PSSR2 2.1.1

• Predicted images are saved when initially processed rather than simultaneously at the end
• Can now use increasing n_frames (e.g. n_frames=[5,10])
• Fixes

Full Changelog: v2.1.0...v2.1.1

PSSR2 2.1.0

• Added reassemble_sheet function
• Moved normalize_preds to util
• predict_collage works with LR mode
• Fixes

Full Changelog: v2.0.2...v2.1.0

PSSR2 2.0.2

Updated package metadata for PSSR2 consistency

Full Changelog: v2.0.1...v2.0.2

PSSR2 2.0.1

• LR mode is set with lr_scale = -1 instead of None
• SaltPepper applies gain before crappification
• Better dataset repr
• Fixes

Postfix

• Fixed CLI arguments bug

Full Changelog: v2.0.0...v2.0.1

PSSR2 2.0.0

• Added PSSR Napari plugin (training, prediction, runtime monitoring, etc.)
• Renamed loss module to util
• Refactored/removed previous color handling system
• Auto n_frames value changed from None to -1
• Added callbacks on training batch steps
• Lowered required Python version from 3.10 to 3.9
• Model QOL improvements
• CLI properly works with nested objects
• Fixes

Full Changelog: v1.2.2...v2.0.0

PSSR2 1.2.2

• Added SwinIR model
• Altered LR mode functionality
• CLI improvements
• Fixes

Full Changelog: v1.2.1...v1.2.2

PSSR2 1.2.1

• Increased allowed dependency versions
• Fixes

Full Changelog: v1.2.0...v1.2.1

PSSR2 1.2.0

• Added RDResUNet (RDNet encoder + ResUNet decoder)
• Refactored atrous functionality into current models
• Model optimizations
• Fixes

Full Changelog: v1.1.1...v1.2.0

PSSR2 1.1.1

Updated package metadata for repository transfer

Full Changelog: v1.1.0...v1.1.1

PSSR2 1.1.0

• Introduced and reworked existing multiframe stack handling
• 3d and 2.5d dataloading + models
• Sheet preloading
• MultiCrappifier and SaltPepper crappifiers
• Reworked preprocess_hr into preprocess_dataset
• Train/test splitting is now across tiles rather than frame slices
• Improved error handling
• Various ease of use and quality of life features
• Fixes

Full Changelog: v1.0.2...v1.1.0