Two step cross-component prediction mode

1. A method of video processing, comprising:

deriving a prediction block for a chroma block of a current video block of a video based on a temporary video block, wherein the temporary video block is derived from a reconstructed luma block of the current video block, both the reconstructed luma block and the temporary video block have a first size, and the chroma block has a second size which is different from the first size; and

performing, a conversion between the current video block and a bitstream of the video based on the prediction block for the chroma block according to a cross-component prediction mode,

wherein the first size is (M′+W0)×(N′+H0), the second size is M×N, and

wherein either M′ is unequal to M or N′ is unequal to N, or

wherein one or both of W0 and H0 are equal to a value of zero.

2. The method of claim 1 ,

wherein the temporary video block is associated with a first set of samples and the reconstructed luma block is associated with a second set of samples,

wherein at least one sample S Temp (x0, y0) in the first set of samples is derived as a first function of a corresponding sample S(x0,y0) from the second set of samples,

wherein x0 is within a first range of zero to (M′−1), inclusive, and y0 is within a second range of zero to (N′−1), inclusive, and

wherein M′ represents a width of the first size, and N′ represents a height of the second size.

3. The method of claim 2 , wherein in the first function, S Temp (x0, y0) is defined as ((S (x0, y0)*a)>>k)+b.

4. The method of claim 2 , the method further comprising:

clipping the S Temp (x0, y0) to an allowed range of chroma values.

5. The method of claim 1 , wherein the prediction block is generated by applying one or more downsampling filters to samples in the temporary video block.

6. The method of claim 5 , wherein the one or more downsampling filters are pre-defined.

7. The method of claim 5 , wherein a selection of the one or more downsampling filters is based on a relative position of a sample to be predicted.

8. The method of claim 7 , wherein the one or more downsampling filters includes a downsampling filter with coefficient [1 1] for a sample located at (0, 0) relative to the temporary video block.

9. The method of claim 5 , wherein the one or more downsampling filters includes a 6-tap filter for a sample not located at (0, 0) relative to the temporary video block.

10. The method of claim 9 , wherein the 6-tap filter includes coefficients [1 2 1; 1 2 1].

11. The method of claim 7 , wherein a prediction sample of the chroma block only depends on samples located at (2*x, 2*y) and (2*x, 2*y+1), wherein the prediction sample is located at (x, y), x and y are integers and a top-left sample's coordinates of the temporary video block is set to (0, 0).

12. The method of claim 1 , wherein the conversion includes encoding the current video block into the bitstream.

13. The method of claim 1 , wherein the conversion includes decoding the current video block from the bitstream.

14. 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:

derive a prediction block for a chroma block of a current video block of a video based on a temporary video block, wherein the temporary video block is derived from a reconstructed luma block of the current video block, both the reconstructed luma block and the temporary video block have a first size, and the chroma block has a second size which is different from the first size; and

perform, a conversion between the current video block and a bitstream of the video based on the prediction block for the chroma block according to a cross-component prediction mode,

wherein the first size is (M′+W0)×(N′+H0), the second size is M×N, and

wherein either M′ is unequal to M or N′ is unequal to N, or wherein one or both of W0 and H0 are equal to a value of zero.

15. A non-transitory computer-readable storage medium storing instructions that cause a processor to:

derive a prediction block for a chroma block of a current video block of a video based on a temporary video block, wherein the temporary video block is derived from a reconstructed luma block of the current video block, both the reconstructed luma block and the temporary video block have a first size, and the chroma block has a second size which is different from the first size; and

perform, a conversion between the current video block and a bitstream of the video based on the prediction block for the chroma block according to a cross-component prediction mode,

wherein the first size is (M′+W0)×(N′+H0), the second size is M×N, and

wherein either M′ is unequal to M or N′ is unequal to N, or wherein one or both of W0 and H0 are equal to a value of zero.

16. A non-transitory computer-readable recording medium storing a bitstream of a video which is generated by a method performed by a video processing apparatus, wherein the method comprises:

deriving a prediction block for a chroma block of a current video block of the video based on a temporary video block, wherein the temporary video block is derived from a reconstructed luma block of the current video block, both the reconstructed luma block and the temporary video block have a first size, and the chroma block has a second size which is different from the first size; and

generating the bitstream based on the prediction block for the chroma block according to a cross-component prediction mode,

wherein the first size is (M′+W0)×(N′+H0), the second size is M×N, and

wherein either M′ is unequal to M or N′ is unequal to N, or wherein one or both of W0 and H0 are equal to a value of zero.