This project attempts to follow Semantic Versioning and uses Keep-a-Changelog formatting. But I make mistakes sometimes.
- Better segmentation cutout behavior when specifying a list of segment ids.
- Segmentation cutouts are returned as int64 array if it is safe to do so. Can be turned off by setting
convert_to_int64
argument toFalse
.
- Fixed bug in
save_imagery
. composite_overlay
can take arrays and not just masks.
- By default, the client will adopt the lowest mip level (i.e. highest resolution data) that is not labeled as a placeholder in the cloudvolume info.
- Richer integration of specified resolution, to take advantage of better Cloudvolume support for resolutions.
- Split apart the notion of bounding box dimensions from image dimensions. See bbox_size vs image_size in the readme.
- Vastly improved configuration by a CAVEclient
- Order of arguments is more consistent across functions, and more sensible as well.
- 'voxel_dimensions' arguments can be used to specify pixel dimensions instead of voxel dimensions. If used, the 'bounds' argument instead specifies the center
- Switched from annotationframeworkclient to caveclient
- No longer need caveclient (although it's still useful)
- Directly saving files is allowed
- Changed behavior of bounds_from_center to use width/height/depth instead of radius-style values
- More direct use of newer cloudvolume features
- Cloudvolume version 3.1.0 or greater is now required
- Fix install bug
- Adding long description for pypi
- Added seaborn to requirements
- Color generation using the seaborn color palettes.
- Documentation updated.
- Color specification is now optional in overlays and a default color picker is used.
Initial release, for all intents and purposes.