- Cin - Cinelerra-GG Mailing Lists

Crash when using OpenEXR image sequence
by Andrea paz 02 Oct '21

02 Oct '21

>From this forum post: https://www.cinelerra-gg.org/forum/everything-else/youtubes-video-edit-with… I was inspired to try a workflow with a sequence of images. I took a 32 sec video (in AV1) and rendered it as an OpenEXR image sequence. The 32 sec video was rendered in 5 seconds resulting in 881 images and a TOC file (test-exr.exrs). I created a new project in CinGG and imported the files + TOC but every time I have a crash in the middle of the images (about 480 images). >From the terminal you can see that all 881 images are indexed, but then in a second step, it crashes towards the 480th image. The error is as follows: FFMPEG::open_decoder: some stream have bad times: /home/paz/test-exr/test-exr000001.exr I attach the terminal lines file and also test-exr.exrs. I used both "Replace current project and concatenate tracks" and "Create new resources only". In the first case the images are loaded in the timeline but the crash happens anyway. I used both AppImage and Build. Can you identify/reproduce/resolve the problem? NOTE: In the terminal output I removed many repetitive lines. The file "test-exr.exrs" is the TOC file. I have the same crash both trying to import it into CinGG along with the images and excluding this file. NOTE 2: same crash with PNG sequence!

2 2

Re: [Cin] even more OT: time-base correctors
by Terje J. Hanssen 01 Oct '21

01 Oct '21

Den 28.09.2021 18:36, skrev Andrew Randrianasulu via Cin: > I read elsewhere [0] you might want such device for bad quality vhs > tapes - > yet they tend to expensive, either in terms of money or skills. > > Still, it was pleasure to find some schematic I thought completely lost to > sands of time: > > https://web.archive.org/web/20181219122351/http://marcusgun.awardspace.com/ > newvidproc.html > > https://web.archive.org/web/20051101234823/http://www.astro.uu.se/~marcus/p… > > [0] > https://forum.videohelp.com/threads/402960-VHS-recording-advices-would-be-a… > > reviews on Amazon tend to be mixed - complex and costly devices do not > works for some, yet work for others (due to condition of device itself?) > > "Short Overview: A time-base corrector is essentially a device that > reconstructs a composite video signal scan line by scan line. Composite > video signals (both NTSC and PAL) interlace two fields of lines to > construct a single frame. With NTSC, you are viewing (approximately) 60 > fields per second, while PAL uses 50 fields per second. Each line has a > pulse/signal that tells the display device to start the next line. If the > timing is off even slightly for a single field or for a set of scan lines, > the individual scan lines don't align with each other and you get a jagged > effect. A time base corrector tries to correct the timing for the > individual scan lines to improve the picture. This process never works > perfectly, but you can definitely see an improvement on most video > material. > > A full-frame TBC actually stores the lines to build up a single frame and > then sends them to the video display or monitor. This is preferable to a > unit that just tries to adjust each line as it is processed. An advantage > to the full-frame units is that they normally eliminate the copy > protection > signals that are often placed in the unseen scan lines between fields and > frames. They do this because they insert their own synch signals, > discarding the "corrupted" ones that Macrovision and other copy protection > schemes have deliberately inserted. (Yes, I know this is a simplified > description of a complex process.) > > The 8710 is one of a series of TBC units that are reasonably priced and do > a pretty good job of correction. The improvement is most noticeable when > you have vertical lines in the picture (telephone poles, lettering, etc.) > The other features on this unit (saturation control, color correction, and > so forth) are secondary. It is a full-frame unit and it builds each frame > using two fields. This allows better alignment of the scan lines and a > better picture." I want to confirm from my own experience the importance of using a good TBC at the End of the analog workflow from tape source playback, just in front of A/D converting and recording. Especial early generation Hi8 recordings, in the first nineties with Sony CCD-TR805 & TR808 camcorders, got unacceptable images with jitter, flagging and even dropouts. If this was due to unstable tape operation with slippery Hi8 Pro tapes, I don't know. The footages became fine and stable with upgrade/replacement to Sony TR-2000 camcorder, still functional as Hi8 VCR player). * A simple Line based TBC for normal playback were built-in in good, reference VCR decks, like my Panasonic FS-200EC S-VHS VCR and Sony EV-S880E Hi8 VCR. * A full frame or Dual Field TBC was built-in in my Videonics MX-1 digital videomixer, which resulted in wonderful clear and steady analog video images. As Videonics also published a Digital Video Primer that explained the Time Base problems in an easy and useful way, I have scanned and attached these 4 pages to this post. And for further references and reading this old stuff, follow some related urls which are still active: What is a TBC? Time Base Correction for Videotapes www.digitalfaq.com/forum/video-restore/2251-tbc-time-base.html What is a TBC* and why would I need one? https://www.questronix.com.au/tbc.htm Videomaker https://www.videomaker.com/article/f3/641-a-tbc-tale https://www.videomaker.com/article/1210-edit-suite-tbc-trivia Videonics MX-1 videomixer https://www.southernadvantage.com/products/videonics-mx-1-4-input-synchroni… https://www.amazon.com/Videonics-MX-1-Video-Mixer-MX1/dp/B000ZS88N8 Terje J. H

1 0

New AppImage release for Sept. 30 now available
by Phyllis Smith 30 Sep '21

30 Sep '21

The latest releases are at: 1=newer kernel O/S 2=older kernel O/S; even for newly installed O/S, this may be needed because of older kernel 3=32bit system of older variety 4=newer kernel O/S that handles 10 and 12 bit as well as 8 bit 1. https://cinelerra-gg.org/download/images/CinGG-20210930-x86_64.AppImage 2. https://cinelerra-gg.org/download/images/CinGG-20210930-x86_64-older_distro… 3. https://cinelerra-gg.org/download/images/CinGG-20210930-i386.AppImage 4. https://cinelerra-gg.org/download/images/CinGG-20210930-multibit.AppImage Both PDF and HTML versions of the Manual have been updated on the website as well as the releasenotes. *********************************** Copy of latest ReleaseNotes items The *LV2 plugins* related source code has been improved by *MatN*, making it more robust and better able to prevent crashes at startup for new Operating Systems and LV2 plugins. In addition he has updated the blacklist for the plugins. Problem was reported by Fary54 when he upgraded Debian and new Ardour plugins became available – he assisted in reporting and testing the *BT #586* issue. - Note bootstrap.c System Command was increased from 1024 to 2100 to ameliorate some warnings. - The LV2 plugin path is now being printed to the CinGG startup window; same as other plugins. A *libaom patch* was provided by *AndrewR* for allowing libaom to be updated on Ubuntu 16 due to cmake version. The Motion51 plugin with Dissolve transition problem has been fixed by a Freelancer. *BT #564* was reported and documented by Skinkie. New Icons for the 9 missing ones have been added for viewing in the Resources window. These are alpha, boxblur, colorspace, foreground, posterize, sketcher, timeblur, timelapsehelper, and tracer.

1 0

Color erase like GIMP
by Stefan de Konink 30 Sep '21

30 Sep '21

Background: I have created an animation with animejs, using some SVG images. Not knowing how I could render those I have done a screen capture using ffmpeg on a green background. Suggestions on which renderer would be capable in rendering on top an transparent surface would be interesting. What I want: What I would like to achieve is to use the green channel as alpha mask, but it should also do partial transparency while maintaining the input color. Within GIMP I am able to apply color-erase to replace the green background to transparency, ImageMagick is not able to do it. Cinelerra and ffmpeg also seem to mask, but not subtract. An alternative script is capable of creating an alpha for green, but not keeping the original input white, like GIMP. In ./app/operations/layer-modes/gimpoperationlayermode-blend.c the function gimp_operation_layer_mode_blend_color_erase describes the inner working. Could anyone suggest how to achieve this within Cinelerra or first with an external tool? -- Stefan

4 11

even more OT: time-base correctors
by Andrew Randrianasulu 28 Sep '21

28 Sep '21

I read elsewhere [0] you might want such device for bad quality vhs tapes - yet they tend to expensive, either in terms of money or skills. Still, it was pleasure to find some schematic I thought completely lost to sands of time: https://web.archive.org/web/20181219122351/http://marcusgun.awardspace.com/ newvidproc.html https://web.archive.org/web/20051101234823/http://www.astro.uu.se/~marcus/p… [0] https://forum.videohelp.com/threads/402960-VHS-recording-advices-would-be-a… reviews on Amazon tend to be mixed - complex and costly devices do not works for some, yet work for others (due to condition of device itself?) "Short Overview: A time-base corrector is essentially a device that reconstructs a composite video signal scan line by scan line. Composite video signals (both NTSC and PAL) interlace two fields of lines to construct a single frame. With NTSC, you are viewing (approximately) 60 fields per second, while PAL uses 50 fields per second. Each line has a pulse/signal that tells the display device to start the next line. If the timing is off even slightly for a single field or for a set of scan lines, the individual scan lines don't align with each other and you get a jagged effect. A time base corrector tries to correct the timing for the individual scan lines to improve the picture. This process never works perfectly, but you can definitely see an improvement on most video material. A full-frame TBC actually stores the lines to build up a single frame and then sends them to the video display or monitor. This is preferable to a unit that just tries to adjust each line as it is processed. An advantage to the full-frame units is that they normally eliminate the copy protection signals that are often placed in the unseen scan lines between fields and frames. They do this because they insert their own synch signals, discarding the "corrupted" ones that Macrovision and other copy protection schemes have deliberately inserted. (Yes, I know this is a simplified description of a complex process.) The 8710 is one of a series of TBC units that are reasonably priced and do a pretty good job of correction. The improvement is most noticeable when you have vertical lines in the picture (telephone poles, lettering, etc.) The other features on this unit (saturation control, color correction, and so forth) are secondary. It is a full-frame unit and it builds each frame using two fields. This allows better alignment of the scan lines and a better picture." https://www.amazon.com/AVT-AVT-8710-Time-Base-Corrector/product-reviews/B00…

1 0

offtopic: Digitize and record Audio via Line-In Jack or via USB adapter?
by Terje J. Hanssen 23 Sep '21

23 Sep '21

My background info: I have not only an archive of analog tape video records, but also a bunch of 30 years archived analog music audio to digitize. To mime, my audio records mainly started when I bought my first reel tape recorder, Tandberg 1200X about 1970 (my first own earned money), in parallell with some vinyl LPs. https://www.audio-high-store.com/tandberg-1200x/ Later on followed also the popular music compact casette deck period for home and in cars. So far I have just ripped audio music tracs from audio CD and DVD discs to mp3 files on my Linux workstation, using K3b or other suited applications. Indeed, I have just started to look at how to digitize and record analog audio from external connected tape and vinyl players. Analog PC connections Analog AUX ports (3,5 mm Jacks or RCA phono connections) labeled "Line IN" (usual on previous internal sound cards), are have more or less left out on modern workstations and laptops with digital USB ports. However, my most relevant workstation with a MSI Z170A Pro maiboard includes an AUX "Line IN" jack (Light Blue color code) on its backpanel, see attached image if allowed. I am in doubt and have not yet been able to verify or test with analog players if this (Blank: empty?) Jack Line-IN works at all. I think I read on a forum another user had issues with similar. While Audacity didn't find it, possibly the Pulse Audio Volum control (pavucontrol) could enable the Jack Line-IN. If Jack-IN on the mobo back panel won't work, the other option is to find and buy an analog to USB audio interface adapter that works well with Linux. Meanwhile, possibly experience and suggestion to test procedure for this will be welcome ?! ----------------- Some related output from my workstation: inxi -Fxz System: Kernel: 5.3.18-59.19-default x86_64 bits: 64 compiler: gcc v: 7.5.0 Desktop: Gnome 3.34.5 Distro: openSUSE Leap 15.3 Machine: Type: Desktop System: MSI product: MS-7971 v: 1.0 serial: N/A Mobo: MSI model: Z170-A PRO (MS-7971) v: 1.0 serial: <filter> UEFI: American Megatrends v: 1.K0 date: 07/10/2018 Audio: Device-1: Intel 100 Series/C230 Series Family HD Audio vendor: Micro-Star MSI driver: snd_hda_intel v: kernel bus ID: 00:1f.3 Device-2: NVIDIA GK208 HDMI/DP Audio vendor: ASUSTeK driver: snd_hda_intel v: kernel bus ID: 01:00.1 Device-3: Microdia Camera type: USB driver: snd-usb-audio,uvcvideo bus ID: 1-11:3 Sound Server: ALSA v: k5.3.18-59.19-default Drives: Local Storage: total: 5.19 TiB used: 2.87 TiB (55.4%) ID-1: /dev/nvme0n1 vendor: Samsung model: MZVPV512HDGL-00000 size: 476.94 GiB ID-2: /dev/sda vendor: Intel model: SSDSA2CW160G3 size: 149.05 GiB ID-3: /dev/sdb vendor: Seagate model: ST3500418AS size: 465.76 GiB ID-4: /dev/sdc vendor: Seagate model: ST3500418AS size: 465.76 GiB ID-5: /dev/sdd type: USB vendor: Seagate model: Expansion size: 3.64 TiB ID-6: /dev/sdj type: USB vendor: SanDisk model: Ultra size: 28.97 GiB Selected references: https://www.linux-magazine.com/Issues/2018/206/Digitizing-Analog-Music-Media https://ro-che.info/articles/2017-07-21-record-audio-linux https://kimbriggs.com/computer/record-digitize-cassette-ubuntu-linux https://opensource.com/life/14/10/how-clean-digital-recordings-using-audaci… Thanks, Terje

2 1

Question about the power of cinelerra-gg
by Zeke Williams 09 Sep '21

09 Sep '21

Is it true that cinelerra-gg is more powerful than kdenlive or openshot? The fact this program advertises as having hardware acceleration caught my attention. Something other libre video editors have no such feature.

3 3

I wonder if old bug 967 still with us?
by Andrew Randrianasulu 03 Sep '21

03 Sep '21

https://web.archive.org/web/20140213170701/http://bugs.cinelerra.org/ticket… Cinelerra version indicated in bugreport was 2.1.5, but there was no resolution ('resolved/fixed' ), so I wonder if this bug still lurking somewhere in CinGG and messing up some timelines... ====copypasta== I create an imagelist and load it on the timeline. Cinelerra plays it back correctly till the end where the bug appears: after the last frame the last-but-one frame is added and played back. When playing back frame by frame a blank frame is added and played back. The bug depends on the framerate of the imagelist. I attach a folder with images, lists and an XML project where lists with 5, 8, 10, 12, 15, 20, 25 and 30 fps are tested: 5 -> Correct normal playback but 2 blank frames when played back frame by frame. 8 -> Bug! 10 -> Bug! 12 -> Bug! 15 -> OK 20 -> OK 25 -> OK 30 -> OK NB: 12 fps is the traditional framerate for animation. I first noticed this bug in 2008, when editing my first quick pace animation. WORKAROUND: Make your image sequence longer by copying 2-3 times the last image in your image folder, giving appropriate names (keep the numering of the sequence). Add them to your image list file in a text editor or recreate the imagelist file. Reopen your XML project in Cinelerra. The playback should now be OK. Change History Changed 2 years ago by raffa cc raffaella.traniello@… added Just preparing the attachment now I noticed the bug is present only with 720x576 images. Changing the images resolution to 640x480 or 375x300 makes the imagelists to be played back correctly. NOTE: 720x576 is the standard animation resolution when capturing from MiniDV cameras via firewire framegrabbers. The attachment is too big to be attached here (742.5 KB). You can download it at http://g-raffa.eu/TestImageList.tar.gz It contains a Cinelerra project with lists in different resolution and framerates. === i saved tgz archive into archive.org, so it should be downloadable by directly clicking on link from archived page...

2 3

There is a new Plugin in town - Speed PerCent
by Phyllis Smith 02 Sep '21

02 Sep '21

A contributor has made a new Video plugin available and it is checked into GIT as Speed PerCent. Very nice. It is self-explanatory. See attachment.

4 10

small error in download area?
by Andrew Randrianasulu 31 Aug '21

31 Aug '21

i tried to follow (click on) all links in https://cinelerra-gg.org/download/ and very first one (Arch) just takes me to 'page not found' /news section.. https://www.cinelerra-gg.org/arch-package/ other links correctly display their distro-specific pages with notice about switching to appimage..

2 1