Thanks for kdenlive. I really like it. I thought it might be helpful to someone if I record a few of my initial struggles with it, while going up the learning curve - not a bug-report “per se”, but user-feedback from a new (to this application) user may be helpful.
I did manage to achieve my objective to edit a video in a few hours. The KDEnlive docs are very helpful.
I struggled a bit with the “clip editor”. I was trying to get the in/out points exactly right, and couldn’t see how to set the video to play from/to the selected endpoint. There isn’t (?) a way to play the video starting at the “IN” point. Also, there’s no way to zoom the timeline in the clip editor: so a clip that is 20 mins long requires very precise mouse-movements! In the end, I did this roughly, and corrected it precisely using the main timeline instead, and that worked.
It would be nice to have an easy way to fade exponentially (either way up, like _ or -\ ), depending on the type of end we want. Perhaps this could be a property of the fade in/out dialog?
I wanted to superimpose a title (on V2), in the middle of a continuous segment of video on V1 (which happened to be nearly black). Fading the title in/out didn’t work - as I needed a way to “move the clip to the top of the stack”. I couldn’t figure this out - I know I was fading V2 up, but V1 remained opaque.
It took me a while to figure out that the menu option “cut clip” meant to "cut the track into two parts, as opposed to “cut-and-paste”. This would be a good place for a short tooltip with shift.
I was trying to align a second audio track with the video/audio, and needed greater precision than 1-frame. You can’t do this. (I think).
When using the “typewriter effect” on a title, this runs at constant speed. So it can happen that the title is not fully “typed” before the clip moves on.
Anyway, I hope some of this is helpful. Thanks once again.
Hello @RichardNeill . Welcome to the forum.
Thank you for the feedback. Some input to your points:
In clip monitor hold CTRL and scroll with the wheel on the play-line. A new zoom bar appears where you can adjust the clip.
New easing modes for keyframes come with the upcoming version 24.02. Like exponential in/out.
I think you have done it the right way. Video on V1 and title on V2 (on any place). Now zoom into the timeline that you see the title clip. Click on the red circle on top left and right corner of the clip (Hints are shown on the left side in the status bar when hovering over this points). Now fade appears with the length defined in the settings.
I think you mean Cut-Clip=Razor tool (Shift+R). This cuts the selected clip at the play head position. Do you mean “copy and paste” instead of “cut and paste”?
No. Minimum shift/move is 1 frame. To move less then 1 frame, you have to ungroup the clip and add in an audio tool some silent milliseconds on the begin or end of the clip and re-import it into Kdenlive. You can assign in the settings the application (like Audacity) that will open by double click an audio file. See Settings → Configure Kdenlive > Environment Default Apps (Configure Kdenlive — Kdenlive Manual 23.08 documentation).
Yes, minimum speed is 1 letter by 1 frame (variable “frame step”).
Correct! And why would you expect a video editor to be able to further divide the smallest entity you can have in a video?
It’s actually not an entirely unreasonable thing to want, even if the current editing interface doesn’t support it. There’s already an almost certain mismatch in frame sizes and alignment between video frames and any audio tracks - so the current constant (video) frame size we use means we force the first frames to align, but then most won’t until the next common multiple of their sizes.
But that’s all hidden behind the scenes when mlt decodes and then renders the combination of all tracks in the timeline to the output codecs and container.
It might be an interesting future thing to consider, since an editing interface that could support this can then also easily support mixed input video profiles and variable frame rate video, which is surely only going to become more common than it is at present as the push for better compression with greater quality keeps marching on.
We already support audio codecs (like opus) which can have variable frame sizes, because all we care about is the granularity with which we align the start of things. We could allow (sub)millisecond granularity for that editing instead of “project frame rate” chunks. But I suspect that would have to start with mlt …
Thanks for your comments.
Re 2: thanks - that’s exactly what I wanted to do!
Re: 5: yes, I did mean Shift-R = Cut Clip. I was confused by the word “Cut” - which I wrongly understood to mean “cut this entire clip out of the timeline, either because it isn’t wanted, or because it will then be placed on the clipboard ready to paste elsewhere”. I was confused because usually “Cut” is one of the standard editing trio of cut/copy/paste, and it took me a while to figure out that this was the wrong context… So maybe this would be clearer if the right-click menu said something like “Cut Clip (Razor)” or “Cut clip (new split point)” or something.
Re 6: As for audio alignment, I did eventually do this in audacity. I had a slightly unusual editing problem, in that I was trying to mix two different source-recordings of the same sound, which both had to play simultaneously, therefore I needed perfect sync between them.
The context was this: I had put together a 20 minute fireworks + music display. So the sound recorded by the camera had all the audience-reaction + the bangs and crackles + the music - but the music had been messed up by the camera’s AGC. So I wanted to mix back in, slightly louder, another audio track containing the original MP3 of the music, but I had to align it with the camera’s audio to a precision of a couple of ms.
What I actually did here, (if anyone else finds this thread for reference) was:
- split the video file into separate audio (m4a) and video (mv4) tracks with ffmpeg.
- edit the audio file in audacity, adding in the original MP3 files (at higher volume level) using the spectragram view to aid visual alignment.
- export from audacity.
- re-merge into a combined mp4 file (with ffmpeg)
- then do the edits in kdenlive.
Re: 7 - maybe it would help if the typewriter mode warned if the title-duration wasn’t going to be long enough?
Thanks once again - Richard
Just on a related note to my “wouldn’t it be nice to be able to work in the time resolution of the actual smallest entity in a project (probably usually audio samples, but also mixed frame rate video) during the editing phase”, regardless of what the final render profile is …
It would sometimes be nice to be able to work with an intermediate image size that is larger than the input or rendered video is too.
For example, if I currently add two transform effects to a clip, and on the first I translate X +1000, then the second I translate -500, the first effect will crop the right side of the image to be 1000 smaller than what it originally was, so after the second I will have an image with an empty 500 wide stripe down both sides, rather than just one on the left side as it would be if the transforms were just added to give an overall +500 shift to the end result.
My concrete use case for this is I sometimes use the motion tracker to “stabilise” a clip tightly on some subject of interest, and apply its keyframes to a transform. Then use a second transform to scale and ‘center’ the result of that so the image still fills the whole frame. But often that means I must zoom in a lot more than I otherwise might, and select the offset of the first transform very carefully, because it will crop the edges of the image such that the second transform cannot get them back again.