Simplified coding of generalized bi-directional index
Devices, systems and methods for digital video coding, which include decoder side motion vector derivation (DMVD) tools, are described. An exemplary method for video processing includes making a decision, based on a determination that a current block of a video is coded using a multi-hypothesis prediction mode, regarding a selective enablement of a DMVD tool for the current block, wherein the DMVD tool derives a refinement of motion information signaled in a bitstream representation of the video; and performing, based on the decision, a conversion between the current block and the bitstream representation.
1. A method for video processing, comprising:
determining, for a conversion between a current block of a video and a bitstream of the current block, a weight index of a first prediction mode, wherein the first prediction mode is configured to select a weight w from a set of weights based on the weight index to generate a bi-prediction signal for the current block, and wherein the set of weights comprise multiple weights that are different from ½; and
performing, based on the weight index, the conversion between the current block and the bitstream, wherein the weight w and a weight 1-w are respectively applied to two prediction blocks generated from reference picture list 0 and reference picture list 1 to derive the bi-prediction signal;
wherein the weight index is represented by a bin string which is comprised in the bitstream and is used to select the weight w from the set of weights, and the bin string comprises a plurality of bins and at least one bin of the plurality of bins is bypass coded, wherein the set of weight comprises all probable weights different from ½ corresponding to the first prediction mode,
wherein a first bin of the plurality of bins is coded with at least one context, and wherein all other bins of the plurality of bins are bypass coded, and wherein the at least one context consists of one context, and
wherein in response to the weight w being not equal to the weight 1-w, a second prediction mode is not applied for the current block, wherein the second prediction mode uses gradients in different directions to derive a motion offset and a prediction offset.
2. The method of claim 1 , wherein the at least one context consists of three contexts.
3. The method of claim 2 , wherein the three contexts are defined as:
ctxIdx=aboveBlockIsGBIMode+leftBlockIsGBIMode,
wherein aboveBlockIsGBIMode=1 if an above neighboring block to the current block is coded using the GBI mode and zero otherwise,
wherein leftBlockIsGBIMode=1 if a left neighboring block to the current block is coded using the first prediction mode and zero otherwise.
4. The method of claim 1 , wherein each of a first K bins of the plurality of bins is coded with at least one context, wherein all other bins of the plurality of bins are bypass coded, wherein K is a non-negative integer, wherein 0≤K≤maxGBIIdxLen, and wherein maxGBIIdxLen is a maximum length of the plurality of bins.
5. The method of claim 4 , wherein one context is shared for the first K bins except for a first bin of the first K bins.
6. The method of claim 4 , wherein one context is used for each of the first K bins except for a first bin of the first K bins.
7. The method of claim 1 , wherein the conversion comprises decoding the current block from the bitstream.
8. The method of claim 1 , wherein the conversion comprises encoding the current block into the bitstream.
9. An apparatus for processing video data comprising a processor and a non-transitory memory with instructions thereon, wherein the instructions upon execution by the processor, cause the processor to:
determine, for a conversion between a current block of a video and a bitstream of the current block, a weight index of a first prediction mode, wherein the first prediction mode is configured to select a weight w from a set of weights based on the weight index to generate a bi-prediction signal for the current block, and wherein the set of weights comprise multiple weights that are different from ½; and
perform, based on the weight index, the conversion between the current block and the bitstream, wherein the weight w and a weight 1-w are respectively applied to two prediction blocks generated from reference picture list 0 and reference picture list 1 to derive the bi-prediction signal;
wherein the weight index is represented by a bin string which is comprised in the bitstream and is used to select the weight w from the set of weights, and the bin string comprises a plurality of bins and at least one bin of the plurality of bins is bypass coded, wherein the set of weight comprises all probable weights different from ½ corresponding to the first prediction mode,
wherein a first bin of the plurality of bins is coded with at least one context, and wherein all other bins of the plurality of bins are bypass coded, and wherein the at least one context consists of one context, and
wherein in response to the weight w being not equal to the weight 1-w, a second prediction mode is not applied for the current block, wherein the second prediction mode uses gradients in different directions to derive a motion offset and a prediction offset.
10. The apparatus of claim 9 , wherein the conversion comprises decoding the current block from the bitstream.
11. The apparatus of claim 9 , wherein the conversion comprises encoding the current block into the bitstream.
12. A non-transitory computer-readable storage medium storing instructions that cause a processor to:
determine, for a conversion between a current block of a video and a bitstream of the current block, a weight index of a first prediction mode, wherein the first prediction mode is configured to select a weight w from a set of weights based on the weight index to generate a bi-prediction signal for the current block, and wherein the set of weights comprise multiple weights that are different from ½; and
perform, based on the weight index, the conversion between the current block and the bitstream, wherein the weight w and a weight 1-w are respectively applied to two prediction blocks generated from reference picture list 0 and reference picture list 1 to derive the bi-prediction signal;
wherein the weight index is represented by a bin string which is comprised in the bitstream and is used to select the weight w from the set of weights, and the bin string comprises a plurality of bins and at least one bin of the plurality of bins is bypass coded, wherein the set of weight comprises all probable weights different from ½ corresponding to the first prediction mode,
wherein a first bin of the plurality of bins is coded with at least one context, and wherein all other bins of the plurality of bins are bypass coded, and wherein the at least one context consists of one context, and
wherein in response to the weight w being not equal to the weight 1-w, a second prediction mode is not applied for the current block, wherein the second prediction mode uses gradients in different directions to derive a motion offset and a prediction offset.
13. The storage medium of claim 12 , wherein the conversion comprises decoding the current block from the bitstream.
14. The storage medium of claim 12 , wherein the conversion comprises encoding the current block into the bitstream.
15. A method for storing bitstream of a video, comprising:
determining, for a conversion between a current block of a video and a bitstream of the current block, a weight index of a first prediction mode, wherein the first prediction mode is configured to select a weight w from a set of weights based on the weight index to generate a bi-prediction signal for the current block, and wherein the set of weights comprise multiple weights that are different from ½;
generating, based on the weight index, the bitstream from the current block; wherein the weight w and a weight 1-w are respectively applied to two prediction blocks generated from reference picture list 0 and reference picture list 1 to derive the bi-prediction signal; and
storing the bitstream in a non-transitory computer-readable recording medium,
wherein the weight index is represented by a bin string which is comprised in the bitstream and is used to select the weight w from the set of weights, and the bin string comprises a plurality of bins and at least one bin of the plurality of bins is bypass coded, wherein the set of weight comprises all probable weights different from ½ corresponding to the first prediction mode,
wherein a first bin of the plurality of bins is coded with at least one context, and wherein all other bins of the plurality of bins are bypass coded, and wherein the at least one context consists of one context, and
wherein in response to the weight w being not equal to the weight 1-w, a second prediction mode is not applied for the current block, wherein the second prediction mode uses gradients in different directions to derive a motion offset and a prediction offset.