Open xinaesthete opened 3 months ago
ok @ilan-gold with latest luma.gl it does now render volumes correctly - although still using a bit of an ugly hack so we don't get errors about null
texture bindings... that should be cleaned up before this gets merged (in both xr-layer.js
and xr-3d-layer.js
). Can't remember off the top of my head whether there were any other significant things (apart from material-ui / react etc maybe being worth updating).
This is up to date with main branch.
FWIW, the volume rendering is also much quicker with a version of the shader that doesn't try to read all 6 values - would be good for that to be implemented differently at some point.
It might be good to also update React / Material-UI / Zustand;
Agreed, but let's consider a separate PR for that. This PR is large and involves breaking changes, so I suggest:
dev
)dev
Then we can merge dev
into main when we are ready to make a release. This approach will result in smaller, more manageable PRs.
ok @ilan-gold with latest luma.gl it does now render volumes correctly - although still using a bit of an ugly hack so we don't get errors about null texture bindings...
Great
FWIW, the volume rendering is also much quicker with a version of the shader that doesn't try to read all 6 values - would be good for that to be implemented differently at some point.
I'll have to look into this. I vaguely remember trying to optimize this at one point but honestly don't remember super well.
The RGB now renders but it is a bit slow for me. Let me know @xinaesthete if you get my email with data and see the same. IF so, I will try to make a smaller reproducer to focus the issue and maybe report because profiling isn't showing anything.
I'll have a look, I think the performance is similar to what it always was, which could definitely be improved.
Edit, not tried with RGB, was mostly thinking about volume.
Also I'm not sure why the tests hang on (or after?) packages/layers test: ScaleBarLayer#Empty props
- haven't checked whether that's the same with main.
All other tests pass if scale-bar-layer is skipped.
I've noticed there's a bug with 'lens' - switching channel causes something to go badly wrong (edit: should be fixed now).
Addresses # https://github.com/hms-dbmi/viv/issues/804
Background
As discussed https://github.com/hms-dbmi/viv/discussions/790
Fairly extensive changes will be required; discussion should probably migrate to here.