Almar Klein
Almar Klein
> did you ever create that helper function to sample an atlas correctly? not yet
@Korijn This pr is ready from my end.
Short term planning: * Get this pr approved. * Do a release of pygfx. * Merge this PR. * Do text performance improvements. * Atlas offset thingy. * Release again....
I actually meant to do a pygfx release before merging this 😅 , but not a big deal; this PR is surprisingly backwards compatible.
> You can just revert the merge if you need to. nah, it's alright
I think setting the objects `.visible` flag would be an easier (and possibly more performant) way to do this, unless gpu memory is a bottleneck too. As for the update,...
My guess is that this is either a lavapipe glitch, or a corruption in the atlas. Looks like both cases are rare and you were fortunate to somehow trigger it...
I'm fine with merging this. We plan to revisit mage comparison (https://github.com/pygfx/image-comparison) anyway.
I had to think of this again when the image comparison failed in #902. I had an idea what could explain the different results. Writing this down so we can...
No longer relevant with #1002 🚀