VQV x264 vs 10b VP9/AV1 test
If this thread posts it means 4chan supports 10-bit chroma city through VP9.
Alright it took me a while to compromise on some things but I think I've made the best video quality validation test given the meager 4MB /g/ webm file size limit. Original video source was 4K video de-banded and scaled down to 540p and finally converted to a 2Mbps 10b VP9 webm which SHOULD be overkill for this resolution. Beside 4shitter's anemic limits I also chose 540p because it's a sort of "torture test" for modern video codecs meant to exploit redundancies in 1080p+. x264 gets slaughtered in 1080p+ so I see no point in testing the obvious anyway.
Quality validation passes when VMAF exceeds 95.
10b VP9 first then 10b AV1 and finally x264. Might only post a few times and call it quits cause I don't feel too good right now. However I think I figured out the "base" settings for 2 pass 10b VP9 encodes, a lot of these settings won't even work with 8b VP9. No DO loops this time because it's webm to webm.
#!/bin/bash
ffmpeg -i input.webm -c:v libvpx-vp9 \
-profile:v 2 -pix_fmt yuv420p10le \
-auto-alt-ref 6 -arnr_max_frames 7 -arnr_strength 2 \
-b:v 0 -crf 38 \
-quality good -cpu-used 1 -aq-mode 0 \
-tile-columns 1 -row-mt 1 -enable-tpl 0 \
-pass 1 -an -f null -
ffmpeg -i input.webm -c:v libvpx-vp9 \
-profile:v 2 -pix_fmt yuv420p10le \
-auto-alt-ref 6 -arnr_max_frames 7 -arnr_strength 2 \
-b:v 0 -crf 38 \
-quality good -cpu-used 1 -aq-mode 0 \
-tile-columns 1 -row-mt 1 -enable-tpl 0 \
-pass 2 -an output.webm
read -p "Press Enter to get VMAF, else ctrl + c"
ffmpeg -i output.webm -i input.webm -filter_complex libvmaf -f null -
Alright it took me a while to compromise on some things but I think I've made the best video quality validation test given the meager 4MB /g/ webm file size limit. Original video source was 4K video de-banded and scaled down to 540p and finally converted to a 2Mbps 10b VP9 webm which SHOULD be overkill for this resolution. Beside 4shitter's anemic limits I also chose 540p because it's a sort of "torture test" for modern video codecs meant to exploit redundancies in 1080p+. x264 gets slaughtered in 1080p+ so I see no point in testing the obvious anyway.
Quality validation passes when VMAF exceeds 95.
10b VP9 first then 10b AV1 and finally x264. Might only post a few times and call it quits cause I don't feel too good right now. However I think I figured out the "base" settings for 2 pass 10b VP9 encodes, a lot of these settings won't even work with 8b VP9. No DO loops this time because it's webm to webm.
#!/bin/bash
ffmpeg -i input.webm -c:v libvpx-vp9 \
-profile:v 2 -pix_fmt yuv420p10le \
-auto-alt-ref 6 -arnr_max_frames 7 -arnr_strength 2 \
-b:v 0 -crf 38 \
-quality good -cpu-used 1 -aq-mode 0 \
-tile-columns 1 -row-mt 1 -enable-tpl 0 \
-pass 1 -an -f null -
ffmpeg -i input.webm -c:v libvpx-vp9 \
-profile:v 2 -pix_fmt yuv420p10le \
-auto-alt-ref 6 -arnr_max_frames 7 -arnr_strength 2 \
-b:v 0 -crf 38 \
-quality good -cpu-used 1 -aq-mode 0 \
-tile-columns 1 -row-mt 1 -enable-tpl 0 \
-pass 2 -an output.webm
read -p "Press Enter to get VMAF, else ctrl + c"
ffmpeg -i output.webm -i input.webm -filter_complex libvmaf -f null -
