Andrea, thanks for excellent and thorough tests. I just do not understand why I get different results. I repeated my small tests again -- the previous last test was a very big one with 671,435 frames and not a normal case because it was a recording of a TV program with 11 audio channels. But this small one has only 2 audio channels. Maybe the difference is because they are video as well as audio (which does not make any sense)? ** rendered 2712 frames in* 0.189 secs, 14349.206 fps / Size=11474491* / 1.3.2 FLAC library ** rendered 2712 frames in *0.185 secs, 14659.459 fps / Size=11448186* / Compression-level=0 ** rendered 2712 frames in 0.319 secs, 8501.567 fps / Size= 9448273 / Compression-level=5 ** rendered 2712 frames in 0.656 secs, 4134.146 fps / Size= 8948373 / Compression-level=8 As you can see from above, compression-level of 0 most closely matches in time and size of the original output from 1.3.2 versus 1.4.0. I am going to do more tests and try on Ubuntu 16 and Debian 32-bit 11. Any thoughts?? On Sun, Sep 18, 2022 at 1:50 AM Andrea paz <[email protected]> wrote:
I added wav-type sources on top of the previous mp3-type sources to make the project a little bigger. Tests confirm that 'compression 5' is more efficient than 'compression 0'. I do not understand why. Phyllis' results seem more logical to me.
I also add the results obtained using the flac preset of ffmpeg always at compression 0; 5 and 8. The fps are comparable but not the size (due to optimization patches?).
Flac: ** rendered 26152 frames in 2.886 secs, 9061.677 fps, size 63.2 MiB
** rendered 26152 frames in 1.637 secs, 15975.565 fps, size 57.8 MiB
** rendered 26152 frames in 3.368 secs, 7764.846 fps, size 56.8 MiB
ffmpeg flac: ** rendered 26152 frames in 3.102 secs, 8430.690 fps, size 138.4 MiB
** rendered 26152 frames in 1.649 secs, 15859.309 fps, size 127.8 MiB
** rendered 26152 frames in 2.937 secs, 8904.324 fps, size 126.4 MiB
Flac 1.3 (appimage): ** rendered 26152 frames in 1.445 secs, 18098.270 fps, 63.4 MiB**