Chroma factor applied twice after combination of scaling window proposals

[ksuehring]

The inference of the scaling window from the cropping window was integrated exactly as proposed in JVET-Q0399:

scaling_win_left_offset, scaling_win_right_offset, scaling_win_top_offset, and scaling_win_bottom_offset specify the offsets that are applied to the picture size for scaling ratio calculation. When not present, the values of scaling_win_left_offset, scaling_win_right_offset, scaling_win_top_offset, and scaling_win_bottom_offset are inferred to be equal to SubWidthC * pps_conf_win_left_offset, SubWidthC * pps_conf_win_right_offset, SubHeightC * pps_conf_win_top_offset, and SubHeightC * pps_conf_win_bottom_offset, respectively.

In the proposal we had to account for the correct chroma scaling factor in the inference, because the scaling window used luma sample units while the conformance window used chroma sample units.

With proposal JVET-Q0487 the units of the scaling window were changed to chroma units. Thus no scaling is required anymore.

The inference should be corrected as follows:

scaling_win_left_offset, scaling_win_right_offset, scaling_win_top_offset, and scaling_win_bottom_offset specify the offsets that are applied to the picture size for scaling ratio calculation. When not present, the values of scaling_win_left_offset, scaling_win_right_offset, scaling_win_top_offset, and scaling_win_bottom_offset are inferred to be equal to pps_conf_win_left_offset, pps_conf_win_right_offset, pps_conf_win_top_offset, and pps_conf_win_bottom_offset, respectively.

Otherwise the chroma scaling would be applied twice in the case of inference.

[yk]

This was fixed already (I don't know when it was fixed).