color conversion method greyed out in vapoursynth

[Toltek]

I am really impressed by Hybrid so far, using vapoursynth, rock solid, seems to properly utilize GPU and much faster than megui+avisynth, which would sometimes stall or crash after many hours for no apparent reason. 

I digitized our family archive of Video8/Hi8/Digital8 many years ago  and recently started to deinterlace them with QTGMC to mp4 for easier viewing from the phones, network players etc, still keeping both the tapes and interlaced avi files for archival purposes.  Really happy with the results visually, just have a question about color conversion:

When I look at the generated .vpy script I see that the source is properly recognized as 4:1:1, converted to 4:4:4 for QTGMC and then to 4:2:0 for x264. No crop or resizing, the output is 720x480 with 59.94 framerate. I have only basic understanding of color spaces and conversion but I always assumed that if no resizing is performed it is better use resize.Point option instead of resize.Bicubic for color conversion? I see an option in Filtering/Vapoursynth/Misc/ColorResizer but it is greyed out and Bicubic.

Color wise the result looks the same as original, just wonder if I am missing something.

[Selur]

No, you are not missing anything.
Iirc I implemented it since < user requested it, but I never got any feedback from the user from the version I send him for testing and totally forgot about that option.

... but I always assumed that if no resizing is performed it is better use resize.Point option instead of resize.Bicubic for color conversion ...

Got any basis for this? I did some small tests quite a while back and they made it look like the resizer didn't matter, so if you can show me any basis for the assumtion that this isn't the case I'm happy to enable option.

Cu Selur

[Toltek]

Thanks for such fast response! I did try both and couldn't tell the difference, except in overscan lines at the bottom of some tapes, where there would be random colors flickering and if I stop and look at individual frames some edges in that overscan area would look slightly different, but not like one is better then the other. 

After some reading I think now that only  in highly resolved sharp images with strong color changes over single pixels, like red/yellow borders there might be tiny differences if you zoom in on a stopped frame.  I am also convinced  that bicubic is better and might have slightly  lesser artifacts in that case.