ML-based super resolution / upscaling

A new feature for viewpoint could be ML-based super resolution / upscaling. While such a feature exists in other software, it can only be applied after the RAW processing. As Deep Prime already performs demosaicing, it may be adapted in such a way that the image is interpolated onto a larger grid having 2x2 the size of the original image. This deep integration into the RAW processing workflow would probably be beneficial for the upscaling quality as any sharpening tools (e.g. lens sharpness) would only be applied after the upscaling step and could be fine-tuned to the image content.

This has previously been proposed in one form for comments and voting here: