Adam Tyson

> It now uses a smaller brain for the bright brain. With cache it seems to take around 12 min and without cache up to 30min. I think that's ok....

I'm hesitant to routinely pin max versions. Across the whole BrainGlobe ecosystem (and other tools likely to be installed at the same time like napari plugins), pinning versions causes lots...

Out of interest, is there a reason to downsample using a custom function rather than e.g. [scipy.ndimage.zoom](https://docs.scipy.org/doc/scipy/reference/generated/scipy.ndimage.zoom.html)?

Ah, smart! Thanks.

> Detect the difference but don't replace the atlases automatically. What was done automatically? Nothing in BrainGlobe should update the atlases without specific user interaction.

What started to? What process was running?

Can you link to that file?

This isn't a bug in the atlasapi, the validation script [explicitly updates the atlas](https://github.com/brainglobe/brainglobe-atlasapi/blame/9e1e13f5593a42b11bcddfed93e6cdebb4e28522/brainglobe_atlasapi/atlas_generation/validate_atlases.py#L216). Maybe the validation script should have some logic to deal specifically with atlas versions that are...

You can keep it open, just maybe change the title so we know what the actual issue is. As an aside, I've actually just disabled Discussions. We decided a while...

Ideally you would run the software (either python API or napari plugin) on a large dataset, such as the one outlined in [this tutorial](https://brainglobe.info/tutorials/brainmapper/index.html). It's possible that peformance issues may...