InterDigital CE Patent Holdings, SAS

Disclosed herein are systems, methods, and instrumentalities associated with transform coding. Candidate transforms for a video block may be ordered by a video encoder based on respective hypothetical costs associated with processing the video block based on the candidate transforms. A transform index value indicating the position of a suitable transform on the ordered transform list may then be signaled to a video decoder, which may perform similar cost-based sorting operations to derive the ordered transform list and select a transform for the video block from the ordered transform list based on the signaled index.

Systems, methods, and instrumentalities are disclosed for cross-component depth-luma coding. Depth information may be used as information for joint or cross-component coding of texture information (e.g., luma, chroma, or R, G, B colour component information). The depth information may be used to jointly code the corresponding texture information. In examples, a device may determine whether a coding mode, such as a joint depth-color coding mode (JDCC) (e.g., joint depth-luma coding mode (JDLC)) is used for a current block. If JDCC coding mode is used for the block, then the device may decode depth sample(s) in the block and predict the corresponding color sample(s) (e.g., luma sample(s) and/or chroma sample(s)) in the block based on the decoded depth sample(s).

Methods and apparatuses for encoding/decoding data content representative of a volumetric video are provided, wherein, the encoding/decoding comprises encoding in/decoding from a bitstream, an indicator specifying whether data content has information representative of at least one set of depth layers, the information representative of a set of depth layers specifying a number of depth layers and a depth value for each of the depth layers for a layer-based representation of the volumetric video. Methods and apparatuses for reconstructing Computer Generated Holograms from a reconstructed layered-based representation of the volumetric video are also provided.

A method for decoding or encoding comprising: determining if an inter prediction mode called weighted prediction is enabled for a current block; and, if weighted prediction is enabled, modifying a part of a reconstruction process of said current block related to the weighted prediction and/or to at least another prediction tool of a first set of prediction tools, the first set comprising a triangle prediction mode, a geometrical prediction mode, a local illumination compensation mode, a combined intra inter prediction mode, and modes in which motion information used for motion compensation may be refined based on an optical flow model.

For rendering a three-dimensional (3D) scene, a set of parameters for one or more objects in the 3D scene are identified. Intermediate structures and corresponding objects for each of the one or more objects in the 3D scene are determined based on the set of identified parameters. A hybrid acceleration structure is determined based on the determined intermediate structures and a classical acceleration structure. A color contribution is determined for the hybrid acceleration structure. The 3D scene is then rendered based on the determined color contribution of the hybrid acceleration structure.

Flexible neural architectures for end-to-end compression are used for image or video compression. In one embodiment, a first device is an encoder, and a second device is a decoder. Transmission of compressed features is performed between the first device and the second device. Mapping of probability tables is performed in the first and second devices. In one embodiment, mapping is communicated between the first and second devices by updating a bitstream used to encode features. In another embodiment, inference is split between user equipment and a network.

Apparatuses and methods are disclosed for encoding and decoding data, including video blocks. Techniques disclosed include adapting a decoder-side motion vector refinement (DMVR) process to coding tools with an illumination change. For a video block being encoded or decoded, the DMVR process includes determining that at least one coding tool, of the coding tools with the illumination change, is enabled; and performing the DMVR process, where a distortion metric is selected based on the enabled coding tool.

Systems, methods, and instrumentalities are disclosed for performing video coding using residual coefficient sign prediction with an adaptive cost function for intra prediction modes. An adaptive cost function for intra prediction modes (IPMs) may include multiple (e.g., selectable) cost functions, such as a default or standard cost function, a horizontal cost function, a vertical cost function, a diagonal up-right cost function, a diagonal down-right cost function, and/or a diagonal down-left cost function. A cost function may be selected based on one or more rules, which may be based on an IPM and a corresponding IPM index. An adaptive cost function for IPMs may include a cost function for multiple (e.g., all) directions. A cost function for multiple directions may include weighting factors related to an IPM and corresponding IPM index. A weighting parameter between the top and the left border continuity cost may be deduced from the IPM index.

Intra prediction modes are determined for entry in most probable mode lists for encoding in video encoders and decoding in video decoders. In at least one embodiment, coding modes are derived to determine reference samples to use from inter coded neighboring blocks. In one embodiment, up to five neighboring blocks are used. The reference samples are used in an intra prediction mode to determined prediction samples. The reference samples are used for prediction in encoding or decoding. In at least one embodiment, one of several motion models can be used to extract the intra motion mode. The intra motion modes are used to fill a most probable mode list. The motion model, intra mode, or reference frame can be signaled from an encoder to a decoder.

A method and device allow to reduce the pixel values an input image by combining a dimming map to the input image. This results in reducing the energy consumption required to display the dimmed image while preserving as much as possible the quality of experience when displaying the dimmed image. The reduction of the pixel values can be done by either reducing the luminance and optionally the chrominance or reducing the color components of the image. The dimming map is generated by a lightweight deep learning network based on a small set of parameters and with a target pixel values reduction rate. The generated dimming map aims at preserving visual similarity and is explicitly conditioned to respect specific constraints. For example, a smoothness constraint allows to the dimming map to be robust to downsampling. Two architectures and two training methods are proposed.