Hello,
I'm a beginner with Cinelerra. I saw the tutorial "Cinelerra-GG： Multi-cam with audio sync (the easy way)" and understand it.
I have splitted files from one camera (Fat size limits) and from an DSLR and an audio record started before and stopped after the cameras.
For the DSLR, there is unrecorded time between files due to record time limit.
I followed the tutorial and import all my file in "media", insert my audio in the first track, chose my first file from each camera and right-clic to get "open mixers". I sync my two video with the audio without difficulties.
Now, i'm stuck.
1) I want to check with my headphone if my audio tracks are really synchronized. Is there a fast way to hear two audio track at the same time (mixing similar to a "PFL" on mixer or one per side) ? Found it, buttons "play the track" under the tracks name.
2) What is the way to insert others video files on the same "mixer" in order to keep one video track per camera ?
3) If possible, can i use "align mixer" ? When i tried without mixer mode, the name proposed was only the first file, not the ones selected.
4) Is there a simple way to align video too by adding an offset of two or three frame (sound speed vs light speed)
Thanks.
Please, @Kabak, look at the "Multi Camera Editing" to https://www.youtube.com/watch?v=49EExsRsjlY
There are things there that could be useful to understand How Mixers View (Multicamera) works, I think.
Thanks for the response but this tutorial seems to be for next step : the edition.
I'm still preparing resources before edition.
CinGG can only align and synchronize the tracks during editing and not before. I don't quite understand requirements 2) and 3), but I think by linking the various tracks manually to the mixer you can get what you want. Try to see in the manual if you can find answers to your questions:
https://cinelerra-gg.org/download/CinelerraGG_Manual/Multi_Camera_Mixer.html
and:
https://cinelerra-gg.org/download/CinelerraGG_Manual/Audio_Video_sync_using_Wave.html
For answer 4) perhaps the “nudge,” found in the patchbay, may be useful to you:
https://cinelerra-gg.org/download/CinelerraGG_Manual/Patchbay.html
Maybe i'm not using the rights terms, not doing it the right way and missing something.
For point "4)", i tried and it seems to works like expected. Thanks.
For 2) and 3) :
I have multiple file from two camera (17 in my example) :
- camera 1 with angle 1 : MVI001, MVI002, MVI003 ... MVI010
- camera 2 with angle 2 : MOV027, MOV028 ..
If i use the "insert mixer" function, i will have :
- Mixer 1 for MVI001
- Mixer 2 for MVI002
...
- Mixer 11 for MOV027
...
- Mixer 17 for MOV033
All tutorial i have seen show one file per mixer track.
I'm not sure having 17 mixers windows is the best way to work. I must applying 10 times the same effect to one camera ( white balance, denoiser, audio volume ...), making the yo-yo between tracks during mixing to the final video track...
@kabak No, it is not you who is using the wrong terms, it is me who is not very familiar with the English language. You are right: using many clips in multicam is not very practical. Being able to use a mixer window for multiple clips is not possible in CinGG because of its fundamental operation based on tracks in their entirety instead of clips. Mixer windows are based on the sources and therefore are 1:1 and in addition multicam takes into account the tracks, not the clips we put on the timeline.
However, one thing is not clear to me: do you do editing and apply effects before doing the multicam? Isn't it better to do them on the final result? The original clips are made disarmed by default precisely because it is not recommended to act on them, but only on the first track which is the result of multicam editing.
@andreapaz My native language is french. I'm not very good too with english too.
do you do editing and apply effects before doing the multicam? Isn't it better to do them on the final result?
All "clip" are mostly from one continuous recording (concert, show, a play from theater, conference ...) but the camera need to split it in multiple file (2 or 4GB) due to FAT32 limitation.
Maybe it's an mistake but i want to work the same way i do on an audio mixer. On an audio mixer, for each microphone, instruments, music line ... you set the gain to normalize volume, you make the equalization for each microphone, instruments ... and you make you mixing after that.
With "camera track" containing all "clips" from this camera, i will apply effects who affect all "clip" for this specific camera.
Camera 1 is too dark. I put an brightness/contrast correction
Camera 2 is too "red". I put an color correction on all clip from this camera.
Like this, i will not have to correct the final result each time i switch between cameras. If i need to correct a specific part for "artistic reasons" for example, i will do it on the "mixed" video track, not the "mixers"
Since I do not quite understand question #4 as shown below, maybe you mean you want to insert 2 or 3 blank frames in the video so it is better lined up with the audio. If so, then make sure only the video track that you want to do this on is the only one armed (disarm all of the others in the patchbay to the left of the tracks by clicking on the 2nd icon) then go to where you want the offset and execute Shift-Space 2 or 3 times. This will add 2 or 3 blank frames. Be sure to arm your other tracks again!
4) Is there a simple way to align video too by adding an offset of two or three frame (sound speed vs light speed)
Since I do not quite understand question #4 as shown below, maybe you mean you want to insert 2 or 3 blank frames in the video so it is better lined up with the audio.
The goal is to align picture to picture and sound to sound. The sound travel slower than light so you will have to correct that.
"Insert" some blank frame can be a way to do it but offsetting the track (nudge in this software) seems to be more "elegant".
@Kabab
Yes, for me it is also the right way to work. However, not using multicam, I didn't think about the problems it brought.
There are some advanced editing features you might need; nested clips and nested assets (sub-project):
https://cinelerra-gg.org/download/CinelerraGG_Manual/Nesting_clips_assets.html
OpenEDL could also serve you to reduce the number of clips you have to work with. See the tutorial:
https://cinelerra-gg.org/download/CinelerraGG_Manual/Workflow_with_OpenEDL_Neste.html
However, I don't know if they can be adapted to multicam workflow.
@Kabak
So you have some splitted videos from camera_1 and some splitted video from
camera_2; more an audio (stereo) from microphone. It is right?
Maybe it is a stupid question, and maybe you have already done it, but...
Have you tried inserting all the clips, side by side, of the camera_1 in the Video_1 +
Audio_2 + Audio_3 tracks, and all the clips, syde by side, of the camera_2 in theVideo_2 +
Audio_4 + Audio_5 tracks, and the Master stereo audio in the Video_3_empty +
Audio_6 + Audio_7 tracks?
First you have to add the empty tracks. To insert the clips in the Timeline you can see this old tutorial "Insert the clips in the Timeline" https://www.youtube.com/watch?v=jgLAaxmCsNk
Note: for the Master stereo audio you also need to create an empty Video track (Video3_empty) along with the two audio tracks to use the Mixer Viewer.
Once you have inserted the clips you can use Expert Usage of the "MultiCamera/Mixer" feature.
https://cinelerra-gg.org/download/CinelerraGG_Manual/Multi_Camera_Mixer.html
And one ticket open about MTS/AVCHD files loading...
I tried to synchronize multiple times those contiguous file without having a good result. There's always a little gap between them
Those files seems to be "stream" and are divider into data frames (with audio parts and some video frames). It's seem some camera split the last data frames into the two contiguous file.
Cinelerra and conversion tools cant rebuild it because they doesn't make the link between the contiguous file. They see two damage files with damage data frames at the start and end of file.
I test by concatenating two files, missing frames and audio appears.
It will help me to reduce chunk from 14 chunk to 4.
Maybe it is a stupid question, and maybe you have already done it, but...
Have you tried inserting all the clips, side by side, of the camera_1 in the Video_1 +
Audio_2 + Audio_3 tracks, and all the clips, syde by side, of the camera_2 in theVideo_2 +
Audio_4 + Audio_5 tracks, and the Master stereo audio in the Video_3_empty +
Audio_6 + Audio_7 tracks?
My first try was like this but i discover the multicam tutorial but not understanding how to add more clips to the same mixer or convert existing tracks into mixers.
I make another try and i succeed to convert two tracks into mixers. But i have to say the conversion method is unpractical for a beginner ...
I have created multiple useless mixer and don't know of to remove it for the moment ... 🤣
So you have some splitted videos from camera_1 and some splitted video from
camera_2; more an audio (stereo) from microphone. It is right?
That's it. The audio track came directly from the audio mixer. The audio from camera will be used in the final mix as "ambiance" microphone.
Yes, for me it is also the right way to work. However, not using multicam, I didn't think about the problems it brought.
When i see the difficulties i have, i think there is not enough people using multicam and doing suggestions to improve the interface.
If i understand you idea with "nesting" (not sure what it really mean for the moment) :
- Create a new project and sync "camera 1" files with the audio reference track and remove it.
- Same thing for "camera 2".
- Create a new project and insert "camera 1" and "camera 2" into this third project.
After that, i convert my track into mixers like Igorbeg suggest, sync all track with the audio reference and "make the mixing part".
Am i wrong ?
Could you, @Kabak , share the first two video files (2GB +2GB) by Cam1 and the audio from the audio mixer, if there is not privacy problem?
I would like to take a look at it, if possible.
For the privacy and "copyright" , i think it's not an problem. I think i can make a specific record with the same camera and a 7 segments display counter using 25 Hz at input to have a good view of the missing frames issue and using a rhythm generator from audacity for the audio track.
My big concern is about sending those file due to my low bandwidth. About 20 hours if i didn't have interruption. Some FTP can't resume file transfer, transferring 700mb file is already a challenge on them. Maybe using torrent ?
@Kabak
I agree with you for the bandwidth. Well, It doesn't matter to share the media.
I saw your "MTS-concat-needed-2.png" screenshot.
Could you tell me if the last frame of the first video is the same of the first frame of the second video?
I would like to know if each video contains, in the head and in the end, the two same frames of the previous video and the next video:
AAAAA
ABBBBBC
CCCCCC
@igorbeg I think there is no common frame between the two files.
01150.MTS and 01150-01151 start are perfectly aligned
01151.MTS and 01150-01151 ends are perfectly aligned (checked with spotlights switching)
When you see info from the media, you see there is frame missing (1629 + 17277 = 18906 <> 18909 ):
01150-01151.MTS 1 video stream Vid0 (0), id 0x00001b: video1 h264 1920x1080 25.00 pix yuv420p interlace (container level): 2 interlace (codec level): -1 color space:unknown/ range:unknown codec Detected SAR: 1:1 , DAR 16:9 18909+0 frms 756.34 secs 0:12:36.34 01150.MTS 1 video stream Vid0 (0), id 0x00001b: video1 h264 1920x1080 25.00 pix yuv420p interlace (container level): 2 interlace (codec level): -1 color space:unknown/ range:unknown codec Detected SAR: 1:1 , DAR 16:9 17277+0 frms 691.06 secs 0:11:31.06 01151.MTS 1 video stream Vid0 (0), id 0x00001b: video1 h264 1920x1080 25.00 pix yuv420p interlace (container level): 2 interlace (codec level): -1 color space:unknown/ range:unknown codec Detected SAR: 1:1 , DAR 16:9 1629+0 frms 65.14 secs 0:01:05.14
MPEG transportation stream use packets so i think you have something like that :
File 1 : Packet x : Frame A1, A2, A3
File 1 : Packet y : Frame B1, partial B2
File 2 : Packet y : Frame Partial B2, B3
File 2 : Packet z : Frame C1, C2, C3
Packet y is split between two files so CRC integrity failed and so the frames B1, B2, B3 are lost.
When you use "cat" command, you join the files, the packet is entire so it's work fine.
Yes, the concept of nested assets is as you say: create sub-projects to be later inserted into the "mother" project. The advantage is not having to render the sub-project but leaving it "virtual" (i.e., described in the EDL/xml) so that you can vary cuts and plugins when needed. Nested clips can create a clip of one part of the timeline so that it is independent of the rest and we can then work with cuts and plugins we don't want on the remaining clips. It is basically a way of treating multiple clips as one render but with the ability to continue editing on them. But this is all misleading with respect to your problems with multicam, it was better if I didn't talk about it. Sorry.
It is true that multicam is little used, judging by the rare requests for it; I only tried to repeat a user's tutorial ( https://cinelerra-gg.org/download/CinelerraGG_Manual/Workflow_with_multi_cam_ext.html) with the material he provided. If you can find a good workflow and remedy the loss of frame/sync, let us know, that we will improve the manual. One test you can do, probably unnecessary, is to use "samples" as the timebar unit instead of the classic "hours:minutes:seconds:frames." It is done simply by CTRL + click (LMB) on the timebar.
I got 404 error from the link : https://cinelerra-gg.org/download/testing/cinelerra-forum.zip 🤔
But this is all misleading with respect to your problems with multicam, it was better if I didn't talk about it.
Maybe, maybe not.
If we want to build an easier workflow, it can have a place.
If you can find a good workflow and remedy the loss of frame/sync, let us know, that we will improve the manual.
I will tell you.
For the lost frames with splitted MTS file, it's seem to be an know issue with many softwares (even ffmpeg) so i think it must be treated when we load it into the media window.
We will know if i can make "test files" for this part.
True, the link is no longer active. If you feel like downloading it from Google Drive, I still have a copy:
https://drive.google.com/file/d/1XfIlyF60NWGlcZsL1efgfkx2fGria3RL/view?usp=drive_link
I was looking a solution for merging my MTS/TS files into a single virtual file and i think i found the perfect way : concatfs
It use FUSE to make the virtual file. No need loosing disk space and time in files concatenation or making big modification to cinelerra. This methods build the result file few seconds and work fine.
With this, i think we can build a specific assistant to import those files by selecting necessary files and group them.
It seem to be also possible to automatically group the different record from a camera by using metadata (is start date/hour + duration similar to next file) and by using MTS packet counter (4 bits counter, little after sync character 1/16 probability of wrong detection).
In my real case, one camera with 14 files from 4 record loaded in less of 1 minute for an single will be a huge improvement.
Over time I have repeatedly tried to create a script that would take the various media + external audio collected in a folder, do audio/video synchronization, and then delete the internal audio tracks and replace them with the external one. I thought it could be done with ffmpeg and its feature. Unfortunately, I do not have the capability and have always failed. One could also introduce, in such a script, the concatenation of the files.
1- collect the files into one folder (video files+internal audio; external audio).
2- Synchronize the files with the external audio track (your choice of timecode or waveform).
3- Duplicate and break the external audio track so that there are coincidental chunks with the various files.
4- Associate the audio clips with the relevant files.
5- Delete the internal audio tracks leaving only the external one.
6- Concatenate the files for each camera/scene with which they were recorded (also leaving blanks if they are not strictly consecutive).
I guess it would be a real program rather than a simple script.
Yes, i think a "script" is not the right term in this case.
