According to embodiments described herein, sub-pictures and/or virtual boundaries can be used for coding an image. For example, sub-pictures in the current picture can be used for predicting, reconstructing, and/or filtering the current picture. Virtual boundaries can be used for filtering reconstructed samples of the current picture. Through image coding based on the subpictures and/or virtual boundaries according to embodiments described herein, the subjective/objective quality of an image can be improved, and the consumption of hardware resources necessary for the coding can be reduced.

Systems and methods are provided for implementing methods for resolution-adaptive video coding in a motion prediction coding format by obtaining a current frame of a bitstream, obtaining one or more reference pictures from a reference frame buffer, determining the obtained reference pictures that have resolutions different from a resolution of the current frame, translating an inter predictor of the one or more reference pictures, and generating a reconstructed frame from the current frame based on the one or more reference pictures and motion information of one or more blocks of the current frame, the motion information including at least one inter predictor, thereby achieving substantial reduction of network transport costs in video coding and delivery without requiring the transport of additional data that would offset or compromise these savings.

Systems and methods are provided for implementing methods for resolution-adaptive video coding in a motion prediction coding format by obtaining a current frame of a bitstream, obtaining one or more reference pictures from a reference frame buffer, determining the obtained reference pictures that have resolutions different from a resolution of the current frame, translating an inter predictor of the one or more reference pictures, and generating a reconstructed frame from the current frame based on the one or more reference pictures and motion information of one or more blocks of the current frame, the motion information including at least one inter predictor, thereby achieving substantial reduction of network transport costs in video coding and delivery without requiring the transport of additional data that would offset or compromise these savings.

Several techniques for video encoding and video decoding are described. One example method includes performing a conversion between a subpicture in a video picture of a video and a bitstream of the video according to a rule. The rule specifies that, in in case a subpicture is treated as a video picture for the conversion, a cross-layer alignment restriction is applied to less than all of the multiple layers including a current layer that includes the subpicture and a subset of layers associated with the current layer.

Techniques are described for improving intra-subpartitioning (ISP) mode for splitting coding blocks into sub-blocks. In some cases, whether ISP mode is enabled for a coding block is based on size constraints pertaining to data units (e.g., VPDUs, transform blocks, among others). For instance, based on a size constraint related to a VPDU, the ISP mode can be disabled for coding blocks crossing VPDU boundaries. In some cases, whether to enable ISP mode may be based on comparison of the width and/or height of the coding block to size thresholds corresponding to one or more maximum transform block sizes. In some cases, where the ISP mode is enabled for a coding block, a value of a flag used for defining a type of split, horizontal or vertical, for the coding block, can be inferred based on the width and/or height of the coding block relative to one or more thresholds.

A computer-implemented method of training an image generative network f.sub.θ for a set of training images, in which an output image {circumflex over (x)} is generated from an input image x of the set of training images non-losslessly, and in which a proxy network is trained for a gradient intractable perceptual metric that evaluates a quality of an output image {circumflex over (x)} given an input image x, the method of training using a plurality of scales for input images from the set of training images. In an embodiment, a blindspot network b.sub.α is trained which generates an output image {tilde over (x)} from an input image x. Related computer systems, computer program products and computer-implemented methods of training are disclosed.

This application relates to a method and apparatus, a storage medium, and a computer device for video encoding and decoding. The video encoding method includes: determining a sub-pixel interpolation mode, the sub-pixel interpolation mode comprising one of a direct sub-pixel interpolation mode or a sampled sub-pixel interpolation mode; acquiring motion estimation pixel precision corresponding to a current video frame; performing sub-pixel interpolation processing on a reference frame corresponding to the current video frame according to a resolution relationship between the current video frame and the reference frame, the motion estimation pixel precision, and the sub-pixel interpolation mode, to obtain a target reference frame; and encoding the current video frame according to the target reference frame, to obtain encoded data corresponding to the current video frame.

This application relates to a method and apparatus, a storage medium, and a computer device for video encoding and decoding. The video encoding method includes: determining a sub-pixel interpolation mode, the sub-pixel interpolation mode comprising one of a direct sub-pixel interpolation mode or a sampled sub-pixel interpolation mode; acquiring motion estimation pixel precision corresponding to a current video frame; performing sub-pixel interpolation processing on a reference frame corresponding to the current video frame according to a resolution relationship between the current video frame and the reference frame, the motion estimation pixel precision, and the sub-pixel interpolation mode, to obtain a target reference frame; and encoding the current video frame according to the target reference frame, to obtain encoded data corresponding to the current video frame.

Techniques are described for video encoding and decoding using reference picture resampling with switchable filters. One example involves obtaining a current picture and a reference picture, identifying filter index information for a current block of the current picture, and determining that a first picture size value of the current picture and a second picture size value of the reference picture are different. Based on the determining that the first picture size value of the current picture and the second picture size value of the reference picture are different, performing a resampling process using a default filter index in place of a current filter index identified by the filter index information. Additional examples can use the current filter index identified by the filter index information in subsequent blocks. In various examples, the current filter index can be derived or signaled.

Devices, systems and methods for digital video coding, which includes inter prediction with refinement, are described. An exemplary method of video processing includes determining to use, for a conversion between a current block of a video and a bitstream representation of the video, a first linear optimization model for the conversion using a first coding mode, the first linear optimization model being derived from a second linear optimization model that is used for the conversion using a second coding mode, and performing, based on the determining, the conversion. Another exemplary method of video processing includes determining to use, for a conversion between a current block of a video and a bitstream representation of the video, a gradient value computation algorithm for a bi-directional optical flow tool, and performing, based on the determining, the conversion.