- Cin - Cinelerra-GG Mailing Lists

OpenCL again
by Andrew Randrianasulu 27 Feb '23

27 Feb '23

So, I run veeery experimental clvk (OpenCL via Vulkan) on my termux, ffmpeg tried to init filter but died. I guess my day is not today ... Anyway, my test line was: ffmpeg -init_hw_device opencl=gpu -filter_hw_device gpu -i ~/20210419_055507A.mp4 -vf "hwupload, boxblur_opencl, hwdownload,format=yuv420p" -f null /dev/null for different filters format might be yuv420p or nv12 ... I wonder if anyone here have working OpenCL ffmpeg setup ? I'll try with llvmpipe/rusticl on desktop .... but it guaranteed to be slow.

4 29

helper_cuda.h
by Stefan de Konink 20 Feb '23

20 Feb '23

Where should this pretty one come from? nbody.h:27:10: fatal error: helper_cuda.h: No such file or directory 27 | #include <helper_cuda.h> | ^~~~~~~~~~~~~~~ -- Stefan

2 1

Natron, background render and MantisBT #635
by Andrea paz 17 Feb '23

17 Feb '23

I am writing in the mailing list because today I cannot post on MantisBT. However, we had gone OT from the original request. @Andrew-R Your idea of how to do the Natron/CinGG interaction is also mine, otherwise you fall back to rendering to go to another program and then again rendering to go back to CinGG. There is imagelist.sh (https://cinelerra-gg.org/download/CinelerraGG_Manual/Image_Sequence_Creatio…) but I could not get it to work. Other solutions found on the net also do not work for me. Sorry, my inability... Natron (Write node) overwrites the imported images, keeping their exact name (which is the frame number) and path. These are read by CinGG: until a new background render starts, on the Compositor window we see the frames altered in Natron. However, this sequence of images applies only to the timeline view: if we add filters or do a final render of the project, the original source is used and not the frames altered in Natron.

2 1

A resource for Audio-Visual Preservation using FFmpeg
by Terje J. Hanssen 17 Feb '23

17 Feb '23

I came across what seems to be a great web resource for Audio-Visual Preservation using FFmpeg cross platform. This resource is released under a CC BY 4.0 (Creative Commons) License and provided “as is” without warranty or support of any kind. Most of the content is in English, a smaller part in German only https://avpres.net/ Tutorials (under work) * Bash Cookbook for Archivists * FFmpeg Cookbook for Archivists Especially the latter "FFmpeg Cookbook" seems for me to be a great tutorial and reference using FFmpeg code syntax. An example: Filter image - Pad 4:3 into 16:9 https://avpres.net/FFmpeg/4-3_16-9

3 2

Workflow issues for multitrack editing
by Stefan de Konink 17 Feb '23

17 Feb '23

When I am creating content I am typically using 2 or more camera's, each with indepedent recordings. Much of my time goes into the alignment of such recordings. Within Cinelerra single files have some scratch area using the viewer to set in and out points. Still the viewer does not show an audio waveform while editing. And certainly does not allows the alignment of two or multiple tracks. When I am editing I would love to have something like a *multitrack* scratch space. The only way I am aware of that can achieve this is creating a second Cinerella instance working on its own EDL. This would allow me to edit some 'sub project' that can later be imported again. When *reviewing* content, it is not obvious at all which files should be part of these subprojects in the first place. Sure there might be a specific section (or take) that would be on camera A and B, but camera C might have a continuous recording not falling in the category of take. I wonder how people approach these subprojects with respect to workflow and quality. Once you have set up a take, would you bring that take to the main project? Would you first render it? What are best practises for this? -- Stefan

5 25

Working on adding GPL license attribution for all program files
by Phyllis Smith 15 Feb '23

15 Feb '23

As Andrew pointed out: "GPL license states each file should have attribution ...." so I checked into GIT the addition of missing GPL information for the *guicast subdirectory *and am working on the cinelerra subdirectory and eventually plugins subdirectory. Some notes: 1) When header missing, first checked to see if file exists in Cinelerra HV-8 or CV. If files existed there but no header, used header in like files. For example, bccmodels.inc in HV-8 had no header but bccmodels.C did, so used the same one. 2) Files, like bootstrap.c, that exist in HV and have no header, I left unchanged because I have no idea where this file came from and it has an extension of lower case "c" instead of upper case "C". 3) Any file in Cinelerra-HV that did not have the Copyright/Name/Year, I did not change in CinGG, so that it would match. If Andrew has time to verify and thinks something should have been done differently, please let me know.

4 18

ms2130 hdmi-usb dongles
by Andrew Randrianasulu 13 Feb '23

13 Feb '23

https://www.cnx-software.com/2022/11/07/ms2130-based-4k-hdmi-to-usb-3-0-vid… "I learned about the video capture dongle via Mateusz Starzak <https://twitter.com/MateuszStarzak/status/1589595781971181568> who purchased a sample on Aliexpress <https://s.click.aliexpress.com/e/_Dd6uEUB> and confirmed support for uncompressed YUV @ 1080p60. He also mentioned the latency to be around 6 to 11 frames depending on the resolution with 720p input to 720p output only looking slightly faster than 1080p input to 1080p output." well, I hope Terje's camera can be switched into hdmi progressive output mode ..... contain few links. Of course main problem is verification ..... if they just paint them (usb connectors) blue for no reason ...:/

2 3

adv7181d (video decoder chip)
by Andrew Randrianasulu 12 Feb '23

12 Feb '23

May be I read it wrong but table 10 suggest it can even output 20 bit (!!!) yuv for no reason ?? (input ADC only 10 bit) Note, very early version without any letter at the end seems to say "9 bit ADC". Also, some versions may support "16 bit port" or "8 bit port" depending on exact packaging! https://web.archive.org/web/20221006213255/https://www.analog.com/media/en/… Four 10-bit ADCs sampling up to 75 MHz 10 analog input channels SCART fast blank support Internal antialiasing filters GENERAL DESCRIPTION The ADV7181D is a high quality, single-chip, multiformat video decoder and graphics digitizer. This multiformat decoder supports the conversion of PAL, NTSC, and SECAM standards in the form of composite or S-Video into a digital ITU-R BT.656 format. At least 10 bit yuv 4:2:2 seems to be real at chip level "10-Bit SDR ITU-R BT.656" I have no idea if there even way to store 20bit per component yuv .... But board for it will be definitely not beginner's project: https://electronics.stackexchange.com/questions/311491/adv7180-breakout-boa… slightly different board but high-speed challenges outlined ... Apparently some of those chips (same family) were used in some Blackmagick devices but not necessary in best for vhs config? https://www.digitalfaq.com/forum/video-capture/4356-dazzle-dvc107-video.html ==== And I just received a reply from the Australian BlackMagic Design distributor, confirming that their Intensity Pro card uses the ADV7180. Though they tell me it doesn't handle timebase errors well, and they recommend an external TBC! That's disappointing! Sounds like it's been configured for very stable professional equipment. ------------------------------ * Last edited by AusDaz; 07-15-2012 at 09:04 PM. Reason: Updated info * *====* *note, 10 years later we have some software to try ... I successed in compiling vhs-decode branch of ld-decode, at least mains script do not err out on my Slackware, so ..... :-)*

1 1

Analog S-Video or HDV capture via HDMI-USB to FFV1.mkv in realtime
by Terje J. Hanssen 12 Feb '23

12 Feb '23

I have no plan to compete with our National Library located locally here :) https://www-nrk-no.translate.goog/nordland/her-skal-den-norske-kulturarven-… but I prepare myself to test and use one method left with regards to "Personal Video Archiving". Therefore I have ordered two selected, inexpensive devices (the urls here are just for English language): 1) Capture Card USB HDMI 4K to 1080P USB-C adapter (Speedy USB 3.2/3.1/3.0, 60 Hz and 1080p FHD.) https://www.turascandinavia.com/en/products/computer/microphone-webcam/webc… * Fluid video recordings without delays or distortions during live streaming * Linux not mentioned, but assumed this is a standard supported UVC. OBS and VLC is mentioned. 2) Scart AV SV to HDMI Converter Switcher S Video Composite RCA DVD HD TV adapter (3-4 weeks delivery time https://www.ebay.com.au/itm/295472602625 * The 3-in-1 S-/video/RCA/SCART to HDMI converter is a universal converter for analog video input to HDMI output. * Analog to digital converter (ADC), 10 bits maximum 1,62 MSPS sampling, Q I have not yet found a definitiv ffmpeg FFV1 v 3 syntax guide for SD and HDV capturing and hope for suggestions and comments here!? 3.1 FFmpeg recipe (PAL) ffmpeg -i VIDEO_IN \ -c:v ffv1 -level 3 -coder 1 -context 0 -slices 24 -slicecrc 1 \ -color_primaries bt470bg \ -color_trc bt709 \ -colorspace bt470bg \ -color_range mpeg \ -map 0 \ -top 1 \ -c:a copy \ -g 1 -pix_fmt + \ VIDEO_OUT.mkv 3.2 Reference Example https://trac.ffmpeg.org/wiki/Encode/FFV1 Copy audio "as-is" and use FFV1.3 as video codec. Parameters are 8 threads, coder=1, context=1, GOP-size=1, 24 slices and slice-CRC on: ffmpeg -i <input_video> \ -acodec copy \ -vcodec ffv1 -level 3 \ -threads 8 \ -coder 1 \ -context 1 \ -g 1 \ -slices 24 \ -slicecrc 1 \ <output_video> 3.3 Various Encoding examples for Video capture from USB camera /dev/video0: ffmpeg -f v4l2 -framerate 25 -video_size 720x576 -i /dev/video0 output_data.mkv ffmpeg -f v4l2 -framerate 25 -video_size 720x576 -i /dev/video0 -codec:v ffv1 -codec:a pcm_s16le \ -f matroska output_video.mkv Check that video is captured properly using ffplay: ffmpeg -ar 44100 -thread_queue_size 1024 -f alsa -i $AUDIO \ -r 25 -thread_queue_size 1024 -i $VIDEO \ -codec copy -f matroska - | ffplay - Check the overall flow using ffplay before capturing: ffmpeg -ar 44100 -thread_queue_size 1024 -f alsa -i $AUDIO \ -itsoffset -0.5 -r 25 -thread_queue_size 1024 -i $VIDEO \ -filter_complex 'channelmap=FL-0' -filter_complex 'crop=w=700:h=556:x=8:y=0' \ -map 1:v -map 0:a -codec:v ffv1 -codec:a pcm_s16le \ -f matroska - | ffplay - 4, Regarding FFV1 lossless compression in realtime Comparison tests (2013/2015) have shown that FFV1 produced the smallest files at the fastest speed. It can be used to capture SD material in realtime, and HD possibly, too. This according to a 2012/2015 article, where I have extracted the results for FFV1 v.3 in the tables below: http://www.av-rd.com/knowhow/video/comparison_video_codecs_containers.html#… NOTE: This is a consumer grade, off-the-shelf PC setup. Hardware: CPU: Intel(R) QuadCore(TM) i7-2600K CPU @ 3.40GHz RAM: 8 GB Disk: Intel SSDSA2CW080G3 (SSD) Software: Operating System: GNU/Linux (Xubuntu 12.04.1, 64bit) Transcoding tool: FFmpeg (version git N-59183-g3e62654, Dec 17 2013) Video source file: VQEG reference video "football" <http://media.xiph.org/video/derf/y4m/football_422_ntsc.y4m> (NTSC-SD, 720x486px, 30fps, yuv422p, 8bpc) Codec Encoding Decoding Filesize % of uncompressed Implementation Details FFV1 <http://www.av-rd.com/knowhow/video/comparison_video_codecs_containers.html#…> (version 3) 216 fps 277 fps 111 MiB 46.1% libavcodec (FFmpeg) <http://en.wikipedia.org/wiki/Libavcodec> log <http://www.av-rd.com/knowhow/video/test_results/football-720x486i30/ffv1.3-…>, framemd5 <http://www.av-rd.com/knowhow/video/test_results/football-720x486i30/ffv1.3-…> Video source file: SVT reference video "park joy" <http://media.xiph.org/video/derf/y4m/park_joy_1080p50.y4m> (full-HD/1080p, 1920x1080px, 50fps, yuv420p, 8bpc) Codec Encoding Decoding Filesize % of uncompressed Implementation Details FFV1 <http://www.av-rd.com/knowhow/video/comparison_video_codecs_containers.html#…> (version 3) 31 fps 63 fps 879 MiB 57.2% libavcodec (FFmpeg) <http://en.wikipedia.org/wiki/Libavcodec> log <http://www.av-rd.com/knowhow/video/test_results/park_joy-1920x1080p50/ffv1.…>, framemd5 <http://www.av-rd.com/knowhow/video/test_results/park_joy-1920x1080p50/ffv1.…> PAL-SD 720x576 stored as YUV422, 10 bpc (bits per component), instead of the tested 8 bpc above, seems to me would have enlarged the the filesizes with about 25%, or to about 40 GB/hr with FFV1 lossless compressed. Although the FFV1 compression results above obviously used yuv422p, 8bps source SD and FHD files as input, it seems for me that the encoding speeds at 216 fps and 31 fps for SD and FHD respectively, assumingly should manage realtime FFV1 encoding and streaming of PAL SD and HDV at 25 fps as well, especially with a faster i7-6700 x 8 cpu. Comments here? 5. Related 2021-[Cin] threads as background references [Cin] hdmi capture card (with v4l2) https://www.mail-archive.com/[email protected]/msg02865.html [Cin] offtopic: Digitize and record Video with A/D USB adapters https://www.mail-archive.com/[email protected]/msg03269.html [Cin] offtopic: HDMI capture https://www.mail-archive.com/[email protected]/msg02084.html [Cin] Is SD and HD 422 Video Capture obtainable with V4L(2)? https://www.mail-archive.com/[email protected]/msg03719.html [Cin] Uncompressed video to lossless compression FFV1/MKV https://www.mail-archive.com/[email protected]/msg03791.html

2 21

Usb 3.0 capture card (costly?)
by Andrew Randrianasulu 12 Feb '23

12 Feb '23

On vhs-decode Discord server I was pointed at this device: https://www.magewell.com/tech-specs/usb-capture-aio seems to list linux v4l2/alsa (not sure how 10bit stuff represented)

1 1