Thanks for sharing this! Stories like this are priceless guides toward best practice, as we all figure out what that is and what sort of tools will help us the most.
The vertical versions are often shorter clips, or re-edits adapted for social media, sometimes with different pacing or structure.
I think this is a key point for distinguishing between things that just add “infinite flexibility” and things that might actually assist in this kind of workflow. And confirms that the people who are doing this best aren’t just smashing a single edit into different formats. They have a story to tell, and different audiences to tell it to, which is much more than just a different screen layout.
To maybe map it to its old school predecessors:
- There’s the full length feature version
- The trailer(s) of the full length feature version
- The reader’s digest condensed version
- The illustrated children’s version
And maybe a serial and spinoffs. Either way, there’s a pool of common resources and some things which are specific to each format - and our job as toolmakers is to minimise the amount of busy and re- work needed to create them all as richly as possible.
I duplicate the sequence to export different versions, all still in 16:9 but at different resolutions.
Can I get you to elaborate on this a little? Are you actually re-editing them to crop things out of the lower resolution versions or shift the focus of a scene - or just re-rendering them to have 4k, 1080p, 720p etc. copies of an otherwise identical sequence?
The latter I do regularly (I always shoot and edit in 4k, but the lower resolutions are usually more appropriate versions to share) - but that doesn’t need the sequence to be duplicated.
So if you’re doing something that really does need it duplicated, it’s probably a use case that we should also explicitly support well.
The real challenge: switching to 9:16
This one should be a lot easier if all your source material is in a common format. It’s the case where some of your footage is 4k 16:9, and some of it is from a phone shot in some arbitrary resolution/aspect ratio/colour space/frame rate, that things get really messy.
Option 1: Export master > import into new 9:16 project
Yeah, you’ve exactly captured my feelings about the value of investing effort into this as a naive transformation.
However you can actually do this without exporting and reimporting - in the render dialog, you have the option to rescale and select the output aspect ratio.
So from your 4k master project, you can render a 16:9 1080p version (which simply scales it), or a 1:1 or 9:16 version (which preserves the vertical resolution, and simply crops the horizontal excess).
Which means you could have a separate/duplicated sequence in your 4k project where you re-edit the clips to pan and scale the part of the 16:9 frame that you want visible in the 9:16 cut to be in the center of the frame (and you can apply a crop effect to the sequence master or a track, as well as to individual clips, which will help visualise the result before actually rendering). Then you just render that sequence with the aspect selected in the render dialog. Any cropping that you applied in the sequence will just itself be cropped/ignored unless you crop something even harder than the rendering step will.
That lets you do it all in the one project, and lets you operate on your clips in their native resolution and aspect ratio in all of them. Only what is output is different. That should give you much better results and a better working environment than importing your 16:9 clips into a 9:16 project.
It still gets messier if you have mixed source material - but the basic idea is your project format should match your source clip format - or if they are mixed, the highest resolution you want to interact with your source clips in. The format/resolution(s) that you later render each sequence in aren’t strictly linked to that.
Option 2: Open a new Kdenlive project with 9:16 resolution, copy-paste the sequence, and apply “Original Size” + transform effects to every clip
If you really did have a mishmash of source material in different formats - this might be a useful option. eg. if you want to create a 16:9 cut from primarily 16:9 source material, but including just a few 9:16 format source sequences - and a 9:16 cut from primarily 9:16 source material but including a few of the 16:9 clips - then you could have two separate projects and share things between them by having both projects open at the same time so you can copy things between them.
But as you note, lots of things (like keyframes) don’t automatically get rescaled - which is actually a completely separate, and more complex again, problem to supporting sequences and clips with different resolutions and rates to the project configuration. But one that is also interesting to consider. It’s not even clear to me right now whether we can sanely just always “fix that automatically” (what if you don’t want those keyframes rescaled because you’re applying some separate additional transform to that clip?). But having some easy user controllable way to scale a set of keyframes does seem worth investigating.
We have a similar problem with things like time remapping. You can’t do motion tracking on a time remapped clip because the motion tracker keyframes are based on the speed of the original clip. And changing a remapping can screw with already existing keyframes. We need to fix that, but it’s a Complex problem with more than one Right answer that aren’t all compatible with each other.
What would be ideal for me?
- Being able to create multiple sequences with independent resolution settings within the same project file.
I think we need to clarify that there are two separate sets of ‘settings’ here.
There’s the project settings. Which with things the way they are today, the optimal setting is one that matches the majority of (or best of) your source clips.
And there’s the export settings which is the format you want an individual rendering of a sequence to have (and a single sequence can have an infinite number of export settings, you get to choose those every time you render, you don’t need to duplicate a sequence just to change only that, you just need to render it again to a different output file with different rendering settings).
In the case you described here, where all your source footage is 4k 16:9, we should actually support that pretty well. You just create one project to suit your source clips, and then use separate sequences for each of the actually different edits you want create and render.
It might be a little counter-intuitive at first, but you just always have a raw working space the size of your source clips, but edit some sequences to fit what you want in a viewport that is a subset of that - much the same way you would if you had broadcast safe/title safe/safe margin considerations you needed to account for.
You have a 16:9 monitor to see your whole raw clip in, but you’re editing with safe margins appropriate for the 9:16 viewport you’ll be rendering.
- And having those sequences preserve all clip transformations, effects, and full editability (access to raw media) without resetting them.
It’s the definition of ‘preserve’ there which makes the ‘correct’ behaviour somewhat variable. There are use cases where preserve means not scaling them, and cases where it means you do. And there isn’t really a clear technical distinction that separates them - it’s what the person editing them wants to happen. So what I thing we really need here is some way for them to tell us what they want to happen in each of those cases.
Probably something like a reference frame size/rate that the keyframes are scaled to, vs the clip size that they are applied to. A bit like the way ASS subtitles use ‘script resolution’ units, so they can maintain proper proportions regardless of the frame size of the video they are overlayed on.
This is definitely a One For the Roadmap thing.
As is dealing with things using different time scales. If we want to get better at handling audio (which we do), there’s no getting around that one. We maybe could do more to hide the fact we’re transcoding things to match them to a given project frame rate - but I think that’s the wrong approach. If the goal is to be the best at this, the users who care need to be able to control all that to fit their needs. We might “sell more units today” by just targeting the majority who don’t care, but if we get this right then both groups will win.
video sources are often so varied (in resolution, FPS, format…), depending on where they come from, that pre-conforming everything before import becomes tedious and time-consuming.
Yeah, how we handle that definitely has lots of room for improvement. When it comes to combining them for editing, the time consuming part might not be completely avoidable (because creating proxies and intermediates that make real time editing feasible just inherently takes time) - but we certainly should be able to improve on the tedious part of it which makes it become work for editors, not just time.
There probably are two somewhat different groups of users here too - there’s the people mashing together found footage and holiday snaps shot on multiple different cameras with little regard to the issues we’re talking about here.
And then there’s the, let’s go with Industry Professionals - who know why choosing a single format to shoot and process in and having synchronised audio and even camera genlocks and all that sort of stuff will make their life so much easier when it comes time to edit.
In the case you started from, where all the master footage is all in the same format, did I miss something important in what I outlined above, or does that actually mean you can already do what you need for that in a single project today?
That wasn’t quite what I was expecting from the outset, but having a quick poke at it, the only real glitch I hit seems to be that the “Edge crop” effect doesn’t appear to work as a “Master” effect on a sequence, or on a track, only on individual clips. But the other cropping effects don’t seem to have that problem and can be used to set a ‘viewport’ to work to. And it might be useful to have a monitor overlay for alternate aspect ratios a but like the one we have for broadcast safe margins.
