- Cin - Cinelerra-GG Mailing Lists

Debian 32-bit on Qemu
by Andrea paz 12 Oct '21

12 Oct '21

I have installed a Debian 11.1 32-bit on qemu. What packages do I need in order to compile CinGG from git? I remember there used to be a dependency checking script, but I, with Arch, have never used it. Or is it better to use some other installation method?

4 12

[WIP] connecting pluginprefs.C back to codebase
by Andrew Randrianasulu 11 Oct '21

11 Oct '21

Basically just monkeying around (it does not do anything with local plugin set) but I can try and push this a bit more so it will actually search plugins in second location I think main idea was to alliw 'favorities' set of plugins somewhere in user dir, if user can't change stuff in /usr. Today Cingg can look in dufferent places for plugins depending on env. variables, still gui might be handy for some users. Also, it show how to add new category in oreferences, so we can offload few things into new tab in the future (or just add them there)

2 2

libaom 3.1.1 patch for Ubuntu 16.04.7
by Andrew Randrianasulu 10 Oct '21

10 Oct '21

It seems Cmake version bump in libaom 3.1.x was mostly cosmetic - it builds with attached patch (put in thirdparty/src) right now this new (CinelerraGG/ffmpeg-4.4/libaom-3.1.1) build completes - so not tested yet.

4 5

DPX sequence in CinGG
by Andrea paz 09 Oct '21

09 Oct '21

I tried to make an ffmpeg profile of the DPX standard movie format. So now CinGG has DPX support inside ffmpeg and OpenEXR outside of ffmpeg. Remember that these two formats produce a sequence of still images and their relative TOC (it is only the TOC that needs to be imported into CinGG when we want to load this format). So it's good to create a folder where to put all the images and the TOC produced, which are hundreds and even thousands. Big space occupation but super efficient in timeline and great quality. I attach the profile if you want to test it.

3 6

offtopic: Digitize and record Audio via Line-In Jack or via USB adapter?
by mat 06 Oct '21

06 Oct '21

=== Terje:===== 1) My Behringer UFO202 USB Audio interface has arrived. As a first and easy test I have connected it to a PC USB port, the RCA IN L+R to RCA OUT on a Toshiba videoplayer playing an Audio music CD, and third connected earphones to UFO202. 2) Yes, I hear the playback sound in the earphones, however no sound if I plugin the earphones to the normal rear PC backplane's Jack, and no sound in the corresponding speakers. Correct so far? =========== That is normal. By default, incoming (line) signals are not rerouted to outputs, some software must do that. === Terje:===== 3) pavucontrol Configuration | *PCM2902 Audio Codec* Profile: Analog Stereo Duplex * Input Devices | Hardware devices | Hardware Input Devices: PCM2902 Audio Codec Analog Stereo, Port: Analog Input response Built-in Audio Analog stereo, Port: Rear Microphone (plugged in) Respons!? ================ Looks good. Here I have for the non-PCM2902 device (HD audio controller analog stereo) port Line 1, and plugged in. === Terje:===== 4) UFO202 is discovered on my Leap 15.3 as *Device 4 **Texas Instruments PCM2902 Audio Codec* as follows: ... ================ Looks good, same here, except here the video HDMI is the first device and the motherboard HD audio the second. PCM2902 is third also. === Terje:===== pacmd list-sources | egrep '^\s+name: .*alsa_input' ================ Output looks fine. === Terje:===== 5) I am unsure how to practically go from here to record the Audio music files. ================ 8) Which tab/buttons for input in Audiacity (possibly tape/noise filters) - I could not find the PCM2902 input or see any sound response here I suggest to try Audacity first. In Edit->Preferences->Devices, Host Alsa, playback device pulse, recording default (or pulse), 1 or 2 channels as desired. Edit->Preferences->Recording, record on new track (each time you start a recoding you get a new audio track). Then, on the main menu, click on the filed that says "click to start monitoring", this will send input to output, so if your output device is line out it will come out there. For recording, the red button starts recording, the black square button stops. The advantage of using Pulse is that pavucontrol will show recoding applications, you should also see the signal coming in when an application is recording (has opened the input channel). This can be seen on the Input Devices tab. You don't see the device as such, it is "abstracted" by both Also and then PulseAudio. === Terje:===== 9) Would it be reasonable to record and archive in lossless FLAC format and afterwards convert with ffmpeg? to smaller files format of good quality, typical MP3 at 320 kbps (for car usb player) and MP4A for home HiFi? ================ FLAC is very good, it itself is lossless. Audacity can export to a variety of formats (File->Export), I usually use mp3 or wav. Note that if you do a lot of editing on a recording is it better to save the project, that avoid converting back-and-forth between flac and the interanl Audacity 32 bit floats. Audicity is a good editor, you record first, save the original recoding, then apply filters/effects/cuts etc . Like CinGG, keep the original intact. Offhand, I'd say you must have extremely good hearing to hear differences between MP3 320kbps and the original. For cars, I myself would not go above 128 kbps. MP3 at 320 kbps is much better than most home HiFi. And if the source is not digital, (but tape or cassette), 128 kpbs MP3 is already much better than the source material. ========= In CinGG, you could try Alsa, CinGG shows the device directly. Preferences->recording tab. Record driver Also, then you get in the Devoce dropdown box multiple devices, look for something like USB Audio. If you have webcam also connected, you are likely to see two of them. I did have some problems there, my settings are Record driver "Pulse Audio" , leave the device at default. Then pavucontrol determines which devices. I am not sure if the sample rate is set corretly for the device by CinGG, if it does not work or you get lots of errors, try 44100 instead of 4800, the UFO202 supports both. Good luck, Mat

2 3

Re: [Cin] offtopic: Digitize and record Audio via Line-In Jack or via USB adapter?
by mat 04 Oct '21

04 Oct '21

Your MSI Z170-A PRO motherboard has according to specs a Realtec AL892 audio chip. The input converters should have 90 dB S/N (signal-to- noise) ratio (A-weighted), but of course the motherboard might make it a little worse. However, from you record player of tape you will be very lucky to get better than 65 dB S/N. So if an A-D converter has a S/N of 70 dB or better, it is more than good enough for this. Regarding your PC input channels, on current desktop motherboards I usually see three 3.5 mm mini-jacks. Their recording quality for line- in is close to 90 dB S/N or even above 100 on the better ones. In pavucontrol (audio mixer), first check the configuration tab, to see if the Built-in Audio is set up correctly. "Analog stereo duplex" is probably right. Then, on the "Input devices" tab, at the bottom set it to show "Hardware Input Devices", then check that the Port is set to the correct port. For me, "Line In" shows correctly if something is plugged in or not. Make also sure the gain is set to 100%. That said, I sometimes have problems recording LineIn with Audacity, even though all seems OK. have not yet determined what the cause is. External USB sound cards can be quite good, if you stay away from the very cheap ones that are mic/phone only. I have a Behringer UCA-202 which is good enough for this, but there are much better ones (at a price), look at music-oriented webshops. I presume to have an RIAA pre- amp for you record player, these are usually set up for less output that typical PC or usb-audio input, which expect 1 Volt. Drivers are not needed, and disregard any reviews that don't show measurements. MatN

3 7

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