Abstract: Systems and methods are described for refining motion compensated predictions in block-based video coding. In an example embodiment, motion-compensated prediction is used to generate predicted sample values in a current block of samples. A precision difference value and a motion vector refinement for the current block are signaled in the bitstream. For each sample in the current block, a spatial gradient is calculated at the sample, and a scalar product is calculated between the spatial gradient and the motion vector refinement. The scalar product is scaled (e.g. bit-shifted) by an amount indicated by the precision difference value to generate a sample difference value, and the sample difference value is added to the predicted sample value to generate a refined sample value.

Abstract: Cross-plane filtering may be used to restore blurred edges and/or textures in one or both chroma planes using information from a corresponding luma plane. Adaptive cross-plane filters may be implemented. Cross-plane filter coefficients may be quantized and/or signaled such that overhead in a bitstream minimizes performance degradation. Cross-plane filtering may be applied to select regions of a video image (e.g., to edge areas). Cross-plane filters may be implemented in single-layer video coding systems and/or multi-layer video coding systems.

Abstract: Cross plane filtering may be used for enhanced chroma coding. An indication of a cross-plane filter associated with a current picture may be received. The current picture may include an intra-coded video block and a plurality of reference samples. The plurality of reference samples may be used to predict the intra-coded video block. A luma sample region may be determined in the current picture. The luma sample region may determined to enhance a corresponding chroma sample in the current picture. The cross-plane filter may be applied to a plurality of luma samples in the luma sample region to determine an offset. The cross-plane filter may be a high pass filter. The offset may be applied to the corresponding chroma sample to determine an enhanced chroma sample.

Abstract: Systems, methods, and instrumentalities are disclosed for discontinuous face boundary filtering for 360-degree video coding. A face discontinuity may be filtered (e.g., to reduce seam artifacts) in whole or in part, for example, using coded samples or padded samples on either side of the face discontinuity. Filtering may be applied, for example, as an in-loop filter or a post-processing step. 2D positional information related to two sides of the face discontinuity may be signaled. In a video bitstream so that filtering may be applied independent of projection formats and/or frame packing techniques.

Abstract: Systems, methods, and instrumentalities are described herein for calculating local illumination compensation (LIC) parameters for bi-predicted coding unit (CU). The LIC parameters may be used to generate adjusted samples for the current CU and to address local illumination changes that may exist among temporal neighboring pictures. LIC parameters may be calculated based on bi-predicted reference template samples and template samples for a current CU. Bi-predicted reference template samples may be generated based on reference template samples neighboring temporal reference CUs. For example, the bi-predicted reference template samples may be generated based on averaging the reference template samples. The reference template samples may correspond to template samples for the current CU. A CU may be or may include a coding block and/or a sub-block that may be derived by dividing the coding block.

Abstract: External overlapped block motion compensation (OBMC) may be performed for samples of a coding unit (CU) located along an inter-CU boundary of the CU while internal OBMC may be performed separately for samples located along inter-sub-block boundaries inside the CU. External OBMC may be applied based on substantially similar motion information associated with multiple external blocks neighboring the CU. The external blocks may be treated as a group to provide OBMC for multiple boundary samples together in an external OBMC operation. Internal OBMC may be applied using the same sub-block size used for sub-block level motion derivation. Internal OBMC may be disabled for the CU, for example, if the CU is coded in a spatial-temporal motion vector prediction (STMVP) mode.

Abstract: Systems, methods, and instrumentalities are disclosed for clustering-based quantization for neural network (NN) compression. A distribution of weights in weight tensors in NN layers may be analyzed to identify cluster outliers. Cluster inliers may be coded from cluster outliers, for example, using scalar and/or vector quantization. Weight-rearrangement may rearrange weights for higher dimensional weight tensors into lower dimensional matrices. For example, weight rearrangement may flatten a convolutional kernel into a vector. Correlation between kernels may be preserved, for example, by treating a filter or kernels across a channel as a point. A tensor may be split into multiple subspaces, for example, along an input and/or an output channel. Predictive coding may be performed for a current block of weights or weight matrix based on a reshaped or previously coded block or matrix. Arrangement, inlier, outlier, and/or prediction information may be signaled to a decoder for reconstruction of a compressed NN.

Abstract: Systems, methods, and instrumentalities may be provided for discounting reconstructed samples and/or coding information from spatial neighbors across face discontinuities. Whether a current block is located at a face discontinuity may be determined. The face discontinuity may be a face boundary between two or more adjoining blocks that are not spherical neighbors. The coding availability of a neighboring block of the current block may be determined, e.g., based on whether the neighboring block is on the same side of the face discontinuity as the current block. For example, the neighboring block may be determined to be available for decoding the current block if it is on the same side of the face discontinuity as the current block, and unavailable if it Is not on the same side of the face discontinuity. The neighboring block may be a spatial neighboring block or a temporal neighboring block.

Abstract: Sampling grid information may be determined for multi-layer video coding systems. The sampling grid information may be used to align the video layers of a coding system. Sampling grid correction may be performed based on the sampling grid information. The sampling grids may also be detected. In some embodiments, a sampling grid precision may also be detected and/or signaled.

Abstract: Systems, methods, and instrumentalities are disclosed for sub-block/block refinement, including sub-block/block boundary refinement, such as block boundary prediction refinement with optical flow (BBPROF). A block comprising a current sub-block may be decoded based on a sample value for a first pixel that is obtained based on, for example, an MV for a current sub-block, an MV for a sub-block adjacent the current sub-block, and a sample value for a second pixel adjacent the first pixel. BBPROF may include determining spatial gradients at pixel(s)/sample location(s). An MV difference may be calculated between a current sub-block and one or more neighboring sub-blocks. An MV offset may be determined at pixel(s)/sample location(s) based on the MV difference. A sample value offset for the pixel in a current sub-block may be determined. The prediction for a reference picture list may be refined by adding the calculated sample value offset to the sub-block prediction.

Abstract: Implementations of the present disclosure provide a solution for encoding and decoding motion information. In this solution, during a conversion between a current video block of a video and a bitstream of the video, a group type of a subgroup of candidates of motion information for the current video block is determined, wherein the group type indicating whether the subgroup of candidates are to be reordered. Further, a list of candidates are constructed based on the group type; and the motion information for the current video block is derived from the list of candidates.

Abstract: Aspects include a method, apparatus and computer-readable medium of decoding video or blocks of an image, including receiving a bitstream of the image, deriving, for a block of the image in the bitstream, multiple intra-prediction modes (IPMs) to use in decoding the block, determining, based on the multiple IPMs, a final predictor to use in decoding the block, and decoding the block using the final predictor. Other aspects include method, apparatus and computer-readable medium for similarly encoding video or blocks of an image.

Abstract: Media content coded using scalable coding techniques may be cached among a group of cache devices. Layered segments of the media content may be pre-loaded onto the cache devices, which may be located throughout a content distribution network, including a home network. The caching location of the media content may be determined based on multiple factors including a content preference associated with the group of cache devices and device capabilities. A cache controller may manage the caching of the media content.

Abstract: Example implementations include a method and of video processing, comprising determining, during a conversion between a current video unit of a video and a bitstream of the video, to apply a prediction mode derivation tool to the current video unit, wherein in the prediction mode derivation tool, at least one intra prediction mode (IPM) is derived for the current video unit based on cost calculations between reconstructed samples and prediction samples of a template region related the current video unit. The implementations further include performing the conversion based on the determining.

Abstract: Aspects of the present disclosure provide techniques for derive one or more intra prediction modes (IPMs) from a subset of IPM candidates in order to determine a predictor to use for decoding a block of an image. In some aspects, the subset of IPM candidates may include IPMs that are less than the number of IPMs in a full set of all available IPM candidates (e.g., 67 IPMs in VVC or 35 in HEVC). In some aspects, the subset of IPM candidates may be based on a most probable mode (MPM) list that can be used to determine or signal an IPM based on IPMs previously used in decoding other blocks.

Abstract: Example implementations include a method, apparatus and computer-readable medium of video processing, including constructing, during a conversion between a current video block of a video and a bitstream of the video, at least one template set for the current video block from a plurality of sub-templates. The one or more sub-templates may be selected from a plurality of sub-templates including: a left sub-template, an above sub-template, a right-above sub-template, a left-below sub-template, and a left-above sub-template. The implementations further include deriving at least one intra-prediction mode (IPM) based on cost calculations. The implementations include determining, based on the at least one IPM, a final predictor of the current video block. The implementations include performing the conversion based on the final predictor.

Abstract: Example implementations include a method, apparatus and computer-readable medium of video processing, including constructing, during a conversion between a current video block of a video and a bitstream of the video, at least one template set for the current video block from a plurality of sub-templates. The one or more sub-templates may be selected from a plurality of sub-templates including: a left sub-template, an above sub-template, a right-above sub-template, a left-below sub-template, and a left-above sub-template. The implementations further include deriving at least one intra-prediction mode (IPM) based on cost calculations. The implementations include determining, based on the at least one IPM, a final predictor of the current video block. The implementations include performing the conversion based on the final predictor.

Abstract: Processing video data may include capturing the video data with multiple cameras and stitching the video data together to obtain a 360-degree video. A frame-packed picture may be provided based on the captured and stitched video data. A current sample location may be identified in the frame-packed picture. Whether a neighboring sample location is located outside of a content boundary of the frame-packed picture may be determined. When the neighboring sample location is located outside of the content boundary, a padding sample location may be derived based on at least one circular characteristic of the 360-degree video content and the projection geometry. The 360-degree video content may be processed based on the padding sample location.

Abstract: Method, apparatus and systems are disclosed. In one embodiment, a method of decoding includes obtaining a sub-block based motion prediction signal for a current block of the video; obtaining one or more spatial gradients of the sub-block based motion prediction signal or one or more motion vector difference values; obtaining a refinement signal for the current block based on the one or more obtained spatial gradients or the one or more obtained motion vector difference values; obtaining a refined motion prediction signal for the current block based on the sub-block based motion prediction signal and the refinement signal; and decoding the current block based on the refined motion prediction signal.

Abstract: Video coding methods are described for reducing latency in template-based inter coding. In some embodiments, a method is provided for coding a video that includes a current picture and at least one reference picture. For at least a current block in the current picture, a respective predicted value is generated (e.g. using motion compensated prediction) for each sample in a template region adjacent to the current block. Once the predicted values are generated for each sample in the template region, a process is invoked to determine a template-based inter prediction parameter by using predicted values in the template region and sample values the reference picture. This process can be invoked without waiting for reconstructed sample values in the template region. Template-based inter prediction of the current block is then performed using the determined template-based inter prediction parameter.