candidate derivation method which combines affine motion and translational motion may be investigated.

[00196] The candidate derivation methods proposed for affine merge mode, may be extended to other coding modes, such as affine AMVP mode and regular merge mode.

[00197] In the present disclosure, the candidate derivation process for affine merge mode is extended by using not only adjacent neighboring blocks but also non-adjacent neighboring blocks. Detailed methods may be summarized in following aspects including affine merge candidate pruning, non-adjacent neighbor based derivation process for affine inherited merge candidates, non-adjacent neighbor based derivation process for affine constructed merge candidates, inheritance based derivation method for affine constructed merge candidates, HMVP based derivation method for affine constructed merge candidates, candidate derivation method for affine AMVP mode and regular merge mode, and motion information storage.

[00198] Affine Merge Candidate Pruning

[00199] As the affine merge candidate list in a typical video coding standards usually has a limited size, candidate pruning is an essential process to remove redundant ones. For both affine merge inherited candidates and constructed candidates, this pruning process is needed. As explained in the introduction section, CPMVs of a current block are not directly used for affine motion compensation. Instead, CPMVs need to be converted into translational MVs at the location of each sub-block within the current block. The conversion process is performed by following a general affine model as shown below:

[00200] where (a, b) are delta translation parameters, (c, d) are delta zoom and rotation parameters for horizontal direction, (e, f) are delta zoom and rotation parameters for vertical direction, (x, y) are the horizontal and vertical distance of the pivot location (e.g., the center or top-left corner) of a sub-block relative to the top-left corner of the current block (e.g., the coordinate (x, y) shown in FIG. 5), and (v _x , v _y ) is the target translational MVs of the sub-block. [00201] For 6-parameter affine model, three CPMVs, termed as V0, VI and V2, are available. Then the six model parameters a, b, c, d, e and f can be calculated as

[00202] For 4-parameter affine model, if top-left corner CPMV and top-right corner

CPMV, termed as VO and VI, are available, the six parameters of a, b, c, d, e and f can be calculated as [00203] For 4-parameter affine model, if top-left corner CPMV and bottom-left corner

CPMV, termed as V0 and V2, are available, the six parameters of a, b, c, d, e and f can be calculated as

[00204] In above equations (4), (5), and (6), w and h represent the width and height of the current block, respectively.

[00205] When two merge candidate sets of CPMVs are compared for redundancy check, it is proposed to check the similarity of the 6 affine model parameters. Therefore, the candidate pruning process can be performed in two steps.

[00206] In Step 1, given two candidate sets of CPMVs, the corresponding affine model parameters for each candidate set are derived. More specifically, the two candidate sets of CPMVs may be represented by two sets of affine model parameters, e.g., (a _1; b _1; c _1; d _1; e _1; fj and (a ₂ , b ₂ , c ₂ , d ₂ , e ₂ , f ₂ ).

[00207] In Step 2, based on one or more pre-defined threshold values, similarity check is performed between the two sets of affine model parameters. In one embodiment, when the absolute values of (a _r - a ₂ ), (b^ b ₂ ), (<q- c ₂ ), (d _t - d ₂ ), e ₂ ) and ( - f ₂ ) are all below a positive threshold value, such as the value of 1, the two candidates are considered to be similar and one of them can be pruned/removed and not put in the merge candidate list.

[00208] In some embodiments, the divisions or right shift operations in Step 1 may be removed to simplify the calculations in the CPMV pruning process.

[00209] Specifically, the model parameters of c, d, e and f may be calculated without being divided by the width w and height h of the current block. For example, take above equation (4) as an example, the approximated model parameters of c', d' , e' and f may be calculated as below equation (7).

[00210] In the case that only two CPMVs are available, part of the model parameters is derived from the other part of the model parameters, which are dependent on the width or height of the current block. In this case, the model parameters may be converted to take the impact of the width and height into account. For example, in the case of the equation (5), the approximated model parameters of c', d’, e ¹ and f may be calculated based on equation (8) below. In the case of the equation (6), the approximated model parameters of c', d', e' and f may be calculated based on equation (9) below.

(8)

(9)

When the approximated model parameters of c’, d', e' and f are calculated in above Step 1, the calculation of the absolute values that are needed for similarity check in the Step 2 above may be changed accordingly:

[00211] In the Step 2 above, threshold values are needed to evaluate the similarity between two candidate sets of CPMV. There may be multiple ways to define the threshold values. In one embodiment, the threshold values may be defined per comparable parameter. Table 1 is one example in this embodiment showing threshold values defined per comparable model parameter. In another embodiment, the threshold values may be defined by considering the size of the current coding block. Table 2 is one example in this embodiment showing threshold values defined by the size of the current coding block.

Table 1

Table 2

[00236]

[00237] In another embodiment, the threshold values may be defined by considering the weight or the height of the current block. Table 3 and Table 4 are examples in this embodiment. Table 3 shows threshold values defined by the width of the current coding block and Table 4 shows threshold values defined by the height of the current coding block.

[00238] Table 3

Table 4

[00258] In another embodiment, the threshold values may be defined as a group of fixed values. In another embodiment, the threshold values may be defined by any combinations of above embodiments. In one example, the threshold values may be defined by considering different parameters and the weight and the height of the current block. Table 5 is one example in this embodiment showing threshold values defined by the height of the current coding block. Note that in any above proposed embodiments, the comparable parameters, if needed, may represent any parameters defined in any equations from equation (4) to equation (9).

Table 5

[00271]

[00272] The benefits of using the converted affine model parameters for candidate redundancy check include that: it creates a unified similarity check process for candidates with different affine model types, e.g., one merge candidate may user 6-parameter affine model with three CPMVs while another candidate may use 4-parameter affine model with two CPMVs; it considers the different impacts of each CPMV in a merge candidate when deriving the target MV at each sub-block; and it provides the similarity significance of two affine merge candidates related to the width and height of the current block.

[00273] Non-Adjacent Neighbor Based Derivation Process for Affine Inherited Merge Candidates

[00274] For inherited merge candidates, non-adjacent neighbor based derivation process may be performed in three steps. Step 1 is for candidate scanning. Step 2 is for CPMV projection. Step 3 is for candidate pruning.

[00275] In Step 1, non-adjacent neighboring blocks are scanned and selected by following methods.

[00276] Scanning area and distance

[00277] In some examples, non-adjacent neighboring blocks may be scanned from left area and above area of the current coding block. The scanning distance may be defined as the number of coding blocks from the scanning position to the left side or top side of the current coding blocks.

[00278] As shown in FIG. 8, on either the left or above of the current coding block, multiple lines of non-adjacent neighboring blocks may be scanned. The distance shown in FIG. 8 represents the number of coding blocks from each candidate position to the left side or top side of the current block. For example, the area with “distance 2 (D2)” on the left side of the current block indicates that the candidate neighboring blocks located in this area are 2 blocks away from the current block. Similar indications may be applied to other scanning areas with different distances.

[00279] In one or more embodiments, the non-adjacent neighboring blocks at each distance may have the same block size as the current coding block, as shown in the FIG. 13A. As shown in FIG. 13A, the non-adjacent neighbor blocks 1301 on the left side and the non- adjacent neighbor blocks 1302 on the above side have the same size as the current block 1303. In some embodiments, the non-adjacent neighboring blocks at each distance may have a different block size as the current coding block, as shown in the FIG. 13B. The neighbor block 1304 is an adjacent neighbor block to the current block 1303. As shown in FIG. 13B, the non- adjacent neighbor blocks 1305 on the left side and the non-adjacent neighbor blocks 1306 on the above side have the same size as the current block 1307. The neighbor block 1308 is an adjacent neighbor block to the current block 1307.

[00280] Note that when the non-adjacent neighboring blocks at each distance have the same block size as the current coding block, the value of the block size is adaptively changed according to the partition granularity at each different area in an image. Note that when the non-adjacent neighboring blocks at each distance have a different block size as the current coding block, the value of the block size may be predefined as a constant value, such as 4x4, 8x8 or 16x16. The 4x4 non-adjacent motion fields shown in FIG. 10 and FIG. 12 are examples in this case, where the motion fields may be considered as, but not limited to, special cases of sub-blocks.

[00281] Similarly, the non-adjacent coding blocks shown in FIG. 11 may have different sizes as well. In one example, the non-adjacent coding blocks may have the size as the current coding block, which is adaptively changed. In another example, the non-adjacent coding blocks may have a predefined size with a fixed value, such as 4x4, 8x8 or 16x16.

[00282] Based on the defined scanning distance, the total size of the scanning area on either the left or above of the current coding clock may be determined by a configurable distance value. In one or more embodiments, the maximum scanning distance on the left side and above side may use a same value or different values. FIG. 13 shows an example where the maximum distance on both the left side and above side shares a same value of 2. The maximum scanning distance value(s) may be determined by the encoder side and signaled in a bitstream. Alternatively, the maximum scanning distance value(s) may be predefined as fixed value(s), such as the value of 2 or 4. When the maximum scanning distance is predefined as the value of 4, it indicates that the scanning process is terminated when the candidate list is full or all the non-adjacent neighboring blocks with at most distance 4 have been scanned, whichever comes first. [00283] In one or more embodiments, within each scanning area at a specific distance, the starting and ending neighboring blocks may be position dependent.

[00284] In some embodiments, for the left side scanning areas, the starting neighboring blocks may be the adjacent bottom-left block of the starting neighboring block of the adjacent scanning area with smaller distance. For example, as shown in FIG. 8, the starting neighboring block of the “distance 2” scanning area on the left side of the current block is the adjacent bottom-left neighboring block of the starting neighboring block of the “distance 1 (DI)” scanning area. In FIG. 8, DI, D2, D3 respectively indicates distance 1, distance 2, and distance 3. The ending neighboring blocks may be the adjacent left block of the ending neighboring block of the above scanning area with smaller distance. For example, as shown in FIG. 8, the ending neighboring block of the “distance 2” scanning area on the left side of the current block is the adjacent left neighboring block of the ending neighboring block of the “distance 1” scanning area above the current block.

[00285] Similarly, for the above side scanning areas, the starting neighboring blocks may be the adjacent top-right block of the starting neighboring block of the adjacent scanning area with smaller distance. The ending neighboring blocks may be the adjacent top-left block of the ending neighboring block of the adjacent scanning area with smaller distance.

[00286] Scanning Order

[00287] When the neighboring blocks are scanned in the non-adjacent areas, certain order or/and rules may be followed to determine the selections of the scanned neighboring blocks.

[00288] In some embodiments, the left area may be scanned first, and then followed by scanning the above areas. As shown in FIG. 8, three lines of non-adjacent areas (e.g., from distance 1 (DI) to distance 3 (D3)) on the left side may be scanned first, then followed by scanning the three lines of non-adjacent areas above the current block. [00289] In some embodiments, the left areas and above areas may be scanned alternatively. For example, as shown in FIG. 8, the left scanning area with “distance 1” is scanned first, then followed by the scanning the above area with “distance 1.”

[00290] For scanning areas located on the same side (e.g., left or above areas), the scanning order is from the areas with small distance to the areas with large distance. This order may be flexibly combined with other embodiments of scanning order. For example, the left and above areas may be scanned alternatively, and the order for same side areas is scheduled to be from small distance to large distance.

[00291] Within each scanning area at a specific distance, a scanning order may be defined. In one embodiment, for the left scanning areas, the scanning may be started from the bottom neighboring block to the top neighboring block. For the above scanning areas, the scanning may be started from the right block to the left block.

[00292] Scanning Termination

[00293] For inherited merge candidates, the neighboring blocks coded with affine mode are defined as qualified candidates. In some embodiments, the scanning process may be performed interactively. For example, the scanning performed in a specific area at a specific distance may be stopped at the instance when first X qualified candidates are identified, where X is a predefined positive value. For example, as shown in FIG. 8, the scanning in the left scanning area with distance 1 may be stopped when the first one or more qualified candidates are identified. Then the next iteration of scanning process is started by targeting at another scanning area, which is regulated by a pre-defined scanning order/rule.

[00294] In one or more embodiments, the X may be defined for each distance. For example, at each distance, X is set to be 1, which means the scanning is terminated for each distance if the first qualified candidate is found and the scanning process is restarted from a different distance of the same area or the same or different distance of a different area. Note that the value of X may be set as the same value or different values for different distances. If the maximum number of qualified candidates are found from all allowable distances (e.g., regulated by a maximum distance) of an area, the scanning process for one area is completely terminated.

[00295] In another embodiment, the X may be defined for an area. For example, X is set to be 3, which means the scanning is terminated for the whole area (e.g., left or above area of the current block) if the first 3 qualified candidates are found and the scanning process is restarted from the same or different distance of another area. Note that the value of X may be set as the same value or different values for different areas. If the maximum number of qualified candidates are found from all areas, the whole scanning process is completely terminated.

[00296] The values of X may be defined for both distance and areas. For example, for each area (e.g., left or above area of the current block), X is set to 3, and for each distance, X is set to 1. The values of X may be set as the same value or different values for different areas and distances.

[00297] In some embodiments, the scanning process may be performed continuously. For example, the scanning performed in a specific area at a specific distance may be stopped at the instance when all covered neighboring blocks are scanned and no more qualified candidates are identified or the maximum allowable number of candidates is reached.

[00298] During the candidate scanning process, each candidate non-adjacent neighboring block is determined and scanned by following the above proposed scanning methods. For easier implementation, each candidate non-adjacent neighboring block may be indicated or located by a specific scanning position. Once a specific scanning area and distance are decided by following above proposed methods, the scanning positions may be determined accordingly based on following methods.

[00299] In one method, bottom-left and top-right positions are used for above and left non-adjacent neighboring blocks respectively, as shown in FIG. 15A.

[00300] In another method, bottom-right positions are used for both above and left non- adjacent neighboring blocks, as shown in FIG. 15B. [00301] In another method, bottom-left positions are used for both above and left non- adjacent neighboring blocks, as shown in FIG. 15C.

[00302] In another method, top-right positions are used for both above and left non- adjacent neighboring blocks, as shown in FIG. 15D.

[00303] For easier illustration, in FIGS. 15A-15D, each non-adjacent neighboring block is assumed to have the same block size as the current block. Without loss of generality, this illustration may be easily extended to non-adjacent neighboring blocks with different block sizes.

[00304] Further, in Step 2, the same process of CPMV projection as used in the current AVS and WC standards may be utilized. In this CPMV projection process, the current block is assumed to share the same affine model with the selected neighboring block, then two or three corner pixel’s coordinates (e.g., if the current block uses 4-prameter model, two coordinates (top-left pixel/sample location and top-right pixel/sample location) are used; if the current block uses 6-prameter model, three coordinates (top-left pixel/sample location, topright pixel/sample location and bottom-left pixel/sample location) are used) are plugged into equation (1) or (2), which depends on whether the neighboring block is coded with a 4- parameter or 6-parameter affine model, to generate two or three CPMVs.

[00305] In Step 3, any qualified candidate that is identified in Step 1 and converted in Step 2 may go through a similarity check against all existing candidates that are already in the merge candidate list. The details of similarity check are already described in the section of “Affine Merge Candidate Pruning” above. If the newly qualified candidate is found to be similar with any existing candidate in the candidate list, this newly qualified candidate is removed/pruned.

[00306] Non- Adjacent Neighbor Based Derivation Process for Affine Constructed Merge

Candidates

[00307] In the case of deriving inherited merge candidates, one neighboring block is identified at one time, where this single neighboring block needs to be coded in affine mode and may contain two or three CPMVs. In the case of deriving constructed merge candidates, two or three neighboring blocks may be identified at one time, where each identified neighboring block does not need to be coded in affine mode and only one translational MV is retrieved from this block.

[00308] FIG. 9 presents an example where constructed affine merge candidates may be derived by using non-adjacent neighboring block. In FIG. 9, A, B and C are the geographical positions of three non-adjacent neighboring blocks. A virtual coding block is formed by using the position of A as the top-left corner, the position of B as the top-right corner, and the position of C as the bottom-left corner. If considering the virtual CU as an affine coded block, the MVs at the positions of A', B' and C’ may be derived by following the equation (3), where the model parameters (a, b, c, d, e, f) may be calculated by the translational MV at the positions of A, B and C. Once derived, the MVs at positions of A’, B’ and C’ may be used as the three CPMVs for the current block, and the existing process (the one used in the AVS and WC standards) of generating constructed affine merge candidates may be used.

[00309] For constructed merge candidates, non-adjacent neighbor based derivation process may be performed in five steps. The non-adjacent neighbor based derivation process may be performed in the five steps in an apparatus such as an encoder or a decoder. Step 1 is for candidate scanning. Step 2 is for affine model determination. Step 3 is for CPMV projection. Step 4 is for candidate generation. And Step 5 is for candidate pruning. In Step 1, non-adjacent neighboring blocks may be scanned and selected by following methods.

[00310] Scanning Area and Distance

[00311] In some embodiments, to maintain a rectangular coding block, the scanning process is only performed for two non-adjacent neighboring blocks. The third non-adjacent neighboring block may be dependent on the horizontal and vertical positions of the first and second non-adjacent neighboring blocks. [00312] In some embodiments, as shown in FIG. 9, the scanning process is only performed for the positions of B and C. The position of A may be uniquely determined by the horizontal position of C and the vertical position of B.

[00313] To form a valid virtual coding block, the position of A may need to be at least valid. The validity of position A may be defined as whether the motion information at the position A is available or not. In one embodiment, the coding block located at the position A may need to be coded in inter-modes such that the motion information is available to form a virtual coding block.

[00314] In some embodiments, the scanning area and distance may be defined according to a specific scanning direction.

[00315] In some embodiments, the scanning direction may be perpendicular to the side of the current block. One example is shown in FIG. 10, where the scanning area is defined as one line of continuous motion fields on the left or above the current block. The scanning distance is defined as the number of motion fields from the scanning position to the side of the current block. Note that the size of the motion filed may be dependent on the max granularity of the applicable video coding standards. In the example shown in FIG. 10, the size of the motion field is assumed to be aligned with the current WC standards and set to be 4x4.

[00316] In some embodiments, the scanning direction may be parallel to the side of the current block. One example is shown in FIG. 11, where the scanning area is defined as the one line of continuous coding blocks on the left or above the current block.

[00317] In some embodiments, the scanning direction may be a combination of perpendicular and parallel scanning to the side of the current block. One example is shown in FIG. 12. As shown in FIG. 12, the scanning direction may be also a combination of parallel and diagonal. Scanning at position B starts from left to right, and then in a diagonal direction to the left and upper block. The scanning at position B will repeat as shown in FIG. 12. Similarly, scanning at position C starts from top to bottom, and then in a diagonal direction to the left and upper block. The scanning at position C will repeat as shown in FIG. 12. [00318] Scanning Order

[00319] In some embodiments, the scanning order may be defined as from the positions with smaller distance to the positions with larger distance to the current coding block. This order may be applied to the case of perpendicular scanning.

[00320] In some embodiments, the scanning order may be defined as a fixed pattern. This fix-pattern scanning order may be used for the candidate positions with similar distance. One example is the case of parallel scanning. In one example, the scanning order may be defined as top-down direction for the left scanning area, and may be defined as from left to right directions for the above scanning areas, like the example shown in FIG. 11.

[00321] For the case of the combined scanning method, the scanning order may be a combination of fix-pattern and distance dependent, like the example shown in FIG. 12.

[00322] Scanning Termination

[00323] For constructed merge candidates, the qualified candidate does not need to be affine coded since only translational MV is needed.

[00324] Dependent on the required number of candidates, the scanning process may be terminated when the first X qualified candidates are identified, where X is a positive value.

[00325] As shown in FIG. 9, in order to form a virtual coding block, three corners named as A, B and C are needed. For easier implementation, the scanning process in Step 1 may be only performed for identifying the non-adjacent neighboring blocks located at corners B and C, while the coordinate of A may be precisely determined by taking the horizontal coordinate of C and the vertical coordinate of B. In this way, the formed virtual coding block is restricted to be rectangle. In the case when either B or C point is unavailable, e.g., out of boundary, or the motion information at the non-adjacent neighboring blocks corresponding to B or C is unavailable (e.g., the block is coded in intra-mode or screen-content modes), the horizontal coordinate or vertical coordinate of C may be defined as the horizontal coordinate or vertical coordinate of the top-left point of the current block respectively. [00326] In another embodiment, when the corner B and/or corner C is firstly determined from the scanning process in Step 1, the non-adjacent neighboring blocks located at corner B and/or C may be identified accordingly. Secondly, the position(s) of the corner B and/or C may be reset to pivot point within the corresponding non-adjacent neighboring blocks, such as the mass center of each non-adjacent neighboring block. For example, the mass center may be defined as the geometric center of each neighboring block.

[00327] When scanning process is performed for corners B and corner C as shown in FIG. 9, the process may be performed jointly or independently. In examples of independent scanning, the previously proposed scanning methods may be applied separately on the corners B and C. In examples of joint scanning, there may be different methods as follows.

[00328] In one embodiment, pairwise scanning may be performed. In one example of pairwise scanning, the candidate positions for corners B and C are simultaneously advanced. For easier illustration and without loss of generality, it is to take FIG. 17B as an example. As shown in the FIG. 17B, the scanning of corner B is started from the first non-adjacent neighboring block located on the above side of the current block, in a bottom-to-up direction. The scanning of corner C is started from the first non-adjacent neighboring block located on the left side of the current block, in a right-to-left direction. Therefore, in the example shown in FIG. 17B, pairwise scanning may be defined as that the candidate positions of B and C are both advanced with one unit of step size, where one unit of step size is defined as the height of the current coding block for corner B and defined as the width of the current coding block for corner C.

[00329] In another embodiment, alternative scanning may be performed. In one example of alternative scanning, the candidate positions for corners B and C are alternatively advanced. At one step, only the position of B or C may be advanced, while the position of C or B is not changed. In one example, the position of corner B may be progressively increased from the first non-adjacent neighboring block to the distance at the maximum number of non- adjacent neighboring blocks, while the position of corner C remains at the first non-adjacent neighboring block. In the next round, the position of the corner C moves to the second non- adjacent neighboring block, and the position of the corner B is traversed from the first to the maximum value again. The rounds are continued until all combinations are traversed.

[00330] For unification purpose, the methods of defining scanning area and distance, scanning order, and scanning termination proposed for deriving inherited merge candidates may completely or partially reused for deriving constructed merge candidates. In one or more embodiments, the same methods defined for inherited merge candidate scanning, which include but no limited to scanning area and distance, scanning order and scanning termination, may be completely reused for constructed merge candidate scanning.

[00331] In some embodiments, the same methods defined for inherited merge candidate scanning, may be partly reused for constructed merge candidate scanning. FIG. 16 shows an example in this case. In FIG. 16, the block size of each non-adjacent neighboring blocks is same as the current block, which is similarly defined as inherited candidate scanning, but the whole process is a simplified version since the scanning at each distance is limited to be only one block.

[00332] FIGS. 17A-17B represent another example in this case. In FIGS. 17A-17B, both non-adjacent inherited merge candidates and non-adjacent constructed merge candidates are defined with the same block size as the current coding block, while the scanning order, scanning area, and scanning termination conditions may be defined differently.

[00333] In FIG. 17A, the maximum distance for left side non-adjacent neighbors is 4 coding blocks, while the maximum distance for above side non-adjacent neighbors is 5 coding blocks. Also, at each distance, the scanning direction is bottom-up for left side and right-to- left for above side. In FIG. 17B, the maximum distance of non-adjacent neighbors is 4 for both left side and above side. In addition, the scanning at a specific distance is unavailable because there is only one block at each distance. In FIG. 17A, the scanning operations within each distance may be terminated if M qualified candidates are identified. The value of M may be a predefined fixed value such as the value of 1 or any other positive integer, or a signaled value decided by the encoder, or a configurable value at the encoder or the decoder. In one example, the value of M may be the same as the merge candidate list size.

[00334] In FIGS. 17A-17B, the scanning operations at different distances may be terminated if N qualified candidates are identified. The value of N may be a predefined fixed value such as the value of 1 or any other positive integer, or a signaled value decided by the encoder, or a configurable value at the encoder or the decoder. In one example, the value of N may be the same as the merge candidate list size. In another example, the value of N may be the same as the value of M.

[00335] In both FIGS. 17A-17B, the non-adjacent spatial neighbors with closer distance to the current block may be prioritized, which indicates that non-adjacent spatial neighbors with distance i is scanned or checked before the neighbors with distance i+1, where i may be a non-negative integer representing a specific distance.

[00336] At a specific distance, up to two non-adjacent spatial neighbors are used, which means at most one neighbor from one side, e.g., the left and above, of the current block is selected for inherited or constructed candidate derivation, if available. As shown in FIG. 17A, the checking orders of the left side and above side neighbors are bottom-up and right-left, respectively. For FIG. 17B, this rule may be also applied, where the difference may be that at any specific distance there is only one option for each side of the current block.

[00337] For constructed candidates, as shown in the FIG. 17B, the positions of one left and above non-adjacent spatial neighbors are firstly determined independently. After that, the location of the top-left neighbor can be determined accordingly which can enclose a rectangular virtual block together with the left and above non-adjacent neighbors. Then, as shown in the FIG. 9, the motion information of the three non-adjacent neighbors is used to form the CPMVs at the top-left (A), top-right (B) and bottom-left (C) of the virtual block, which is finally projected to the current CU to generate the corresponding constructed candidates. [00338] In Step 2, the translational MVs at the positions of the selected candidates after step 1 are evaluated and an appropriate affine model maybe determined. For easier illustration and without loss of generality, FIG. 9 is used as an example again.

[00339] Due to factors such as hardware constrains, implementation complexity and different reference indexes, the scanning process may be terminated before enough number of candidates are identified. For example, the motion information of the motion field at one or more of the selected candidates after Step 1 may be unavailable.

[00340] If the motion information of all three candidates are available, the corresponding virtual coding block represents a 6-parameter affine model. If the motion information of one of the three candidates is unbailable, the corresponding virtual coding block represents a 4-parameter affine model. If the motion information of more than one of the three candidates is unbailable, the corresponding virtual coding block may be unable to represent a valid affine model.

[00341] In some embodiments, if the motion information at the top-left corner, e.g., the corner A in FIG. 9, of the virtual coding block is unavailable, or the motion information at both the top-right corner, e.g., the corner B in FIG. 9, and bottom-left comer, e.g., the corner C in the FIG. 9, is unavailable, the virtual block may be set to be invalid and unable to represent a valid model, then Step 3 and Step 4 may be skipped for the current iteration.

[00342] In some embodiments, if either the top-right corner, e.g., the corner B in the FIG. 9, or bottom-left corner, e.g., the corner C in FIG. 9, is unavailable, but not both are unavailable, the virtual block may represent a valid 4-parameter affine model.

[00343] In Step 3, if the virtual coding block is able to represent a valid affine model, the same projection process used for inherited merge candidate may be used.

[00344] In one or more embodiments, the same projection process used for inherited merge candidate may be used. In this case, a 4-parameter model represented by the virtual coding block from Step 2 is projected to a 4-parameter model for the current block, and a 6- parameter model represented by the virtual coding block from Step 2 is projected to a 6- parameter model for the current block.

[00345] In some embodiments, the affine model represented by the virtual coding block from Step 2 is always projected to a 4-parameter model or a 6-parameter model for the current block.

[00346] Note that according to equation (5) and (6), there may be two types of 4- parameter affine model, where the type A is that the top-left corner CPMV and top-right corner CPMV, termed as Vo and Vi, are available, and the type B is that the top-left corner CPMV and bottom-left corner CPMV, termed as Vo and V2, are available.

[00347] In one or more embodiments, the type of the projected 4-parameter affine model is the same type of the 4-parameter affine model represented by the virtual coding block. For example, the affine model represented by the virtual coding block from Step 2 is type A or B 4-parameter affine model, then the projected affine model for the current block is also type A or B respectively.

[00348] In some embodiments, the 4-parameter affine model represented by the virtual coding block from Step 2 is always projected to the same type of 4-parameter model for the current block. For example, the type A or B of 4-parameter affine model represented by the virtual coding block is always projected to the type A 4-parameter affine model.

[00349] In Step 4, based on the projected CPMVs after Step 3, in one example, the same candidate generation process used in the current WC or AVS standards may be used. In another embodiment, the temporal motion vectors used in the candidate generation process for the current WC or AVS standards may be not used for the non-adjacent neighboring blocks based derivation method. When the temporal motion vectors are not used, it indicates that the generated combinations do not contain any temporal motion vectors.

[00350] In Step 5, any newly generated candidate after Step 4 may go through a similarity check against all existing candidates that are already in the merge candidate list. The details of similarity check are already described in the section of “Affine merge candidate pruning.” If the newly generated candidate is found to be similar with any existing candidate in the candidate list, this newly generated candidate is removed or pruned.

[00351] In some embodiments, a virtual coding block is formed by determining three corner points A, B and C, and then the translational MVs of the 4x4 blocks located at the three corners are used to represent an affine model for the virtual coding block. At the end, the affine model of the virtual coding block is projected to the current coding block. This whole process may be used to derive the first type of affine candidates constructed from non-adjacent spatial neighbors (e.g., the sub-blocks located by the three corner points A, B and C are non- adjacent spatial neighbors). In some embodiments, this method may be applied to an affine mode, such as affine merge mode and affine AMVP mode, and this method may be also applied to regular mode, such as regular merge mode and regular AMVP mode, because the projected affine model can be used to derive a translational MV based on a specific position (e.g., the center position) inside of a prediction block or a coding block.

[00352] Inheritance Based Derivation Method for Affine Constructed Merge Candidates [00353] For each affine inherited candidate, all the motion information is inherited from one selected spatial neighboring block which is coded in affine mode. The inherited information includes CPMVs, reference indexes, prediction direction, affine model type, etc. On the other hand, for each affine constructed candidate, all the motion information is constructed from two or three selected spatial or temporal neighboring blocks, while the selected neighboring blocks could be not coded in affine mode and only translational motion information is needed from the selected neighboring blocks.

[00354] In this section, a new candidate derivation method which combines the features of inherited candidates and constructed candidates is disclosed.

[00355] In some embodiments, the combination of inheritance and construction may be realized by separating the affine model parameters into different groups, where one group of affine parameters are inherited from one neighboring block, while other groups of affine parameters are inherited from other neighboring blocks. [00356] In one example, the parameters of one affine model may be constructed from two groups. As shown in Equation (3), an affine model may contain 6 parameters, including a, b, c, d, e and f. The translational parameters {a, b] may represent one group, while the non-translational parameters {c, d, e, f} may represent another group. With this grouping method, the two groups of parameters may be independently inherited from two different neighboring blocks in the first step and then concatenated/constructed to be a complete affine model in the second step. In this case, the group with non-translational parameters has to be inherited from one affine coded neighboring block, while the group with translational parameters may be from any inter-coded neighboring block, which may or may not be coded in affine mode. Note that the affine coded neighboring block may be selected from adjacent affine neighboring blocks or non-adjacent affine neighboring blocks based on previously proposed scanning methods for affine inherited candidates, such as the methods shown in FIG. 17A, that is the scanning method/rule including the scanning area and distance, scanning order, and scanning termination used in the Section of “Non-Adjacent Neighbor Based Derivation Process for Affine Inherited Merge Candidates” while the scanning method may be performed on both adjacent neighbor blocks or non-adjacent neighbor blocks. Alternatively, the affine coded neighboring block may be not physically existed, but virtually constructed from regular inter-coded neighboring blocks, such as the methods shown in FIG. 17B, that is the scanning method/rule including the scanning area and distance, scanning order, and scanning termination used in the Section of “Non-Adjacent Neighbor Based Derivation Process for Affine Constructed Merge Candidates.”

[00357] In some examples, the neighboring blocks associated with each group may be determined in different ways. In one method, the neighboring blocks for different groups of parameters may be all from non-adjacent neighboring/neighbor areas, while the scanning methods may be similarly designed as the previously proposed methods for non-adjacent neighbor based derivation process. In another method, the neighboring blocks for different groups of parameters may be all from adjacent neighboring/neighbor areas, while the scanning methods may be the same as the current WC or AVS video standards. In another method, the neighboring blocks for different groups of parameters may be partly from adjacent areas and partly from non- adjacent neighboring/neighbor areas.

[00358] When neighboring blocks are scanned from non-adjacent neighboring/neighbor areas for constructing current type of candidates, the scanning process may be differently performed from the non-adjacent neighbor based derivation process for affine inherited candidates. In one or more embodiments, the scanning area, distance and order may be similarly defined, but the scanning termination rule may be differently specified. For example, the non-adjacent neighboring blocks may be exhaustively scanned within a defined maximum distance at each area. In this case, all non- adjacent neighboring blocks within a distance may be scanned by following a scanning order. In some embodiments, the scanning area may be different. For example, in addition to the left and above area, the right bottom adjacent and non-adjacent area of the current coding block may be scanned to determine neighbors for generating translational or/and non-translational parameters. In addition, the neighbors scanned at the right bottom area may be used to find collocated temporal neighbors, instead of spatial neighbors. One scanning criteria may be conditionally based on whether the right-bottom collocated temporal neighbor(s) is/are already used for generating affine constructed neighbors. If used already, the scanning is not performed, otherwise the scanning is performed. Alternatively, if used already, which means the rightbottom collocated temporal neighbor(s) is/are available, the scanning is performed, otherwise the scanning is not performed.

[00359] When several groups of affine parameters are combined to construct a new candidate, there may be several rules to be followed. The first is eligibility criteria. In one example, the associated neighboring block or blocks for each group may be checked whether to use the same reference picture for at least one direction or both directions. In another example, the associated neighboring block or blocks for each group may be checked whether use the same precision/resolution for motion vectors. [00360] When certain criteria are checked, the first X associated neighboring block(s) for each group may be used. The value of X may be defined as the same or different values for different groups of parameters. For example, the first 1 or 2 neighboring blocks containing non-translational affine parameters may be used, while the first 3 or 4 neighboring blocks containing translational affine parameters may be used.

[00361] The second is construction formula. In one example, the CPMVs of the new candidates may be derived in equation below:

[00362] where (x, y) is a corner position within the current coding block (e.g., (0, 0) for top-left corner CPMV, (width, 0) for top-right corner CPMV), {c, d, e, f } is one group of parameters from one neighboring block, { a, b } is another group of parameters from another neighboring block.

[00363] In another example, the CPMVs of the new candidates may be derived in below equation:

CPMV _x (x, y) = a + c * (x + div) -I- d * (y -I- dh) CPMV _y (x, y) = b + e * (x + div) + f * (y + dh) where the (div, dh) is the distance between the top-left comer of the current coding block and the top-left corner of one of the associated neighboring block(s) for one group of parameters, such as the associated neighboring block of the group of { a, b }. The definitions of the other parameters in this equation are the same as the example above. The parameters may be grouped in another way: (a, b, c, d, e,f) are formed as one group, while the (div, dh) are formed as another group. And the two groups of parameters are from two different neighboring blocks. Alternatively, the value of (div, dh) may be predefined as fixed values such as (0, 0) or at any constant values, which is not dependent on the distance between a neighboring block and the current block.

[00365] FIG. 18 shows an example of inheritance based derivation method for deriving affine constructed candidates. In FIG. 18, there are three steps to derive an affine constructed candidate. In Step 1, according to a specific grouping strategy, the encoder or the decoder may perform scanning of adjacent and non-adjacent neighboring blocks for each group. In the case of the FIG. 18, two groups are defined, where neighbor 1 is coded in affine mode and provides non-translational affine parameters, while neighbor 2 provides translational affine parameters. Neighbor 1 may be obtained according to the process in the Section of “Non-Adjacent Neighbor Based Derivation Process for Affine Inherited Merge Candidates” as shown in FIGS. 15A-15D and 17A while neighbor 1 may be an adjacent or non-adjacent neighbor block of the current block. Furthermore, neighbor 2 may be obtained according to the process as shown in FIGS. 16 and 17B.

[00366] In some embodiments, the neighbor 1, which is coded in the affine mode, may be scanned from adjacent or/and non-adjacent areas, by following above proposed scanning methods. In some embodiments, the neighbor 2, which is coded in the affine or a non-affine mode, may be also scanned from adjacent or non-adjacent areas. For example, the neighbor 2 may be from one of the scanned adjacent or non-adjacent areas if the motion information is not already used for deriving some affine merge or AMVP candidates, or from right-bottom positions of the current block if a collocated TMVP candidate at this position is available or/and already used for deriving some affine merge or AMVP candidates. Alternatively, a small coordinate offset (e.g., +1 or +2 or -1 or -2 for vertical or/and horizontal coordinates) may be applied when determining the position of neighbor 2 in order to provide a slightly diversified motion information for constructing a new candidate.

[00367] In Step 2, with the parameters and positions decided in Step 1, a specific affine model may be defined, which can derive different CPMVs according to the coordinate (x, y) of a CPMV. For examples, as shown in FIG. 18, the non-translational parameters {c, d, e, f] may be obtained based on neighbor 1 obtained in Stepl, and the translational parameters {a, b] may be obtained based on neighbor 2 obtained in Step 1. Furthermore, the distance parameters Aw, Ah may thus obtained based on the position of the current block and the position of neighbor 2 (x ₂ ,y ₂ ). The distance parameters Aw, Ah may respectively indicate a horizontal distance and a vertical distance between the current block and neighbor 1 or neighbor 2. For example, the distance parameters Aw, Ah may respectively indicate the horizontal distance ( ^ — x ₂ ) between the current block and neighbor 2 and the vertical distance (y _± — y ₂ ) between the current block and neighbor 2. Specifically, Aw = x — x ₂ and Ah = y _x - y ₂ .

[00368] In Step 3, two or three CPMVs are derived for the current coding block, which can be constructed to form a new affine candidate

[00369] In some embodiments, other prediction information may be further constructed. The prediction direction (e.g., bi or uni-predicted) and indexes of reference pictures may be the same as the associated neighboring blocks if neighboring blocks are checked to have the same directions and/or reference pictures. Alternatively, the prediction information is determined by reusing the minimum overlapped information among the associated neighboring blocks from different groups. For example, if only the reference index of one direction from one neighboring block is the same as the reference index of the same direction of the other neighboring block, the prediction direction of the new candidate is determined as uni-prediction, and the same reference index and direction are reused.

[00370] In some embodiments, an affine model may be constructed by combining model parameters from different inheritances. In one example, the translational model parameters may be inherited from translational blocks (e.g., from adjacent or/and non-adjacent spatial neighboring 4x4 blocks), while the non-translational model parameters may be inherited from affine coded blocks (e.g., from adjacent or/and non-adjacent spatial neighboring affine coded blocks). Alternatively, the non-translational model parameters may be inherited from historically coded affine blocks instead of explicitly scanned non-adjacent spatial neighboring affine coded blocks, while the historically coded affine blocks may be adjacent or nob-adjacent spatial neighbors. This whole process may be used to derive the second type of affine candidates constructed from non-adjacent spatial neighbors (e.g., the non-translational model parameters maybe inherited from non-adjacent spatial neighbors). In some embodiments, this method may be applied to an affine mode, such as affine merge mode and affine AMVP mode, and this method may be also applied to regular mode, such as regular merge mode and regular AMVP mode, because the generated affine model can be used to derive a translational MV based on a specific position (e.g., the center position) inside of a prediction block or coding block.

[00371] HMVP Based Derivation Method for Affine Constructed Merge Candidates

[00372] In the case of adjacent neighbor based derivation process, which is already defined in the current video standards WC and AVS and described in the sections above and FIG. 7, a fixed order of scanning on adjacent neighbors is performed to identify two or three adjacent neighboring blocks. In the case of non-adjacent neighbor based derivation process, as proposed in earlier section and FIG. 17B, two non-adjacent neighbors are identified during another fixed order of scanning. In other words, for both the adjacent and non-adjacent neighbor based derivation methods, certain depth of local scanning is inevitable to identify a number of neighbors. This scanning process is dependent on the local buffering around each current block and also incurs certain amount of computation complexity.

[00373] On the other hand, the HMVP merge mode is already adopted in the current WC and AVS, where the translational motion information from neighboring blocks are already stored in a history table, as described in the introduction section. In this case, the scanning process may be replaced by searching the HMVP table.

[00374] Therefore, for the previously proposed non-adjacent neighbor based derivation process and inheritance based derivation process, the translational motion information may be obtained from HMVP table, instead of the scanning method as shown in the FIG. 17B and FIG. 18. However, in order to derive affine constructed candidate afterwards, the position information, width, height and reference information are also needed, which may be accessible if the current HMVP table can be modified. Therefore, it is proposed to extend the HMVP table to store additional information in addition to the motion information of each history neighbor. In one embodiment, the additional information may include positions of an affine or non-affine neighboring blocks, or affine motion information such as CPMVs or

1 equivalent regular motion derived from CPMVs (e.g.., this regular motion may be from the internal sub-blocks of an affine coded neighboring block) reference index, etc.

[00375] In one or more examples, the above provided non-adjacent neighbor based derivation process and inheritance based derivation process for affine mode may be replaced by modifying the existing HMVP table.

[00376] In some examples, the translational motion of the derived affine model may be directly obtained from the existing HMVP table, where the translational motion is previously saved when neighboring blocks are previously coded at regular inter mode. Alternatively, the translational motion of the derived affine model may be still obtained from the existing HMVP, where the translational motion is previously saved when neighboring blocks are previously coded at affine mode.

[00377] In some examples, the non-translational motion of the derived affine model is obtained from the existing HMVP table, where the non-translational motion is previously saved when neighboring blocks are previously code at affine mode. Alternatively, the reused non-translational motion may not be directly saved while original CPMVs of the previously coded affine neighboring blocks are saved. In this case, the existing HMVP table is updated with not only original CPMVs but also the position and size information of the previously coded affine neighboring blocks.

[00378] In some other examples, the above proposed non-adjacent neighbor based derivation process and inheritance based derivation process for affine mode may be replaced by creating one or more new HMVP tables.

[00379] In some examples, the translation motion and non-translational motion of the derived affine model may be similarly obtained as the method of modifying the existing HMVP table

[00380] Candidate Derivation Method for Affine AMVP And Regular Merge Mode [00381] As described in the sections above, for affine AMVP mode, an affine candidate list is also needed for deriving CPMV predictors. As a result, all the above proposed derivation methods may be similarly applied to affine AM VP mode. The only difference is that when the above proposed derivation methods are applied in AMVP, the selected neighboring blocks must have the same reference picture index as the current coding block.

[00382] For regular merge mode, a candidate list is also constructed, but with only translational candidate MVs, not CPMVs. In this case, all the above proposed derivation methods can still be applied by adding an additional derivation step. In this additional derivation step, it is to derive a translation MV for the current block, which may be realized by selecting a specific pivot position (x, y) within the current block and then follow the same equation (3). In other words, for deriving CPMVs of an affine block, the three corner positions of the block are used as the pivot position (x, y) in equation (3), while for deriving translation MVs of regular inter-coded block, the center position of the block may be used as the pivot position (x, y) in equation (3). Once the translational MV is derived for the current block, it can be inserted to the candidate list as other candidates.

[00383] When new candidates are derived based on above proposed methods for affine AMVP and regular merge mode, the placement of the new candidates may be reordered.

[00384] In one embodiment, the newly derived candidates may be inserted into the affine AMVP candidate list by following the order as below:

(1) Inherited from adjacent spatial neighbors;

(2) Constructed from adjacent spatial neighbors;

(3) Inherited from non-adjacent spatial neighbors;

(4) Constructed from non-adjacent spatial neighbors;

(5) Translational MVs from adjacent spatial neighbors;

(6) Temporary MVs from adjacent temporal neighbors; and

(7) Zero MVs.

[00385] In another embodiment, the newly derived candidates may be inserted into the affine AMVP candidate list by following the order as below:

(1) Inherited from adjacent spatial neighbors; (2) Constructed from adjacent spatial neighbors;

(3) Inherited from non-adjacent spatial neighbors;

(4) Translational MVs from adjacent spatial neighbors;

(5) Constructed from non-adjacent spatial neighbors;

(6) Temporary MVs from adjacent temporal neighbors; and

(7) Zero MVs.

[00386] In another embodiment, the newly derived candidates may be inserted into the affine AMVP candidate list by following the order as below:

(1) Inherited from adjacent spatial neighbors;

(2) Constructed from adjacent spatial neighbors;

(3) Translational MVs from adjacent spatial neighbors;

(4) Inherited from non-adjacent spatial neighbors;

(5) Constructed from non-adjacent spatial neighbors;

(6) Temporary MVs from adjacent temporal neighbors; and

(7) Zero MVs.

[00387] In another embodiment, the newly derived candidates may be inserted into the affine AMVP candidate list by following the order as below:

(1) Inherited from adjacent spatial neighbors;

(2) Constructed from adjacent spatial neighbors;

(3) Translational MVs from adjacent spatial neighbors;

(4) Temporary MVs from adjacent temporal neighbors;

(5) Inherited from non-adjacent spatial neighbors;

(6) Constructed from non-adjacent spatial neighbors; and

(7) Zero MVs.

[00388] In another embodiment, the newly derived candidates may be inserted into the affine AMVP candidate list by following the order as below:

(1) Inherited from adjacent spatial neighbors; (2) Constructed from adjacent spatial neighbors;

(3) Translational MVs from adjacent spatial neighbors;

(4) Temporary MVs from adjacent temporal neighbors;

(5) Inherited from non-adjacent spatial neighbors; and

(6) Zero MVs.

[00389] In another embodiment, the newly derived candidates may be inserted into the affine AMVP candidate list by following the order as below:

(1) Inherited from adjacent spatial neighbors;

(2) Constructed from adjacent spatial neighbors;

(3) Translational MVs from adjacent spatial neighbors;

(4) Translational MVs from temporal neighbors;

(5) Inherited from non-adjacent spatial neighbors; and

(6) Zero MVs.

[00390] Note that the candidates constructed from non-adjacent spatial neighbors may be referred to as the first type or/and the second type of candidates constructed from non- adjacent spatial neighbors.

[00391] In another embodiment, the newly derived candidates may be inserted into the regular merge candidate list by following the order as below:

(1) Spatial MVP from adjacent spatial neighbors;

(2) Temporal MVP from adjacent collocated neighbors;

(3) Spatial MVP from non-adjacent spatial neighbors;

(4) Inherited MVP from non-adjacent spatial affine neighbors;

(5) Constructed MVP from non-adjacent spatial neighbors;

(6) History-based MVP from a FIFO table;

(7) Pairwise average MVP; and

(8) Zero MVs.

[00392] Reordering of Affine Merge Candidate List [00393] In one embodiment, the non-adjacent spatial merge candidates may be inserted into the affine merge candidate list by following the order below: 1. Subblock-based Temporal Motion Vector Prediction (SbTMVP) candidate, if available; 2. Inherited from adjacent neighbors; 3. Inherited from non-adjacent neighbors; 4. Constructed from adjacent neighbors; 5. Constructed from non-adjacent neighbors; 6. Zero MVs.

[00394] In another embodiment, the non-adjacent spatial merge candidates may be inserted into the affine merge candidate list by following the order below: 1. SbTMVP candidate, if available; 2. Inherited from adjacent neighbors; 3. Constructed from adjacent neighbors; 4. Inherited from non-adjacent neighbors; 5. Constructed from non-adjacent neighbors; 6. Zero MVs.

[00395] In another embodiment, the non-adjacent spatial merge candidates may be inserted into the affine merge candidate list by following the order below: 1. SbTMVP candidate, if available; 2. Inherited from adjacent neighbors; 3. Constructed from adjacent neighbors; 4. One set of zero MVs; 5. Inherited from non-adjacent neighbors; 6. Constructed from non-adjacent neighbors; 7. Remaining zero MVs, if the list is still not full.

[00396] In another embodiment, the non-adjacent spatial merge candidates may be inserted into the affine merge candidate list by following the order below: 1. SbTMVP candidate, if available; 2. Inherited from adjacent neighbors; 3. Inherited from non-adjacent neighbors with distance smaller than X; 4. Constructed from adjacent neighbors; 5. Constructed from non-adjacent neighbors; 6. Constructed from inherited translational and non-translational neighbors; 7. Zero MVs, if the list is still not full.

[00397] In another embodiment, the non-adjacent spatial merge candidates may be inserted into the affine merge candidate list by following the order below: 1. SbTMVP candidate, if available; 2. Inherited from adjacent neighbors; 3. Inherited from non-adjacent neighbors; 4. The first candidate constructed from adjacent neighbors; 5. The first X candidates constructed from inherited translational and non-translational neighbors; 6. Constructed from non-adjacent neighbors; 7. Other Y candidates constructed from inherited translational and non-translational neighbors; 8. Zero MVs, if the list is still not full.

[00398] In some examples, where the value of X and Y may be a predefined fixed value such as the value of 2, or a signaled value (sequence/slice/block/CTU level signaled parameters) received by the decoder, or a configurable value at the encoder/decoder, or a dynamically decided value (e.g., X <= 3, Y <=3) according to the number of available neighbors on the left and above of each individual coding block, or any combination of methods determining the value of X and Y. In one example, the value of X may be the same as the value of Y. In another example, the value of X may be different from the value of Y.

[00399] In another embodiment, the non-adjacent spatial merge candidates may be inserted into the affine merge candidate list by following the order below: 1. SbTMVP candidate, if available; 2. Inherited from adjacent neighbors; 3. Inherited from non-adjacent neighbors with distance smaller than X; 4. Constructed from adjacent neighbors; 5. Constructed from non-adjacent neighbors with distance smaller than Y; 6. Inherited from non- adjacent neighbors with distance bigger than X; 7. Constructed from non-adjacent neighbors with distance bigger than Y; 8. Zero MVs. In this embodiment, the value X and Y may be a predefined fixed value such as the value of 2, or a signaled value decided by the encoder, or a configurable value at the encoder or the decoder. In one example, the value of X may be the same as the value of Y. In another example, the value of N may be different from the value of M.

[00400] In some embodiments, if a new candidate is derived by using the inheritance based derivation method which constructs CPMVs by combining affine motion and translational MV, the placement of this new candidate may be dependent on the placement the other constructed candidates.

[00401] In one embodiment, for different constructed candidates, the reordering of the affine merge candidate list may follow the order as below:

(1) Constructed from adjacent spatial neighbors; (2) Constructed from combining adjacent spatial affine neighbors and translational MVs;

(3) Constructed from non-adjacent spatial neighbors; and

(4) Constructed from combining non-adjacent spatial affine neighbors and translational MVs.

[00402] In another embodiment, for different constructed candidates, the reordering of the affine merge candidate list may follow the order below:

(1) Constructed from adjacent spatial neighbors;

(2) Constructed from non-adjacent spatial neighbors;

(3) Constructed from combining adjacent spatial affine neighbors and translational MVs; and

(4) Constructed from combining non-adjacent spatial affine neighbors and translational MVs.

[00403] In another one or more embodiments, the reordering of the affine merge candidates may be partially or completely interleaved among different categories of candidates (e.g., interleaving may indicate that the candidates from the same category may not be adjacently placed in the candidates list). In some embodiment, there may be seven categories of affine merge candidates placed in the affine merge candidate list:

(1) SbTMVP candidate, if available;

(2) Inherited candidates from adjacent neighbors;

(3) Inherited candidates from non-adjacent neighbors;

(4) Constructed candidates from adjacent neighbors;

(5) The second type of constructed candidates from non-adjacent neighbors;

(6) The first type of constructed candidates from non-adjacent neighbors; and

(7) Zero MVs;

The specific order discussed above may be applied in any candidate list including an affine AMVP candidate list, a regular merge candidate list, and an affine merge candidate list. [00404] As the candidates put in the later positions of the affine merge list may cost higher signaling overhead if selected and signaled by the encoder, the order of the above different categories of candidates may be designed in different methods.

[00405] In one or more embodiments, the order of the candidates may remain the same as the above insertion order. An adaptive reordering method may be applied to reorder the candidates afterwards; the adaptive reordering may be template based methods (ARMC) or non-template based method such as bilateral matched based methods.

[00406] In one or more embodiments, the order of the candidates may be reordered in a specific pattern. The specific pattern may be applied in any candidate list including an affine AMVP candidate list, a regular merge candidate list, and an affine merge candidate list.

[00407] In some embodiments, the reordering pattern may depend on the number of available candidates for each category.

[00408] In one example, the reordering pattern may be defined as below:

(1) SbTMVP candidate, if available;

(2) Inherited candidates from adjacent neighbors;

(3) Constructed candidates from adjacent neighbors;

(4) The first X inherited candidates from non-adjacent neighbors (e.g., X may be a prefixed number such as 1 or a signaled number);

(5) The first Y constructed candidates from the second type of constructed candidates from non-adjacent neighbors (e.g., Y may be similarly defined as X);

(6) The first Z constructed candidates from the first type of constructed candidates from non-adjacent neighbors (e.g., Z may be similarly defined as X);

(7) The remaining inherited candidates from non-adjacent neighbors (e.g., X may be a prefixed number such as 1 or a signaled number); (8) The remaining constructed candidates from the second type of constructed candidates from non-adjacent neighbors (e.g., Y may be similarly defined as

X);

(9) The remaining constructed candidates from the first type of constructed candidates from non-adjacent neighbors (e.g., Z may be similarly defined as X); and

(10) Zero MVs.

[00409] In one or more embodiments, the reordering pattern may be an interleaved method which may merge different candidates from different categories. In one example, the interleaved pattern may be defined as below:

(1) SbTMVP candidate, if available;

(2) Inherited candidates from adjacent neighbors;

(3) The first XI constructed candidates from adjacent neighbors;

(4) The first Y1 constructed candidates from the second type of constructed candidates from non-adjacent neighbors;

(5) The first Z1 inherited candidates from non-adjacent neighbors;

(6) The first KI constructed candidates from the first type of constructed candidates from non-adjacent neighbors;

(7) The first X2 constructed candidates from adjacent neighbors;

(8) The first Y2 constructed candidates from the second type of constructed candidates from non-adjacent neighbors;

(9) The first Z2 inherited candidates from non-adjacent neighbors;

(10) The first K2 constructed candidates from the first type of constructed candidates from non-adjacent neighbors;

(11) >

(12) The first Xi constructed candidates from adjacent neighbors; (13) The first Yi constructed candidates from the second type of constructed candidates from non-adjacent neighbors;

(14) The first Zi inherited candidates from non-adjacent neighbors;

(15) The first Ki constructed candidates from the first type of constructed candidates from non-adjacent neighbors; and

(16) Zero MVs.

The value of (Xi, Yi, Zi, Ki) may be a prefixed number such as 1 or a signaled number. If the number of available candidates of one category is smaller than other categories, the position of the candidates for this category is skipped and the remaining available candidates of other categories would take over this position.

[00410] In one or more embodiments, the reordering pattern may be a combined version which considers both availability and interleaving method. In one example, the combined pattern may be defined as below:

(1) SbTMVP candidate, if available;

(2) Inherited candidates from adjacent neighbors;

(3) The first XI constructed candidates from adjacent neighbors;

(4) All the available constructed candidates from the second type of constructed candidates from non-adjacent neighbors;

(5) The remaining constructed candidates from adjacent neighbors;

(6) The first ZI inherited candidates from non-adjacent neighbors;

(7) The first KI constructed candidates from the first type of constructed candidates from non-adjacent neighbors;

(8) ...

(9) The first Zi inherited candidates from non-adjacent neighbors;

(10) The first Ki constructed candidates from the first type of constructed candidates from non-adjacent neighbors; and

(11) Zero MVs. [00411] In some examples, the above reordering methods may be selected and applied according to different factors.

[00412] In one or more examples, the reordering methods may be selected based on the types of the video frames/slices. For example, for low-delay pictures or slices, all the candidates of the first type of constructed candidates from non-adjacent neighbors may be placed after all the constructed candidates from adjacent neighbors. While for non-low-delay pictures or slices, the first KI candidates of the first type of constructed candidates from non-adjacent neighbors may be placed after the first K2 constructed candidates from adjacent neighbors, and the remaining candidates of the first type of constructed candidates from non-adjacent neighbors may be placed after the remaining constructed candidates from adjacent neighbors. [00413] Improved Reordering of Affine Candidate List

[00414] Based on the above proposed candidate derivation methods, one or more candidates may be derived for an existed affine merge candidate list, or an affine AMVP candidate list, or a regular merge candidate list, where the size of the corresponding list may be statically (e.g., configurable size) or adaptively (e.g., dynamically changed according to availability at encoder and then signaled to decoder) adjusted. Note that, when one or more new candidates is/are derived for regular merge candidate list, the new candidates are firstly derived as affine candidates, and then converted to translational motion vectors by using a pivot position (e.g., center sample or pixel position) within a coding block and associated affine models before insert into the regular merge candidate list.

[00415] In one or more embodiments, an adaptive reordering method such as ARMC may be applied to one or more of the above candidate lists after the candidate lists are updated or constructed by adding some new candidates which are derived by above proposed candidate derivation methods.

[00416] In another embodiment, a temporal candidate list may be created first, where the temporal candidate list may have a larger size than the existed candidate list (e.g., affine merge candidate list, affine AMVP candidate list, regular merge candidate list). Once the temporal candidate list is constructed by adding newly derived candidates and statically ordered by using above proposed insertion methods, an adaptive reordering method such as ARMC may be applied to reorder the temporal candidate list. After the adaptive reordering, the first N candidates of the temporal candidate list are inserted to the existed candidate lists., where the value of N may be a fixed or configurable value. In one example, the value of N may be the same as the size of the existed candidate list, where the selected N candidates from of the temporal candidate list are located.

[00417] In the above application scenarios of applying an adaptive reordering method such as ARMC, following methods may be used to improve the performance or/and reduce the complexity of the applied reordering methods.

[00418] In some embodiments, when a template matching cost is used for reordering different candidates, a cost function such as the sum of absolute differences (SAD) between samples of a template of the current block and their corresponding reference samples may be used. The reference samples of the template may be located by the same motion information of the current block. In case fractional motion information is used for the current block, an interpolation filtering process may be used to generate prediction samples of the template. Since the generated prediction samples are just used to comparing the motion accuracy between different candidates, not for final block reconstructions, the prediction accuracy of the template samples may be relaxed by using an interpolation filter with smaller tap. For example, in the case of adaptively reordering affine merge candidate list, a 2-tap or 4-tap any other shorter length (e.g., 6-tap, 8-tap) interpolation filter may be used to generate prediction samples for the selected template of the current block. Or even the nearest integer samples (completely skip the interpolation filtering process) may be used as the prediction samples of the template. An interpolation filter with smaller tap may be similarly used when a template matching method is used to adaptively reorder the candidates in other candidate list such as regular merge candidate list or affine AMVP candidate list. [00419] In some embodiments, when a template matching cost is used for reordering different candidates, a cost function such as the SAD between samples of a template of the current block and their corresponding reference samples may be used. The corresponding reference samples may be located at integer positions or fractional positions. When fractional positions are located, a certain level of prediction accuracy may be achieved by performing an interpolation filter process. Due to the limited prediction accuracy, the calculated matching costs for different candidates may contain noise level differences. To reduce the impact of the noise level cost difference, the calculated matching costs may be adjusted by removing a few bits of the least significance bits before candidate sorting process.

[00420] In some embodiments, a candidate list may be padded with zero MVs at the end of each list, if not enough candidates could be derived by using different derivation methods. In this case, the candidate cost may be only calculated for the first zero MV, while the remaining zero MVs may be statically assigned with an arbitrarily large cost value, such that these repeated zero MVs are placed at the end of the corresponding candidate list.

[00421] In some embodiments, all zero MVs may be statically assigned with an arbitrarily large cost value, such that all zero MVs are placed at the end of the corresponding candidate list.

[00422] In some embodiments, an early termination method may be applied for a reordering method to reduce complexity at the decoder side.

[00423] In one or more embodiments, when a candidate list is constructed, different types of candidates may be derived and inserted into the list. If one candidate or one type of candidates is not participated in the reorder process, but selected and signaled to the decoder, the reordering process, which is applied to other candidates, may be early terminated. In one example, in the case of applying ARMC for the affine merge candidate list, the SbTMVP candidate may be excluded from the reordering process. In this case, if the signaled merge index value for an affine coded block indicates a SbTMVP candidate at the decoder side, the ARMC process may be skipped or early terminated for this affine block. [00424] In another embodiment, if one candidate or one type of candidates is not participated in the reorder process, but not selected and signaled to the decoder, both the derivation process and the reorder process for this specific candidate or this specific type of candidates may be skipped. Note that the skipped derivation process and reordering process are only applied to the specific candidate or the specific type of candidates, while the remaining candidates or types of candidates are still performed, where the derivation process is skipped indicates that the related operations of deriving the specific candidate or this specific type of candidates are skipped, but the predefined list position (e.g., according to a predefined insertion order) of the specific candidate or this specific type of candidates may be still kept, just the candidate content such as the motion information may be invalid due to skipped derivation process. Similarly, during the reordering process, the cost calculation of this specific candidate or this specific type of candidates may be skipped and the list position of this specific candidate or this specific type of candidates may be not changed after reordering other candidates.

[00425] Motion Information Storage

[00426] When non-adjacent spatial neighbors are scanned based on above proposed candidate derivation methods, the selected non-adjacent spatial neighbors may be affine coded blocks or non-affine coded blocks (e.g., regular inter AMVP or merge coded blocks). In the case of non-affine coded blocks, the motion information may include translational MVs and corresponding reference index at each direction. In the case of affine coded blocks, the motion information may include CPMVs and corresponding reference index at each direction, and also the positions and the sizes of the affine coded blocks.

[00427] For either affine-coded blocks or non-affine-coded blocks, the motion information of these blocks may need to be saved in a memory once these blocks have been coded. In order to save memory usage, the non-adjacent spatial neighbors may be restricted to a certain area. [00428] As shown in FIG. 21, the allowed non-adjacent area for scanning non-adjacent spatial neighboring blocks may be restricted to a limited area size.

[00429] In one or more embodiments, the restricted area may be applied to affine or non- affine spatial neighboring blocks.

[00430] The size of the allowed non-adjacent area may be defined according to the size of current CTU, e.g., integer (e.g., 1 or 2 or other integer) or fractional number (e.g., 0.5 or 0.25 or other fractional number) of current CTU size.

[00431] The size of the allowed non-adjacent area may be defined according to a fixed number of pixels or samples, e.g., 128 samples on the above of the current CTU or/and on the left of the current CTU.

[00432] The size (e.g., according to the CTU size or number of samples) may be a prefixed value or a signaled value determined at the encoder and carried in the bit-stream.

[00433] In some other examples, the size of the restricted area may be separately defined for top and left non-adjacent neighboring blocks.

[00434] In one example, the above non-adjacent neighboring blocks may be restricted to be within the current CTU, or outside of the current CTU but within at most fixed number samples/pixels away from the top of the current CTU such that no additional line buffer is needed for saving the motion information of above non-adjacent neighboring blocks. For example, the fixed number may be defined as 8, if 8 sample rows of neighboring/neighbor area away from the current CTU top is already covered by the existing line buffer.

[00435] In another example, the left non-adjacent neighboring blocks may be restricted to be within the current CTU, or outside of the current CTU but within a predefined or a signaled number of samples/pixels away from the left boundary of the current CTU.

[00436] In some examples, as shown in FIGS. 27A-27B, the allowed non-adjacent area (for either non-adjacent affine neighbors or non-affine neighbors) above the current CU may have large memory cost if the allowed non-adjacent area is beyond the current CTU. In this case, the actual memory cost is proportionally increased with the picture width and the maximum allowable scanning distance in the vertical direction. To reduce the line buffer (i.e., the above non-adjacent area beyond the current CTU) cost, the height of the above non- adjacent area outside of the current CTU may be limited to a value of h, as shown in FIG. 27A. In some examples, this value of h may be configurable or signaled to decoder. In case the affine motion and non-affine motion are stored in a separate buffer, for the example shown in FIG. 27B, there may be different methods to save the motion in the line buffer as follows.

[00437] In one example, the line buffer used to store affine motion may indicate that the buffer area where the CU B is located is set to be invalid since CU B is not affine CU.

[00438] In another example, the line buffer used to store affine motion may indicate that the buffer area where the CU B is located is set to be valid and the affine motion is copied from CU A, since CU A is CU B’s adjacent affine neighbor.

[00439] In FIGS. 27A-27B, the height value h and the width value w may be set to be multiples of 4 for easier implementation. In one example, the value h and w may be set to be the minimum value (e.g., 4) of a non-affine CU. When the allowed non-adjacent area for scanning non-adjacent spatial neighboring blocks is restricted to a limited area size, such as the examples in FIGS. 21 and 27A-27B, the scanned non-adjacent neighbor position may be out of the allowed non-adjacent area. In this case, different methods may be used to solve this issue as follows.

[00440] In one example, the scanning process may indicate that this scanned position has no valid neighbor information.

[00441] In another example, the scanning process may project or clip this out-of-range position to another position which is within the allowed non-adjacent area. As shown in FIG. 28, there are two positions 2801 (i.e., the two dotted spots 2801) out of the range of the allowable non-adjacent area. These two positions 2801 are projected/clipped to another two positions 2802 which are at the same vertical/horizontal coordinate but within the allowable non-adjacent area, respectively. The projected/clipped new position 2802 may be located on the boundary of the allowable non-adjacent area which is closest to the original position. In case the allowed non-adjacent area beyond the current CTU is set to be has a value w and h set to the minimum value (e.g., 4) of a non-affine CU, the projected/clipped new position 2802 may be interchangeably set to be on the one boundary or the other boundary because the buffer is so small that the motion information from only one CU may be stored and, in this case, there is no difference to be clipped to one boundary side or the other side.

[00442] In some other examples, the allowed non-adjacent spatial area may include three regions. As shown in FIG. 29, the three regions are the spatial areas located on the topleft, left, and above area outside of the current CTU and adjacent to the current CTU. The Height of the top-left region and above region in the allowed non-adjacent spatial area is defined as h, while the width may depend on the picture width. The width of the left allowed non-adjacent spatial area is defined as w, while the height is equal to the height of the current CTU. As shown in FIG. 29, if the scanned non-adjacent position 2906-1/2906-2/2906-3 is beyond the allowed spatial area, the projected/clipped new position 2907-1/2907-2/2907-3 may be defined in different ways. For the top-left non-adjacent area, the projected/clipped new position is always the pixel position adjacent to the top-left position of the current CTU. For example, if the top-left position of the current CTU is (ctu_x, ctu_y), the projected/clipped new position is (ctu_x - 1, ctu_y - 1). For the above non-adjacent area, the projected/clipped new position has the same horizontal coordinate, but the vertical coordinate becomes (ctu_y - 1). For the left non-adjacent area, the projected/clipped new position has the same vertical coordinate, but the horizontal coordinate becomes (ctu_x - 1).

[00443] When motion information of an affine-coded block is saved in memory, the motion information, including CPMVs, reference index, block size and positions may be saved at the granularity of minimum affine block size (e.g., an 8x8 block). In case the current affine- coded block is a coding unit with larger size than the minimum affine block, the motion information may be saved in different methods.

[00444] In one or more embodiments, the motion information saved at each minimum affine block (e.g., 8x8 block) within the current block is just a repeated copy of the motion information of the current block. In this case, the position and size of the current block (termed as parent block in the FIGS. 22A-22B) may need to be repeatedly saved at each minimum affine block (termed as sub-block in the FIGS. 22A-22B) as well. An example for this case is shown in the FIG. 22A, where the current block (termed as the parent block) is at the size of 24x16, and the minimum affine block (termed as sub-block) is at a fixed size of 8x8.

[00445] In another or more embodiments, the motion information saved at each minimum affine block (termed as sub-block in FIGS. 22A-22B) is the motion information already projected to this minimum affine block. Since the position of each minimum affine block is known (the top-left corner of each minimum affine block), and the size of each minimum affine block is also known (the minimum size, 8x8), the position and size information of the current block (termed as the parent block in FIGS. 22A-22B) do not need to be saved. An example for this case is shown in the FIG. 22B.

[00446] When the storage method of FIG. 22B is used, the conversion from affine motion to regular/translational motion may be also simplified as follows.

[00447] In some examples, assume the minimum non-affine block is at the size of 4x4, for each 8x8 affine block, the regular/translational motion at each inside non-affine block may be computed as follows.

[00448] Taking FIG. 23 as an illustrative example. For this minimum affine block, it has three already projected CPMVs following the method shown in FIG. 22B. Based on the affine model shown in the equation (2), the regular/translational motion may be derived as below (e.g., ignoring some precision related shifts). The examples provided below are based on the center point of each 4x4 block as shown in FIG. 23, but the present disclosure is not limited to using center point to derive translation MV for each block.

[00449] In some examples, for the top-left sub-block Bl, MVl_x = e + (a >> 2) + (c >> 2), and MVl_y = f + (b » 2) + (d » 2), where a = CPMV2_x - CPMVl_x, b = CPMV2_y - CPMVl_y, c = CPMV3_x - CPMVl.x, d = CPMV3_y - CPMVl.y, e = CPMVl.x, and f = CPMVl_y. [00450] In some examples, for the top-right sub-block B2, MV2_x = MVl_x + (a »1), and MV2_y = MVl_y + (b>>l).

[00451] In some examples, for the bottom-left sub-block B3, MV3_x = MVl_x + (c »1), and MV3_y = MVl_y + (d»l).

[00452] In some examples, for the bottom-right sub-block B4, MV4_x = MVl_x + ((a+c) >>1), and MV4_y = MVl_y + ((b+d)>>l).

[00453] Alternatively, or additionally, the motion information of an affine-coded block may be saved at a different granularity a x b (e.g., 16x16 or 16x32 or 32x16 or 32x32 granularity etc.) instead of the minimum affine block size (e.g., 8x8 granularity), where the granularity values of the a and b may be configurable or decided at the encoder then signaled to the decoder. Without loss of generality, it takes a granularity of 16x16 (e.g., a = b = 16) as an illustrative example. If further assumes that the minimum affine block size is 8x8, it indicates that each 16x16 block may only save one set of affine motion information, which includes two or three CPMVs and represents one single affine model, even though the four 8x8 sub-blocks within this 16x16 block may be from more than one affine blocks, which is shown in the FIG. 25. In FIG. 25, the four 8x8 sub-blocks A, B, C and D form a 16x16 block/area, and only one affine model information is saved. However, the four 8x8 sub-blocks are from four different affine blocks, which represent four affine models and include four sets of affine motion information. In this case, there may be different ways to derive and save one single set of affine information.

[00454] In one or more examples, one of the multiple sets of available affine motion information may be selected and saved. In one example, the affine motion information at one fixed or configurable position (e.g., the top-left minimum affine block) is selected for motion storage. In another example, an averaged affine motion information of multiple models may be calculated for motion storage.

[00455] In some examples, the affine motion information at a selected neighboring affine block may be simplified/compressed before storage. [00456] In one example, it is proposed that the selected neighboring affine block is always 4-parameter model and only two CPMVs are saved.

[00457] In another example, it is proposed that the selected neighboring affine block is always uni-predicted, and only one direction of affine motion is saved.

[00458] In another example, instead of directly saving the CPMVs, it is proposed to save the affine model parameters converted from the corresponding CPMVs such that the size information (e.g., width and height) of the neighboring block does not need to be saved. In this case, the top-left CPMV may still need to be saved to provide translational motion.

[00459] In another example, each saved CPMV may be compressed before storage to further reduce the memory size. One example is to use general techniques for data compression. For example, it is provided to save a compounded value from one exponent and mantissa to approximately represent each saved CPMV.

[00460] In another method, the affine motion information of non-adjacent affine coded blocks may be accessible by derivation from existing storage of translational motion information. An example of this method is shown in FIGS. 30A-30C.

[00461] In the first step as illustrated in FIG. 30A, each scanned non-adjacent position is normalized/projected/clipped to the top-left position of the predefined/configured sub-block where the scanned non-adjacent position is located. The size of sub-block depends on the storage granularity, as shown in FIG. 30A. The locations of the spot 3001 and the spot 3002 are evenly distributed within a CTU, where the spot 3001 indicates that there is no valid affine model information at this location, while the spot 3002 indicates that there is valid affine model information at this location.

[00462] In the second step as illustrated in FIG. 30B, the information at each normalized/projected/clipped position is accessed to decide whether valid affine model information is available. As shown in FIG. 30B, the three pieces of information includes the position, the width, and the height of a non-adjacent affine CU. If the width and the height are both positive values, it indicates that valid affine information is available, otherwise not available.

[00463] The position information including horizontal coordinate and vertical coordinate may be stored in different ways. In one example, the absolute position values of each non-adjacent affine CU may be saved. In some examples, the position of each affine CU may be at least multiples of the minimum CUs, such that the position values may be right shift 2 bits (e.g., if minimum CU size is 4x4) or 3 bits (e.g., if minimum CU size is 8x8) before storage for reduced memory cost.

[00464] In another example, the relative position values of each non-adjacent affine CU may be saved. In some examples, the CTU position information where each non-adjacent affine CU is located is known. In this case, for each non-adjacent affine CU, its position relative to the top-left pixel position of its associated CTU may be saved. For example, if the maximum CTU size is 256x256 and the minimum CU size is 4x4, the dynamic range of position information may be represented as 4 x [0, 63] , where the values in the range of [0, 63] may be represented in 6 bits.

[00465] The size information including width and height may be stored in different ways. In one method, the size may be stored at actual values and the dynamic range is (0, maximum CTU size] . In another example, the size information may be stored at a look up table, where the table only includes a set of all possible size values, including 8, 16, 32, 64, 128 and up to the maximum CTU size. In this case, 3 bits are enough to represents width or height values. In another example, if the size information is always two to the power of N, then the right shift operations may be applied for size information storage (e.g., right shift 3 bits before storing size information).

[00466] In the third step as shown in FIG. 30C, if valid affine information is available, the pivot positions (e.g., the center position) of three sub-blocks (e.g., the spots 3003 in FIG. 30C) are able to be derived by using the position, the width and the height of the associated affine CU. In some examples, the three sub-blocks are the top-left, top-right and bottom-left internal sub-blocks of the associated affine CU.

[00467] In one example, if the position 3004 of the represented non-adjacent affine CU is (x, y), the size of this affine CU is w and h, and the size of each internal sub-block is 4x4, the center position of the top-left sub-block may be derived as (x+2, y+2), the center position of the top-right sub-block may be derived as (x+w-2, y+2), and the center position of the bottomleft sub-block may be derived as (x+2, y+h-2).

[00468] As shown in FIG. 30C, since the pivot location used for deriving the translational/regular motion information of each sub-block is different from the pixel location where the original CPMV is derived, there may be different methods for deriving the affine model information of the non-adjacent affine CU. In one example, the translational/regular motion information of the three sub-blocks, which are already available in the existing motion storage, are directly used as the CPMVs for the selected non-adjacent affine CU to derive affine model parameters. In another example, the translational/regular motion information of the three sub-blocks are directly used as the CPMVs, but the size and position of the non-adjacent affine CU may be modified to match the position difference between the sub-block MV and original CPMV. In one or more examples, the size is reduced from original width x height to (width - sub-block width) x (height- sub-block height), and the position is changed from original (x, y) to (x + half of sub-block width, y + half of sub-block height).

[00469] In another method, the affine motion information of non-adjacent affine coded blocks may be still accessible by derivation from existing storage of translational motion information, but the required information and memory size may be further reduced. One example in this case is shown in FIGS. 31A-31C.

[00470] The example in FIG. 31 A is exactly the same as FIG. 30A, which indicates that the storage granularity in this method may be similarly defined as the method in FIGS. 3OA-3OC, [00471] In FIG. 3 IB, only one piece of information, the affine CU position, is stored. This indicates that lower memory storage is required for this method. The required position information may be similarly stored as FIG. 30B, where the absolute position values or relative position values (e g., to the current CTU top-left comer) may be stored. To indicate whether the stored position information is valid, multiple methods may be proposed. In one method, a specially selected position may be used to represent invalid position information. For example, the bottom-right corner position of the current CTU (if relative position values are stored) or the current picture/ slice (if absolute position values are stored) may be used to differentiate valid or invalid position information.

[00472] In FIG. 31C, the internal sub-blocks are differently selected for affine model derivation. Note that since width and height information are not stored in FIG. 3 IB, it is unable to determine the top-right and bottom-left internal sub-blocks within the current affine CU. Depending on the actual definition of minimum width and height, the selection of internal subblocks may be adaptively decided. In one or more embodiments, in some traditional video codecs, such as VVC/AVS/ECM, the minimum width and height of an affine CU may be constantly defined as value of 8, and the width and height of any internal sub-blocks within an affine CU may be always defined as value of 4. In this case, there are at least four sub-blocks within each affine CU, and the three sub-blocks 3103 may be selected as: the top-left sub-block (accessible due to the affine CU position information), the immediate right sub-block of the top-left sub-block, and the immediate bottom sub-block of the top-left sub-block. The position 3104 of the represented non-adjacent affine CU is (x, y). Once the three sub-blocks are selected, the remaining procedure details for affine model derivation may be similarly defined as the method shown in FIG. 30C. Note that, if the three sub-blocks are selected as shown in FIG. 31C, the width and height of the associated affine CU may be set to the minimum width and height value because the actual width and height values are not stored and inaccessible.

[00473] In the examples above, methods may be applied for motion information storage in any combinations. For example, the defined restricted area for non-adjacent neighboring blocks may be combined with the usage of compressed affine motion information.

[00474] MMVD based candidate derivation method for affine AMVP and merge mode [00475] In MMVD mode, the best MVD information is selected at the encoder side based on rate-distortion optimization (RDO) method, and then signaled to the decoder side. The tradeoff between the signaling cost and the potential coding bits saving decides whether the MMVD mode is applied to a coding block. [00476] In some examples, it is provided to use MVD information to refine the existing candidates in the affine AMVP or/and affine merge candidate list. The new candidates after refinements are then inserted into the existing affine AMVP or/and affine merge candidate list. [00477] In some examples, the available number of combinations for MVD information, such as the motion magnitude (e.g., offset value) and motion direction (e.g., sign value), may be the same or different as the existing MMVD mode in the WC. In one example, a smaller number of offset values, such as {1, 2, 4, 8, 16}, may be used. In another example, a different or the same set of direction values as the existing MMVD mode may be used.

[00478] When selecting one or more base MVs, in some examples, the base MV is any one of the candidates from the existing affine AMVP and/or merge candidate list, and there may be multiple ways to determine the selection of a potential base MV. In one or more examples, the base MV may be selected from a candidate list before or after an adaptive reordering method such as ARMC is applied to this candidate list.

[00479] In one or more examples, a single base MV or multiple base MVs may be selected from a candidate list. For example, when a single base MV is selected, one or multiple combinations (e.g., Y combinations) of MVD information may be selected to refine this single base MV, which indicates that Y new candidates (e.g., each combination of MVD information is applied to the base MV and generate one new candidate) may be generated and inserted into the candidate list. For another example, when multiple base MVs are selected (e.g., X base MVs), one or multiple combinations (e.g., Y combinations) of MVD information may be selected to refine each selected base MV, which indicates X multiplied by Y new candidates may be generated and inserted into the candidates list.

[00480] When one or multiple base MVs are selected, the index of each selected base MV may be determined in different ways. In one or more examples, the index of the selected base MV may be determined by avoiding the base MVs which are already selected in MMVD mode, if affine MMVD mode is enabled for the current coding process. [00481] In one or more examples, the index of the selected base MV may be determined by following a predefined order. For example, N candidates from the beginning of the list are sequentially selected as the base MVs. If any base MV is already selected by the current affine MMVD mode, this base MV may be skipped.

[00482] The newly generated candidates after different combinations of MVD refinements may be directly inserted into the existing candidate list. Alternatively, another round of reordering process may be applied to all the new candidates and the top Z candidates with smaller matching cost (e.g., template matching cost or bilateral matching cost) may be selected to be inserted into the candidate list.

[00483] Improved regression based affine candidate derivation

[00484] In the regression based affine candidate derivation method, there are several steps performed by the encoder/decoder as follows:

• Step 1 : Scan and determine a selected neighboring affine CU.

• Step 2: Use only the motion information of the internal sub-blocks from the selected neighboring affine CU as input and solve a linear regression model to derivate one affine model of parameters

• Step 3 : Use both of the neighboring (adjacent or non-adjacent) affine CU internal motion information and the adjacent 4x4 sub-block motion information (the above and left gray sub-blocks shown in figure 32) of the current CU as the input and solve a linear regression model to derivate another affine model of parameters

• Step 4: Convert the derived affine model parameters into a set of affine CPMVs and insert the CPMVs into the existing affine merge and AMVP candidate lists.

[00485] Note that in the above step 2 and step 3, the difference is that whether subblock motion information from affine CU only or together with adjacent 4x4 sub-block motion information is used as the input for linear regression calculation. In addition, for each selected neighboring affine CU, two affine models may be derived: one is derived from only neighboring affine sub-blocks, while the other one is derived from both neighboring affine sub-blocks and adjacent 4x4 sub-blocks. [00486] In some examples, several methods are proposed to improve the regression method.

[00487] For the scanning and determination of neighboring affine CU, the above proposed methods may be used in any combinations. In one embodiment, the scanning and determination method proposed in FIG. 17A may be used. In order to constrain the memory usage, the methods proposed in FIG. 28 or 29 may be used to differentiate the access of non-adjacent neighboring affine CU motion information when the selected non-adjacent affine CU is located within the current CTU or beyond the current CTU. When the selected non-adjacent affine CU is located within the current CTU, the associated motion information may be accessed at different granularity and availability, and the method proposed in FIG. 30 or 31 may be applied as an alternative solution.

[00488] As explained above, for each selected neighboring affine CU, two affine models may be derived. For the first type of affine models, only the motion information of the internal sub-blocks of the selected neighboring affine CU are used as the input for the linear regression calculation. As the affine model currently used in the major traditional video codecs such as VVC/AVS is actually a linear model, all the internal sub-blocks essentially follow a same linear model with only rounding variations. In this case, it is proposed to replace the linear regression calculation for the first type of affine model derivation with a simple affine model inheritance, while the inheritance procedure is the same as the projection method used in the current major traditional video codecs or the method proposed above in the section of non-adjacent neighbor based derivation process for affine inherited merge candidates. For the memory constrained case, the affine motion before the inheritance projection may be obtained by following the method proposed in FIG. 30C or 31C.

[00489] For the second type of affine models, both the motion information of the internal sub-blocks of the selected neighboring affine CU and motion information of adjacent 4x4 subblocks are used as the input for the linear regression calculation. It is proposed to control the motion information ratio from the affine internal sub-blocks and from the adjacent 4x4 sub-blocks. Especially when the memory constrained case such as FIG. 31C is used, the internal affine subblocks may be available at a reduced number (e g., only four internal sub-blocks are available for the affine CU). In one or more embodiments, when the number of available affine sub-blocks is much smaller (e.g., a threshold ratio value may be defined as thl) than the number of available adjacent 4x4 sub-blocks, the derivation process may be skipped or early terminated. Tn another one or more embodiments, when the number of available affine sub-blocks is much bigger (e.g., a threshold ratio value may be defined as th2) than the number of available adjacent 4x4 sub-blocks, the derivation process may be skipped or early terminated. Note that the threshold ratio values may be predefined as constant values, or directly signaled to the decoder or a look up table index signaled to the decoder.

[00490] FIG. 24 shows a computing environment (or a computing device) 2410 coupled with a user interface 2460. The computing environment 2410 can be part of a data processing server. In some embodiments, the computing device 2410 can perform any of various methods or processes (such as encoding/decoding methods or processes) as described hereinbefore in accordance with various examples of the present disclosure. The computing environment 2410 may include a processor 2420, a memory 2440, and an I/O interface 2450.

[00491] The processor 2420 typically controls overall operations of the computing environment 2410, such as the operations associated with the display, data acquisition, data communications, and image processing. The processor 2420 may include one or more processors to execute instructions to perform all or some of the steps in the above-described methods. Moreover, the processor 2420 may include one or more modules that facilitate the interaction between the processor 2420 and other components. The processor may be a Central Processing Unit (CPU), a microprocessor, a single chip machine, a GPU, or the like.

[00492] The memory 2440 is configured to store various types of data to support the operation of the computing environment 2410. Memory 2440 may include predetermine software 2442. Examples of such data include instructions for any applications or methods operated on the computing environment 2410, video datasets, image data, etc. The memory 2440 may be implemented by using any type of volatile or non-volatile memory devices, or a combination thereof, such as a static random access memory (SRAM), an electrically erasable programmable read-only memory (EEPROM), an erasable programmable read-only memory (EPROM), a programmable read-only memory (PROM), a read-only memory (ROM), a magnetic memory, a flash memory, a magnetic or optical disk. [00493] The I/O interface 2450 provides an interface between the processor 2420 and peripheral interface modules, such as a keyboard, a click wheel, buttons, and the like. The buttons may include but are not limited to, a home button, a start scan button, and a stop scan button. The I/O interface 2450 can be coupled with an encoder and decoder.

[00494] In some embodiments, there is also provided a non-transitory computer- readable storage medium including a plurality of programs, such as included in the memory 2440, executable by the processor 2420 in the computing environment 2410, for performing the above-described methods. For example, the non-transitory computer-readable storage medium may be a ROM, a RAM, a CD-ROM, a magnetic tape, a floppy disc, an optical data storage device or the like.

[00495] The non-transitory computer-readable storage medium has stored therein a plurality of programs for execution by a computing device having one or more processors, where the plurality of programs when executed by the one or more processors, cause the computing device to perform the above-described method for motion prediction.

[00496] In some embodiments, the computing environment 2410 may be implemented with one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field- programmable gate arrays (FPGAs), graphical processing units (GPUs), controllers, microcontrollers, microprocessors, or other electronic components, for performing the above methods.

[00497] FIG. 33 is a flowchart illustrating a method for video decoding according to an example of the present disclosure.

[00498] In step 3301, the processor 2420, at the decoder side, may obtain, based on a storage granularity, a normalized position for an affine CU at a non-adjacent position obtained by scanning area that is not adjacent to a current CU.

[00499] As shown in FIG. 31 A, each scanned non-adjacent position may be normalized/projected/clipped to the top-left position of the predefined/configured sub-block where the scanned non-adjacent position is located, e.g., to proximate to one spot at 3101 or 3102.

[00500] In step 3302, the processor 2420, at the decoder side, may determine whether valid affine model information is stored for the normalized position, where the valid affine model information includes a position of the affine CU. As shown in FIG. 3 IB, only one piece of information, the affine CU position, is stored, thus lower memory storage is needed.

[00501] In step 3303, the processor 2420, at the decoder side, may derive affine model information for the affine CU based on the valid affine model information that is stored for the normalized position in response to determining that the valid affine model information is stored for the normalized position.

[00502] In some examples, the processor 2420, at the decoder side, may determine that the valid affine model information is stored for the normalized position, where the specific position may include a corner position that is located at right bottom of a current CTU in response to determining that the position of the affine CU is not located at a specific position. For example, the bottom-right corner position of the current CTU (if relative position values are stored) or the current picture/slice (if absolute position values are stored) may be used to differentiate valid or invalid position information.

[00503] In some examples, the position of the affine CU may be indicated using absolute position values or relative position values. Furthermore, the position may be stored in one of following manners: a horizontal coordinate and a vertical coordinate with the absolute position values; or a horizontal coordinate and a vertical coordinate with the relative position values.

[00504] In some examples, the processor 2420, at the decoder side, may obtain three pivot positions based on the position of the affine CU and a minimum size of the affine CU, obtain regular motion information of three sub-blocks at the three pivot positions, and derive the affine model information for the affine CU based on the regular motion information of the three sub-blocks. In some examples, the three sub-blocks may include a first sub-block at the top-left of the affine CU with the valid affine model information, a second sub-block that is an immediate right sub-block of the first subblock and a third sub-block that is an immediate bottom sub-block of the first subblock. As shown in FIG. 31C, the three sub-blocks are respectively indicated by 3103.

[00505] In some examples, the processor 2420, at the decoder side, may further obtain CPMVs for the affine CU by directly using the regular motion information stored for the three sub-blocks.

[00506] In some other examples, the processor 2420, at the decoder side, may further derive a new affine model by using the regular motion information and obtain CPMVs for the affine CU by using the new affine model.

[00507] FIG. 34 is a flowchart illustrating a method for video encoding corresponding the method for video decoding as shown in FIG. 33.

[00508] In step 3401, the processor 2420, at the encoder side, may obtain, based on a storage granularity, a normalized position for an affine CU at a non-adjacent position obtained by scanning area that is not adjacent to a current CU.

[00509] As shown in FIG. 31 A, each scanned non-adjacent position may be normalized/projected/clipped to the top-left position of the predefined/configured sub-block where the scanned non-adjacent position is located, e.g., to proximate to one spot at 3101 or 3102.

[00510] In step 3402, the processor 2420, at the encoder side, may determine whether valid affine model information is stored for the normalized position, where the valid affine model information includes a position of the affine CU. As shown in FIG. 3 IB, only one piece of information, the affine CU position, is stored, thus lower memory storage is needed.

[00511] In step 3403, the processor 2420, at the encoder side, may derive affine model information for the affine CU based on the valid affine model information that is stored for the normalized position in response to determining that the valid affine model information is stored for the normalized position. [00512] In some examples, the processor 2420, at the encoder side, may determine that the valid affine model information is stored for the normalized position, where the specific position may include a corner position that is located at right bottom of a current CTU in response to determining that the position of the affine CU is not located at a specific position. For example, the bottom-right corner position of the current CTU (if relative position values are stored) or the current picture/slice (if absolute position values are stored) may be used to differentiate valid or invalid position information.

[00513] In some examples, the position of the affine CU may be indicated using absolute position values or relative position values. Furthermore, the position may be stored in one of following manners: a horizontal coordinate and a vertical coordinate with the absolute position values; or a horizontal coordinate and a vertical coordinate with the relative position values.

[00514] In some examples, the processor 2420, at the encoder side, may obtain three pivot positions based on the position of the affine CU and a minimum size of the affine CU, obtain regular motion information of three sub-blocks at the three pivot positions, and derive the affine model information for the affine CU based on the regular motion information of the three sub-blocks. In some examples, the three sub-blocks may include a first sub-block at the top-left of the affine CU with the valid affine model information, a second sub-block that is an immediate right sub-block of the first subblock and a third sub-block that is an immediate bottom sub-block of the first subblock. As shown in FIG. 31C, the three sub-blocks are respectively indicated by 3103.

[00515] In some examples, the processor 2420, at the encoder side, may further obtain CPMVs for the affine CU by directly using the regular motion information stored for the three sub-blocks.

[00516] In some other examples, the processor 2420, at the encoder side, may further derive a new affine model by using the regular motion information and obtain CPMVs for the affine CU by using the new affine model. [00517] FIG. 35 is a flowchart illustrating a method for video decoding according to an example of the present disclosure.

[00518] In step 3501, the processor 2420, at the decoder side, may scan a neighboring area of a current CU to obtain a neighboring affine CU that is previously coded, where the neighboring affine CU may include a plurality of internal sub-blocks.

[00519] In some examples, the neighboring affine CU may be located in an area that is adjacent or non-adjacent to the current CU.

[00520] In some examples, the processor 2420 may scan in a manner as provided in FIG. 17A as discussed above. For example, the processor 2420 may scan the neighboring area of the current CU based on a scanning rule including at least one scanning area, the at least one scanning distance, and at least one scanning direction. The at least one scanning area may include a first scanning area and a second scanning area, the first scanning area may be determined according to a first maximum scanning distance indicating a maximum number of blocks away from a first side of the current CU, the second scanning area may be determined according to a second maximum scanning distance indicating a maximum number of blocks away from a second side of the current CU, and the first maximum scanning distance may be the same as or different than the second maximum scanning distance.

[00521] Furthermore, the at least one scanning direction may include a first scanning direction and a second scanning direction, and the processor 2420 may predetermine the first maximum scanning distance or the second maximum scanning distance as a fixed value, and scan the first scanning area in the first scanning direction and the second scanning area in the second scanning direction.

[00522] In some examples, in order to constrain the memory usage, the methods proposed in FIG. 28 or 29 may be used to differentiate the access of non-adjacent neighboring affine CU motion information when the selected non-adjacent affine CU is located within the current CTU or beyond the current CTU. For example, the processor 2420, at the decoder side, may obtain a restricted area that is not adjacent to the current CU according to a value associated with the restricted area, obtain affine model information of the neighboring affine CU based on the restricted area, and obtain the motion information of the plurality of internal subblocks in the neighboring affine CU based on the affine model information of the neighboring affine CU. In some examples, the restricted area may be a predefined area associated with the current CU. Such association may be a spatial relationship between the restricted area and the CU, or a mapping relationship predefined between the restricted area and the CU.

[00523] In some examples, when the selected non-adjacent affine CU is located within the current CTU, the associated motion information may be accessed at different granularity and availability, and the method provided in FIGS. 30A-30C and 31A-31C may be applied as an alternative solution. For example, the processor 2420, at the decoder side, may obtain, based on a storage granularity, a normalized position for the neighboring affine CU, determine whether valid affine model information is stored for the normalized position, and derive affine model information for the neighboring affine CU based on the valid affine model information that is stored for the normalized position in response to determining that the valid affine model information is stored for the normalized position.

[00524] In step 3502, the processor 2420, at the decoder side, may derive a plurality of first affine model parameters based on the neighboring affine CU.

[00525] In some examples, the processor 2420, at the decoder side, may derive the plurality of first affine model parameters based on the neighboring affine CU and a first linear regression model, where an input of the first linear regression model may include motion information of a first set of internal sub-blocks selected from the plurality of internal subblocks, a number of the first set of internal sub-block may be smaller than or equal to the number of all the plurality of internal sub-block included in the neighboring affine CU.

[00526] In some examples, the processor 2420, at the decoder side, may obtain three pivot positions based on at least one of following information: a position of the neighboring affine CU, a width of the neighboring affine CU, or a height of the neighboring affine CU. Furthermore, the processor 2420, at the decoder side, may obtain motion information of three internal sub-blocks at the three pivot positions. Moreover, the processor 2420, at the decoder side, may derive the plurality of first affine model parameters based on the motion information of three internal sub-blocks.

[00527] In step 3503, the processor 2420, at the decoder side, may derive a plurality of second affine model parameters based on a second regression model, where an input of the second linear regression model may include motion information of a second set of internal sub-blocks and motion information of a plurality of adjacent sub-blocks of the current CU, the second set of internal sub-blocks may be selected from the plurality of internal sub-blocks, and a number of the second set of internal sub-blocks may be smaller than or equal to a number of all the plurality of internal sub-block included in the neighboring affine CU, and the plurality of adjacent sub-blocks may be located in an adjacent area that is adjacent to the current CU.

[00528] In some examples, the adjacent area may include a top area and a left area, the top area may be above the current CU, and the left area may be on the left of the current CU and below the top area, a width of the top area may equal to double widths of the current CU, and a height of the left area may equal to one and half of a height of the current CU, as shown in FIG. 32.

[00529] In step 3504, the processor 2420, at the decoder side, may obtain a plurality of CPMVs based on the plurality of first affine model parameters and the plurality of second affine model parameters.

[00530] In some examples, the processor 2420, at the decoder side, may obtain a ratio of a number of the second set of internal sub-blocks and a number of the plurality of adjacent sub-blocks, and skip or terminate derivation of the plurality of second affine model parameters in response to determining that the ratio is greater than a first threshold ratio or smaller than a second threshold ratio.

[00531] FIG. 36 is a flowchart illustrating a method for video encoding corresponding the method for video decoding as shown in FIG. 35. [00532] In step 3601, the processor 2420, at an encoder side, may scan a neighboring area of a current CU to obtain a neighboring affine CU that is previously coded, where the neighboring affine CU may include a plurality of internal sub-blocks.

[00533] In some examples, the neighboring affine CU may be located in an area that is adjacent or non-adjacent to the current CU.

[00534] In some examples, the processor 2420 may scan in a manner as provided in FIG. 17A as discussed above. For example, the processor 2420 may scan the neighboring area of the current CU based on a scanning rule including at least one scanning area, the at least one scanning distance, and at least one scanning direction. The at least one scanning area may include a first scanning area and a second scanning area, the first scanning area may be determined according to a first maximum scanning distance indicating a maximum number of blocks away from a first side of the current CU, the second scanning area may be determined according to a second maximum scanning distance indicating a maximum number of blocks away from a second side of the current CU, and the first maximum scanning distance may be the same as or different than the second maximum scanning distance.

[00535] Furthermore, the at least one scanning direction may include a first scanning direction and a second scanning direction, and the processor 2420 may predetermine the first maximum scanning distance or the second maximum scanning distance as a fixed value, and scan the first scanning area in the first scanning direction and the second scanning area in the second scanning direction.

[00536] In some examples, in order to constrain the memory usage, the methods proposed in FIG. 28 or 29 may be used to differentiate the access of non-adjacent neighboring affine CU motion information when the selected non-adjacent affine CU is located within the current CTU or beyond the current CTU. For example, the processor 2420, at the encoder side, may obtain a restricted area that is not adjacent to the current CU according to a value associated with the restricted area, obtain affine model information of the neighboring affine CU based on the restricted area, and obtain the motion information of the plurality of internal sub- blocks in the neighboring affine CU based on the affine model information of the neighboring affine CU. In some examples, the restricted area may be a predefined area associated with the current CU. Such association may be a spatial relationship between the restricted area and the CU, or a mapping relationship predefined between the restricted area and the CU.

[00537] In some examples, when the selected non-adjacent affine CU is located within the current CTU, the associated motion information may be accessed at different granularity and availability, and the method provided in FIGS. 30A-30C and 31A-31C may be applied as an alternative solution. For example, the processor 2420, at the encoder side, may obtain, based on a storage granularity, a normalized position for the neighboring affine CU, determine whether valid affine model information is stored for the normalized position, and derive affine model information for the neighboring affine CU based on the valid affine model information that is stored for the normalized position in response to determining that the valid affine model information is stored for the normalized position.

[00538] In step 3602, the processor 2420, at the encoder side, may derive a plurality of first affine model parameters based on the neighboring affine CU.

[00539] In some examples, the processor 2420, at the encoder side, may derive the plurality of first affine model parameters based on the neighboring affine CU and a first linear regression model, where an input of the first linear regression model may include motion information of a first set of internal sub-blocks selected from the plurality of internal subblocks, a number of the first set of internal sub-block may be smaller than or equal to the number of all the plurality of internal sub-block included in the neighboring affine CU.

[00540] In some examples, the processor 2420, at the encoder side, may obtain three pivot positions based on at least one of following information: a position of the neighboring affine CU, a width of the neighboring affine CU, or a height of the neighboring affine CU. Furthermore, the processor 2420, at the encoder side, may obtain motion information of three internal sub-blocks at the three pivot positions. Moreover, the processor 2420, at the encoder side, may derive the plurality of first affine model parameters based on the motion information of three internal sub-blocks. [00541] In step 3603, the processor 2420, at the encoder side, may derive a plurality of second affine model parameters based on a second regression model, where an input of the second linear regression model may include motion information of a second set of internal sub-blocks and motion information of a plurality of adjacent sub-blocks of the current CU, the second set of internal sub-blocks may be selected from the plurality of internal sub-blocks, and a number of the second set of internal sub-blocks may be smaller than or equal to a number of all the plurality of internal sub-block included in the neighboring affine CU, and the plurality of adjacent sub-blocks may be located in an adjacent area that is adjacent to the current CU.

[00542] In some examples, the adjacent area may include a top area and a left area, the top area may be above the current CU, and the left area may be on the left of the current CU and below the top area, a width of the top area may equal to double widths of the current CU, and a height of the left area may equal to one and half of a height of the current CU, as shown in FIG. 32.

[00543] In step 3604, the processor 2420, at the encoder side, may obtain a plurality of CPMVs based on the plurality of first affine model parameters and the plurality of second affine model parameters.

[00544] In some examples, the processor 2420, at the encoder side, may obtain a ratio of a number of the second set of internal sub-blocks and a number of the plurality of adjacent sub-blocks, and skip or terminate derivation of the plurality of second affine model parameters in response to determining that the ratio is greater than a first threshold ratio or smaller than a second threshold ratio.

[00545] In some examples, there is provided an apparatus for video decoding. The apparatus includes a processor 2420 and a memory 2440 configured to store instructions executable by the processor; where the processor, upon execution of the instructions, is configured to perform any method as illustrated in FIGS. 33 and 35.

[00546] In some examples, there is provided an apparatus for video encoding. The apparatus includes a processor 2420 and a memory 2440 configured to store instructions executable by the processor; where the processor, upon execution of the instructions, is configured to perform any method as illustrated in FIGS. 34 and 36.

[00547] In some other examples, there is provided a non-transitory computer readable storage medium, having instructions stored therein. When the instructions are executed by a processor 2420, the instructions cause the processor to perform any method as illustrated in FIGS. 33-36. In one example, the plurality of programs may be executed by the processor 2420 in the computing environment 2410 to receive (for example, from the video encoder 20 in FIG. 1G) a bitstream or data stream including encoded video information (for example, video blocks representing encoded video frames, and/or associated one or more syntax elements, etc.), and may also be executed by the processor 2420 in the computing environment 2410 to perform the decoding method described above according to the received bitstream or data stream. In another example, the plurality of programs may be executed by the processor 2420 in the computing environment 2410 to perform the encoding method described above to encode video information (for example, video blocks representing video frames, and/or associated one or more syntax elements, etc.) into a bitstream or data stream, and may also be executed by the processor 2420 in the computing environment 2410 to transmit the bitstream or data stream (for example, to the video decoder 30 in FIG. 2B). Alternatively, the non-transitory computer-readable storage medium may have stored therein a bitstream or a data stream including encoded video information (for example, video blocks representing encoded video frames, and/or associated one or more syntax elements etc.) generated by an encoder (for example, the video encoder 20 in FIG. 1G) using, for example, the encoding method described above for use by a decoder (for example, the video decoder 30 in FIG. 2B) in decoding video data. The non-transitory computer-readable storage medium may be, for example, a ROM, a Random Access Memory (RAM), a CD-ROM, a magnetic tape, a floppy disc, an optical data storage device or the like.

[00548] Other examples of the disclosure will be apparent to those skilled in the art from consideration of the specification and practice of the disclosure disclosed here. This application

Il l is intended to cover any variations, uses, or adaptations of the disclosure following the general principles thereof and including such departures from the present disclosure as come within known or customary practice in the art. It is intended that the specification and examples be considered as exemplary only.

[00549] It will be appreciated that the present disclosure is not limited to the exact examples described above and illustrated in the accompanying drawings, and that various modifications and changes can be made without departing from the scope thereof.