Does kdenlive have feature usage analytics ? I’m curious.
I see where you’re coming from, each and every plan for custom features anyone would want will be at least a little vague, it’s a snake biting its own tail, you can’t really describe precisely what you want as there’s currently no way to implement it.
I have ideas for things that suit MY workflow, but I can’t really describe or imagine how it’d be done with an imaginary middleware that does not exist (yet), will it be an effect ? a new window ? a generator ? who knows, I certainly don’t.
The ‘place image accordingly’ in my head is not even part of the tool, since it’d be just a Transform effect.
My (imaginary) tool definitely does not need to be added to the whole of kdenlive, I know it’s niche, I know I’ll probably be the only one using it, and that’s why a broad plugin / module system™️would suit it perfectly.
I’ll attempt to see what’s possible to do myself in the meantime
But for the love of me I can’t make it work inside kdenlive (I’m on Windows so I do things manually, build, try to put the right files in the right place), no can do.
If there’s a defined workflow to add custom effects / generators, I’d love to see it, because I searched quite a lot and found nothing yet.
That’s kind of the essence of what I’ve been trying to explain. If you’re not on the pointy end of actually implementing it, or experimenting with possible implementations, for a problem that you personally understand, then starting with assertions of how it should be done is unlikely to anywhere near as productive as simply describing what you want to do and why.
And if you can’t describe what you want to do, then it’s extremely unlikely that anything someone else implements will do it. Well or at all.
If you design an interface with no real use case in mind, then that’s exactly what you will get. An interface that supports no real use case at all. And the first time a real use case ever comes along, if it ever does, that interface will be entirely insufficient for the real world problems that come with it, and now you’ll have two problems.
And that’s why it really wouldn’t. How could it possibly provide the things your problem needs if you can’t or don’t communicate what problem you want it to solve. “Make it broad” is not a specification for what it needs to be able to do. Solutions created without a problem they were designed to solve are just self-inflicted problems. That’s the only thing they do perfectly.
I’m not on windows, so the precise details might vary a little, but frei0r looks for its plugins in specific directories (either hard-coded or overridden from the environment). You need to drop the plugin there, and put the XML file for it in the directory where kdenlive looks for those resources. And then it should Just Work.
The containerised builds can be a bit more complicated if they don’t look outside the container or let you place things in it, but the principle is the same for all of them.
Having plugins is not only nice but necessary I think. Look at tools like vscode which is now the most used editor for almost any programming language. It is only got there because of its extensibility. people are writing all kind of extensions and everyone use what he/she likes. The other factor here is the language used in the plugin so if I had to write in c++ or c that would restrict the number of people writing plugins by a factor of thousands if not more. And people will move to a new tool (unfortunately, that is actually what I am doing now).
Let me tell what I have in mind and was not able to see how to achieve that in kdenlive directly, I wanted to use sound source separation so that I can directly run it from inside kdenlive and choose which sound source to use. That will require me to be able to run AI models in python ideally (because that is the only productive way to do it). Now what I do I export clips from kdenlive then run the tools then import again to kdenlive. Same for many things such as speech enhancements using advanced AI tools. I also would like to run AI models for differnt things also on videos to do all sort of cool things so imagine a transition between two clips done using AI model such as LTX 2.3.
I can list more. but basically I would imagine a plugin to be something like:
1- normal effect so clip input clip output
2- transition so two clips as input and you produce and clip
3- more general like taking multiple clips and producing multiple for example take a group or a track of sound clips and cut it based on some criteria such as silent.
Please try to consider that. I know that security can be an issue but that is a risk we can be warned about and can be enhanced as we go with the feature.
If you want to know what is coming to Kdenlive in the future, please read our roadmap. Scripting support is just one of the many things we want to do.
Please note that the team is rather small and we cannot make any promises wrt when and what and how. The roadmap is just that: a roadmap, not a schedule.
If you want to help getting things done, please consider donating time or money to the project.