Yes! Got it! 🥳
Switched from #Vulkan to #WebGPU in my video synthesizer / mixer.
Using #eframe instead of plain #egui now and also changed the API in that process.
The application compiles again. Loading pictures and connecting them to other nodes works.
The core, which binds everything together, is fully functional.
Not much to see at the moment. After removing piles of old code, I will add the rest of the filter and then release.
Cleaning up before was so worth it!
Now I can concentrate to work on the dictionary again to add more signs. I completely stopped that until the model was finished.
The alphabet for french #LSF is still incomplete. Will release a new version, once this is done.
And I added an option to switch to another #SignLanguage like #ASL and #DGS. But only the letter 'A' is available in the dictionary for both yet. Still interesting to be able to switch.
Building a wasm/gui version of my dithering thing.
last night, I got the file picker and image display to work!
next up, the porting over my dithering code, giving users some params to mess with, then display and save!
As I use #RustLang with #egui | #eframe and #WebGPU, the code could be compiled into a web app.
But I don't think, my front-end is good enough. The UX has a ... lot of opportunities for improvement.
All in all, it would be the best to separate the sign data from the front-end, wouldn't it?
It would make cooperative working on the signs much easier. More people could get write access to just the sign data with better available tools than git.
But I still want to add my part to it. How I see a solution.
So I worked a couple of weeks on a #Rust project and learned to integrate #WebGPU into #egui (using #eframe this time).
Too bleeding edge for a lot of documentation, but it was enough to make it run.
https://gitlab.freedesktop.org/AdeptVeritatis/lisa
Prototype is running as expected.
Only two signs for french sign language #lsf are available at the moment.
AdeptVeritatis
LiSA
nathan teoh