September 2024 - Cin - Cinelerra-GG Mailing Lists

FranceBB's LUT collection
by Andrew Randrianasulu 16 May '25

16 May '25

forum https://forum.doom9.org/showthread.php?t=176091 it says those LUTs should be usable for ffmpeg's lut3d and therefore for CinGG? github (each *.cube can be several mb's big) https://github.com/FranceBB/LinearTransformation author says they were verified on professional equipment... list includes PAL and NTSC variants for rec. 601....

3 6

I tried to add AVC-Intra 100 output profiles
by Andrew Randrianasulu 08 May '25

08 May '25

But not sure if I did this correctly. https://en.wikipedia.org/wiki/AVC-Intra says for AVC-Intra 100: AVC-Intra 100: nominally 100 Mbit/s, size of each frame is fixed CAVLC entropy coding only. All formats are High 4:2:2 Intra Profile, Level 4.1 4:2:2 chrominance sampling frames are not scaled and yeah, this French-language pdf actually says the same: https://web.archive.org/web/20110720215759/http://www.ficam.fr/upload/docum… 50M 100M Standard MPEG-4 part10 AVC/H.264 Profile High 10 Intra High 4:2:2 Intra Level Level 3.2/4.0 Level 4.1 Resolution 1440x1080 / 960x720 1920x1080 / 1280x720 Sampling 4:2:0 4:2:2 Bit rate 54Mbps 111Mbps Block size Y: 8x8, C: 4x4 Intra Prediction Y: 9mode, C: 4mode MBAFF Yes De-block filter No Slice/frame 10 Entropy coding CABAC CAVLC so, I looked up ffmpeg cmd line https://trac.ffmpeg.org/ticket/3597 ffmpeg -v info -i /mnt-data/backup4/vol0/export/CUHK\ 2014-1/V017b.mov -vcodec libx264 -intra -pix_fmt yuv422p10le -vb 100M -x264opts avcintra-class=100 -x264opts colorprim=bt709 -x264opts transfer=bt709 -x264opts colormatrix=bt709 -threads 4 -acodec copy V017b-ffmpeg-git-avcintra100.mov Huh, said I and entered those params in opts file, actually in two files (mp4 and mov containers) mov libx264 cin_pix_fmt=yuv422p10le flags -global_header keyint_min=0 x264-params=keyint=0:colorprim=bt709:transfer=bt709:colormatrix=bt709 avcintra-class=100 bufsize=100000000 level=4.1 mp4 libx264 cin_pix_fmt=yuv422p10le flags -global_header keyint_min=0 x264-params=keyint=0:colorprim=bt709:transfer=bt709:colormatrix=bt709 avcintra-class=100 bufsize=100000000 level=4.1 both for progressive content, for interlace you must add some more flags: https://lists.ffmpeg.org/pipermail/ffmpeg-user/2017-January/035049.html -flags +ildct+ilme -x264opts interlaced=1 Also, there is still lack of (GUI) user notification if file writing failed due to NO SPACE on device :} FFStream::write_packet: write packet failed. file:/dev/shm/avci_new.mp4 err: На устройстве не осталось свободного места FFStream::encode_frame: encode failed. file: /dev/shm/avci_new.mp4 err: На устройстве не осталось свободного места FFStream::flush failed :file:/dev/shm/avci_new.mp4 err: Операция не позволена Render::render_single: Session finished. ** rendered 273 frames in 85.585 secs, 3.190 fps Render::render_single: Session finished. ** rendered 273 frames in 85.485 secs, 3.194 fps Render::render_single: Session finished. ** rendered 273 frames in 82.176 secs, 3.322 fps Profile sort of assumes you set project to some supported resolution, framerate, etc, otherwise libx264 will fail with console error message..... I found some talk about AVC-Ultra, there even was patch for x264, but it was not merged: https://mailman.videolan.org/pipermail/x264-devel/2015-September/011412.html Also, I think sound must be uncompressed, but I'm not sure.

2 5

vpl-inspect
by Andrew Randrianasulu 11 Oct '24

11 Oct '24

May be ffmpeg tries onevpl on new hardware? https://bugs.gentoo.org/590752 example output https://bugs.gentoo.org/attachment.cgi?id=835909

2 11

x265 4.0 quietly out
by Andrew Randrianasulu 09 Oct '24

09 Oct '24

https://bitbucket.org/multicoreware/x265_git/commits/6318f223684118a2c71f67… ====== Version 4.0 =========== Release date - 13th September, 2024. New feature ----------- 1. Alpha Channel feature. 2. Screen Content Coding (SCC). 3. MV-HEVC feature. Enhancements to existing features --------------------------------- 1. Added support for the VMAF v3.x. API changes ----------- 1. Add command line parameter for Alpha Channel feature :option:`--alpha`. 2. Add command line parameter for SCC feature :option:`--scc 1`. 3. Add command line parameters for the MV-HEVC feature :option:`--multiview-config "multiview_config.txt"`. Optimizations --------------------- 1. Arm SIMD optimizations: Several time-consuming scalar C functions now have SIMD implementations on Arm platforms. Existing Arm SIMD implementations have also been optimized. These optimizations result in up to 57% faster encoding compared to release 3.6. 2. Arm SIMD optimizations include use of Armv8.4 DotProd, Armv8.6 I8MM, and Armv9 SVE2 instruction set extensions. The following algorithms now have optimized SIMD implementations: SAD, SSE, DCT, SAO, convolution, quantization, intra_planar, intraFilter, intrapred DC and IDCT16x16. Bug fixes --------- 1. Fix for y4m pipe input broken. 2. Fix SCC crash on multipass encode. 3. Fix mcstf when :option:`--bframes` value was less than 5. 4. Fix lowpass DCT for high bit depth. 5. Added build support for Visual Studio 17. 6. Fix issue in default code flow and memory leak. 7. Framethreads tuning for Windows ARM devices. 8. Fix scc crash on multipass encode. ==== I think only ffmpeg 7.1 (already branched) will have multiview (stereo?) HEVC decoding, not even sure how(if) it can be represented in cinelerras ... Time to test, I guess.

2 4

August/September changes with new AppImages
by Phyllis Smith 05 Oct '24

05 Oct '24

New AppImages are now available at: https://cinelerra-gg.org/download/images/ but the special Red Hat older version which is Fedora Core 29 is not available because the build went into an infinite loop and I have not analyzed the problem yet (CinGG-20240930-x86_64-older-distros-multibit.AppImage). For the newer Operating Systems, the packages that Andrey handles are always up to date at (note it is split into 2 different days because of time difference): https://github.com/einhander/cin-gg-packages/releases/tag/20240927 https://github.com/einhander/cin-gg-packages/releases/tag/20240926 Always THANKS to the contributors,package creators, testers, forum questions/answers, and email questions/answers/updates/notes. Release notes are included below. *************************** GIT for Cinelerra-GG has the following changes from 08/01/2024-09/30/2024 *Library Upgrades:* *Dcraw* has been updated to the latest version of 1.478. It was 1.477 of 2016 and should have been updated earlier to the last 2018 version from Dave Coffin. It looks like there are 50+ additional cameras added, such as the GoPro HERO5 Black, Nikon COOLPIX 8700, Fujifilm X-T2, Olympus TG-5, Sony ILCE-9, etc. *Flac* library was upgraded from 1.4.2 to 1.4.3 and may be slightly faster on encoding. *Giflib* has been updated from 5.2.1 to 5.2.2 just to keep it to the most recent version. *SVT_AV1* has been upgraded to 2.2.1 from 1.8.0 and appears to be somewhat faster on encoding. Terje’s test show about 14% faster in 1 case and 41% faster in another case. File size could be about 12% bigger – but your results could be wildly different. *Openjpeg *was upgraded from 2.5.0 to 2.5.2. *Tiff *4.6.0 has been updated to 4.6.0t which puts back a lot of things that had been deleted in 4.6.0 due to security. *X264* is now up to the date of May, 2024 as release r3191 from previous release of r3106. Render format, av1.webm, has been renamed *av1_aom.webm* and keyframes placed judiciously in order that seeks will work better in CinGG. Change courtesy of *Terje*’s testing work. *Andrew mods* *and fixes *include: - An additional info/detail in the Resources window for ffmpeg files that have transfer/primaries color attributes like bt709/bt601/bt2020. - Updated for newer O/S with Boost 1.85 is tools/makeappimagetool/appdir.cpp and core.cpp. - Fixed linking error if esound is enabled - found while working on Termux.

2 1

Is one pre-built dynamic and static binary for user selection possible?
by Terje J. Hanssen 30 Sep '24

30 Sep '24

If my understanding is correct, the standard pre-built CinGG packages are static binaries with all required thirdparty, ffmpeg and libs included(?) Would it be possible to combine static and dynamic in one pre-built so that users via setting could select what they want, ffmpeg and libs included or from the system? Or is two different pre-builts required? Experiences with the latter? --------- Refer to extracted sections from the CinGG manual Chapter 1 - Installation below: If you prefer to not have to take the time to build CINELERRA-GG Infinity yourself, there are pre-built dynamic or static binaries for various versions of Ubuntu, Mint, Suse, Fedora, Debian ........ The pkgs directory contains the standard packaged application for various distros. This will install a dynamic system version for users who prefer to have the binaries in the system area --------- Notes about Building from Git in your Customized Environment Getting a build to work in a custom environment is not easy. If you have already installed libraries which are normally in the thirdparty build, getting them to be recognized means you have to install the devel version so the header files which match the library interfaces exist. If you want to build using only the thirdparty libraries installed in your system, just include "–without-thirdparty" to your configure script. For example: ./confgure --with-single-user --disable-static-build --without-thirdparty Below is the list of thirdparty builds, but this list may have changed over time. Table 1.1: List of thirdparty builds a52dec yes djbfft yes ffmpeg yes fftw auto flac auto giflib yes ilmbase auto lame auto libavc1394 auto libraw1394 auto libiec61883 auto libdv auto .......... The yes means force build and auto means probe and use the system version if the build operation is not static. To get your customized build to work, you need to change the probe options for the conflicting libraries from yes to auto, or even rework the configure.ac script. There may be several libraries which need special treatment. An example of a problem you might encounter with your customized installation is with a52dec which has probes line (CHECK_LIB/CHECK_HEADERS) in configure.ac, but djbfft does not. In this case, djbfft is only built because a52dec is built, so if your system has a52dec, set a52dec to auto and see if that problem is solved by retrying the build with: ./confgure --with-single-user -enable-a52dec=auto . With persistence, you can get results, but it may take several tries to stabilize the build. If you need help, email the log and config.log, which is usually sufficient to determine why a build failed. If you have already installed the libfdk_aac development package on your computer because you prefer this version over the default aac, you will have to do the following to get this alternative operational. The libfdk_aac library is not a part of CINELERRA-GG by default because it is not license free. export FFMPEG_EXTRA_CFG=" --enable-libfdk-aac --enable-nonfree" export EXTRA_LIBS=" -lfdk-aac" for f in `grep -lw aac cinelerra-5.1/ffmpeg/audio/*`; do sed -e 's/\<aac\>/libfdk_aac/' -i $f done

5 31

so, it looks like qsv integration a bit harder than just adding few lines ....
by Andrew Randrianasulu 29 Sep '24

29 Sep '24

got answer from ffmpeg devs https://ffmpeg.org//pipermail/ffmpeg-devel/2024-September/333906.html

2 26

uncompressed video in mp4 - may be coming to ffmpeg
by Andrew Randrianasulu 26 Sep '24

26 Sep '24

https://ffmpeg.org//pipermail/ffmpeg-devel/2024-September/333964.html but I guess post-7.1 https://ffmpeg.org//pipermail/ffmpeg-devel/2024-September/333685.html "[FFmpeg-devel] DNxUncompressed decoder" ah, cute Dirac decoding via Vulkan! https://ffmpeg.org//pipermail/ffmpeg-devel/2024-September/333405.html but not sure what kind of driver expose it?

1 0

fix linking with libesd
by Andrew Randrianasulu 26 Sep '24

26 Sep '24

was playing with esound (enlightenment sound daemon) on termux, found linking error if esound was enabled. Can anyone please test this patch? hopefully it does not break default build ... but sound still does not work in termux - I get segfault.

2 2

Multitrack performance issues
by Stefan de Konink 26 Sep '24

26 Sep '24

I think have a significant opportunity to improve the Cinerella performance, under normal editing conditions, by n_videotracks-fold. At this point in time I notice that the CPU usage (decoding) for a section where multiple tracks are stacked, but only the top one is actually on screen (read opacity 100%) could be improved. I observe the following: - for a disabled video effect, the CPU still runs - for stacked video, the CPU still runs for both videos - the effect must be detached and the video below must have play track disabled to prevent the CPU to go from 20% to 150% - this is independent of colourspace (RGB, RGBA, YUV or YUVA) -- Stefan

3 6