I have recently spent some time recording bands, I get an audio recording of the entire show and multiple camera shots, sometimes handheld short clips, sometimes longer runs. I have imported as many as 30 video clips and cut between them in FCP and it “just works”. I’m trying to do the same in kdenlive as I would much prefer an open source solution.
My latest show is comprised of one audio track and 4 video clips. I imported the 5 files in kdenlive and set the audio clip as the reference and then tried to sync to the audio reference on the video files. So far, the processing time has been significantly longer than FCP and, to date, none of my attempts have successfully aligned the videos to the audio. I get no errors, no warnings, and no re-alignment. Just a bunch of a processing and then nothing. Maybe my workflow is not correct? I’ve been going over documentation and youtube how-tos but, so far, have come up empty. Anyone have any tips for aligning videos off a common audio track? Should kdenlive be able to align on audio tracks? The audio source for the videos is technically a different microphone than the audio track (not sure how else this could be done). Again, this “just works” in FCP with the same files so I was hoping kdenlive could do something similar.
As I am a complete newbie to kdenlive, I will apologize, I still don’t know why/how I got this to work, but a few crashes, few restarts and few attempts later, I got it to work. Once I moved the audio track manually off the timeline of the video tracks, I was able to successfully align. Not sure why I had to move the audio track first, but it got it working. I “think” what happened is that the first video file was started well before the audio file started. This meant that if the video was aligned, there was not enough space to the left of the audio file to place the video there. Not quite sure why kdenlive couldn’t just shift everything but I won’t blame kdenlive when I just need to learn the workflow.
Step 1 done, now I have to figure out how basic multi-camera editing works in kdenlive. If anyone has any tips or a good tutorial to follow, I would appreciate it.
There are still a few issues here that are probably interesting to address.
Did it really give you no warning at all? Usually when some operation can’t fit a track because of timeline constraints it will show (or at least flash) a warning on the status bar at the very bottom of the UI window - but it is a busy UI, and if you don’t know to expect them they can be easy to miss - and in a few cases, something else replaces them very quickly, making them easier to miss, or difficult to read.
It definitely should warn if that really was the problem, so this is potentially a bug to address.
Was it still very slow even in the case when it succeeded? Autocorrelation is a relatively expensive operation mathematically, so if this is still very slow we should check that the library being used for that has been built will all optimisations enabled, as this can make a very big difference. What platform (and kdenlive build) are you working with?
If some of the files had a very long lead in before the first alignment point, that could/would have contributed to this too. If you know that is the case, one way you might speed it up is to cut the initial part of the video off in the timeline clip - then place markers on all the clips once an alignment has been found. After that you can restore the full clip and align the markers.
I know how the DSP for this works under the hood, but haven’t really used or looked at the code for what kdenlive is doing with this yet - so I’m not sure how it chooses what to move (and/or what to refuse to move), but there might be some real possibilities for improvement with that too.
The audio alignment code and multicam support has been in kdenlive for a while now, but it is one of the more ‘specialised’ use cases which surely doesn’t get as much live-firing exercise as some other features - so if this is something you’ve got a lot of experience with, your feedback is definitely valuable to us - both for improving the code, and if/where necessary, the user manual.
As you build more perspective on what is different in kdenlive to how you’ve been doing this previously, and what does or doesn’t work well - please do feel welcome to share your thoughts on what could work better. There’s always going to be a learning curve for new users - but it’s a process that goes both ways for people with real lived experience in the more specialised use cases.
Some of these things seem like they should usefully be bug reports if they still look like sharp edges that might cut you again on a later project.
There’s documentation on multicam in the user manual, and a few user-made tutorials that shouldn’t be hard to find, though I can’t particularly vouch for how useful they’ll be to you - but the manual at least we can definitely improve if you find things that it doesn’t explain or could explain better.
@stauf Multicam editing was introduced in Kdenlive as a kind of “live cutting while playing” in mind. We got only few feedback on mulitcam editing so far.
It would be very interesting to have a feedback what is missing and what is not working to meet your workflow.
I apologize for disappearing, I have had a big change in my personal life (parent passing) so my time on this has been minimal of late.
Basically my goal is to find a FCP alternative. What I tend to do is video events myself. To accomplish this, I usually have a couple static cameras, an audio recorder and then a roaming camera I use for spot work. Then, in post, I want to sync each of the cameras to the audio recording (via the audio) and then make cuts to the various “angles” as I see fit.
This tends to be labor intensive so I don’t do it a ton. Every time I go back to it, I feel like I need to re-learn FCP so, this time, I felt like checking out kdenlive would be a good use of my time. I am considering an investment in a new computer, that is really my drive. I hate that I seem to use a different OS for different applications. Mainly, my reason for OSX is FCP. If I can eliminate FCP from my workflow then that makes a big difference in how I think about this upgrade.
I am hoping to provide more feedback soon but it certainly appears as if kdenlive has potential in being the solution I am looking for.
One question to start, what Linux variant would you say is most supported/tested? I saw one person saying EndeavourOS was their recommendation but I have little experience with it and have noticed a couple crashes. I really don’t mind changing distros if there is one better suited. Ubuntu Studio maybe?
Put all videos into the timeline. Define the which audio is the “master”. Then follow these steps to align all other videos to the audio “master/reference”. Maybe small adjustments of each video are needed due to camera distance from the sound source (speed of sound).
Enable multicam view and make a raw cut. Afterwards move all clips onto one track. Make final adjustments with Slip and Ripple. Add effects and title.