Almar Klein
Almar Klein
Very nice! I can definitely see this as an example. And we can also consider moving (some of) it as a helper class in `pygfx`.
The short answer is that the GPU requires data in little-endian order, and that the buffer does not do auto-conversions, but requires the data to be in a supported format....
The second statement should be the case, but I'm not sure whether it is currently implemented as such; we'd have to double-check.
Done in #795. Big endian data is treated similarly to non-contiguous data: it supported (and automatically converted), but when `force_contiguous` is set, it is disallowed, to make sure it does...
> I could maybe take this thing out of your hands I was hoping this'd nerd-snipe you 😉 But no rush from my end.
> My understanding that mpeg4 would be a huge savings in transmission performance (bytes sent/received) and would make up for any encoding/decoding time increase I think this sentence is missing...
I was planning to experiment/benchmark some ideas for improving the performance. Related: @kushalkolar is interested in encoding to jpeg on the GPU.
> Looks like pywidgets depends on jupyterlab-widgets here: Right, so it's already an implicit dependency. Let's remove it and see what happens.
This weekend I was breaking my head thinking of different ways to approach the problem, when my 2yo took my pen and started scribbling on my drawings, and then suddenly...
> For some reason it looks like the second screenshot has more "jagginess"? The poc has no aa 😉