Search Results
!!m8ZMOnWSI0a/g/106025382#106037830
7/26/2025, 9:59:50 PM
Eureka!
None of the colorspace parameters would fucking stick but by editing the output webm now they do. Unfortunately this creates a temp webm file that you have to delete later but I think I have successfully committed VP9 video stream FRAUD. Updated batch script:
for %%f in (*.png) do (
ffmpeg -framerate 1 -stream_loop 1 -i "%%f" -c:v libvpx-vp9 -quality best -crf 10 -qmin 10 -qmax 10 -pix_fmt yuv420p -y temp.webm && ffmpeg -i temp.webm -color_range 1 -colorspace 1 -color_primaries 1 -color_trc 1 -c:v copy "%%~nf.webm"
)
Not completely sure what these things do but all 4 seem necessary to mimic actual VP9 video streams. If anyone has any playback problems, I don't know what to tell you, your software is borked beyond reason.
-color_range 1
-colorspace 1
-color_primaries 1
-color_trc 1
None of the colorspace parameters would fucking stick but by editing the output webm now they do. Unfortunately this creates a temp webm file that you have to delete later but I think I have successfully committed VP9 video stream FRAUD. Updated batch script:
for %%f in (*.png) do (
ffmpeg -framerate 1 -stream_loop 1 -i "%%f" -c:v libvpx-vp9 -quality best -crf 10 -qmin 10 -qmax 10 -pix_fmt yuv420p -y temp.webm && ffmpeg -i temp.webm -color_range 1 -colorspace 1 -color_primaries 1 -color_trc 1 -c:v copy "%%~nf.webm"
)
Not completely sure what these things do but all 4 seem necessary to mimic actual VP9 video streams. If anyone has any playback problems, I don't know what to tell you, your software is borked beyond reason.
-color_range 1
-colorspace 1
-color_primaries 1
-color_trc 1
!!m8ZMOnWSI0a/g/106025382#106035552
7/26/2025, 6:47:19 PM
>>106035279
We now have the ability to use a GPU instead of a CPU to decode an image now, here on 4chan. Apple is already doing this for HEIF like it's been discussed in this thread. Their solution is obviously more refined than a shitty hack that we're trying to punk out but it's there if you want it.
>>106035334
Good point but I still find this amusing. HEIC support would have been better but it only works on Apple devices which is bullshit.
https://caniuse.com/?search=HEIC
Since this thread is still up maybe some of you anons can help me improve this hack somehow. Currently there are playback issues and they might have something to do with this:
https://trac.ffmpeg.org/wiki/colorspace
When you open my webms in MPC you'll find this in mediainfo tab:
>Color primaries : BT.709
>Transfer characteristics : sRGB/sYCC
Anyway maybe we're supposed to mirror HEIC images regarding this info? I curiously opened a few HEIC images in MPC and I found this in their mediainfo tab:
>Color primaries : Display P3
>Transfer characteristics : BT.709
>Matrix coefficients : BT.601
SOOOOO the playback issues might not be because of -framerate 1 -stream_loop 1 but instead they're due to that missing matrix coefficient information and wrong color primaries/transfer characteristics????
We now have the ability to use a GPU instead of a CPU to decode an image now, here on 4chan. Apple is already doing this for HEIF like it's been discussed in this thread. Their solution is obviously more refined than a shitty hack that we're trying to punk out but it's there if you want it.
>>106035334
Good point but I still find this amusing. HEIC support would have been better but it only works on Apple devices which is bullshit.
https://caniuse.com/?search=HEIC
Since this thread is still up maybe some of you anons can help me improve this hack somehow. Currently there are playback issues and they might have something to do with this:
https://trac.ffmpeg.org/wiki/colorspace
When you open my webms in MPC you'll find this in mediainfo tab:
>Color primaries : BT.709
>Transfer characteristics : sRGB/sYCC
Anyway maybe we're supposed to mirror HEIC images regarding this info? I curiously opened a few HEIC images in MPC and I found this in their mediainfo tab:
>Color primaries : Display P3
>Transfer characteristics : BT.709
>Matrix coefficients : BT.601
SOOOOO the playback issues might not be because of -framerate 1 -stream_loop 1 but instead they're due to that missing matrix coefficient information and wrong color primaries/transfer characteristics????
Page 1