I’m using Shotcut 23-07-29. I’m a rank amateur processing videos for an organization to post on YouTube playlists. This issue has to do with FaceBook recordings of Zoom classes. The recordings are in Gallery view, but I want the teacher’s box to fill the screen. I have done this successfully in the past, but this time I’m having difficulty.
It’s easy enough to adjust the size and position at the start. But when someone joins or leaves the session, often the size and position has to be readjusted.
Here’s what I’m doing:
Set S&P at the beginning and add a smooth keyframe.
At the last frame before a change, add another keyframe.
In the next frame readjust S&P and add a keyframe.
Continue for each change.
Here’s the problem: about an hour of the hour and a half session is all at 500%, then at the end I revert to 100% to get the gallery view of participants thanking the teacher. But within that hour at 500%, I spot-check to find that the percentage changes on it’s own.
I thought that setting a key frame at the start and end of a change would make the size stay the same throughout, but that is not happening in this case.
One follow-up, if I may. I think it is directly related to the Keyframe issue.
I got my 90 minute mlt set up with 5 or 6 keyframes, plus a 6 second intro and two additional tracks, one for a logo that extends the full 90 minutes (with a single size/position filter) and a 15 second simple text datestamp.
I know that all adds processing time, but this is ridiculous: I’m exporting it now and it’s estimating 7 hours to export as an MP4. When I do one of these videos the same but without the keyframes, the export time is usually the same length as the video (in this case 90 minutes).
Is there any help for this, or do I have to abandon the effort to center the teacher?
The Size, Position, & Rotate filter is heavy, not the keyframes. It does not matter how many of these filters exist (unless there are multiple for each frame). It depends on the resolution, frame rate, and number of frames. The GPU Effects version of this is much faster, but you cannot convert an existing project to GPU Effects and vice versa. Also, you cannot rely on an early estimate of the export time, especially in version 23.07, and it cannot predict the future complexity of your project. But if your project has a consistent complexity, then around 5% it should be a decent estimate. You can try turning on Export > Advanced > Parallel processing, and it might help. The export time also depends on which export preset and whether hardware encoder is being used. Among the modern codecs, only H.264 or hardware encoders are fast - definitely not software AV1, HEVC, or VP9 encoders. There are many variables.
I use a YouTube preset. Resolution is 1280x720. Parallel processing is on. The codec is h264_qsv. I also use “preset=veryfast” under Other, which was recommended to me in another inquiry, although I’ve never noticed a difference with that.
In the past I have noticed the estimation adjustments you mention, but in this case it’s been working for over an hour and still says 6-1/2 hours to go.
When this is done, I’ll update to the newer version and try again with the next video. But at this point I’m starting to think I just need to keep the Zoom gallery view.
Just for the record, I got the answer I needed 9 hours ago. I finished the job and started the export shortly after that. The video has been exporting all this time. It says 20 minutes to go. If it isn’t the number of keyframes, I’m guessing it must be the extreme enlargement. The vast majority of the video is 500% enlarged and the rest of it is 400%.
Anyway, there’s no question here. Just thought I’d report back.