contain a lot of experimental/debug stuff.. Mostly for Andrea
On Wednesday, August 11, 2021, Andrew Randrianasulu <[email protected]> wrote:
contain a lot of experimental/debug stuff..
Mostly for Andrea
aha, this series breaks mpeg video encoding, I fixed mjpegtools build differently and restored it (still it doesn't fully utilize even 8 cpus..) . Add also patch 0070 (attached) on top of those.. sorry
I would like to remove the Termux patches from randrik15. They work for Andrew, I've tested that applying them on git sources does not create errors; at this point I would remove them until a decision is made on whether to implement them. I would avoid a lot of confusion about the remaining patches. PS: I don't understand well the function of "compile_multibit_X265.txt". Even without patching it and moving it to .../src, the 10 and 12 bit x265 profiles appear anyway and work fine. (they also work without applying the 32-60-61-64 patch).
On Mon, 2021-08-16 at 10:22 +0200, Andrea paz via Cin wrote:
I would like to remove the Termux patches from randrik15. They work for Andrew, I've tested that applying them on git sources does not create errors; at this point I would remove them until a decision is made on whether to implement them. I would avoid a lot of confusion about the remaining patches.
PS: I don't understand well the function of "compile_multibit_X265.txt". Even without patching it and moving it to .../src, the 10 and 12 bit x265 profiles appear anyway and work fine. (they also work without applying the 32-60-61-64 patch).
How do you test that the result is indeed 10 or 12 bits? My TV does not seem to support it. MatN
How do you test that the result is indeed 10 or 12 bits? My TV does not seem to support it.
Not even my old TV supports 10-bit (By January I will have to buy the decoder that supports 10-bit Hevc). I simply rendered without problems in CinGG and then parsed the file with mediainfo and ffprobe. I don't know if that's adequate.
Andrea, I do not pretend to understand the 10/12 bit world, but about: PS: I don't understand well the function of
"compile_multibit_X265.txt". Even without patching it and moving it to .../src, the 10 and 12 bit x265 profiles appear anyway and work fine. (they also work without applying the 32-60-61-64 patch).
The default pixel value for h265-10bit and h265-12bit is yuv42*2p10le* and yuv42*2p12le* respectively. There is a whole lot of discussion about 420 versus 422 (again, which I do not understand). If you try to render with these pixel values in the 8-bt version of CinGG you will get this error message and immediate failure: [libx265 @ 0x7f2cd4efc1c0] Specified pixel format yuv422p12le is invalid or not supported However, in the multibit CinGG, they will work AND there is definite color differences noticeable through out the whole video when I run ydiff on a pixel format of yuv422p10le versus yuv420.
participants (4)
-
Andrea paz -
Andrew Randrianasulu -
mat -
Phyllis Smith