Update v1.9.3

New features

• Added txt2vid (only works with graphics cards that have compute capability of 7.5 and above)
• Modified the design of layers

Bug fixes

• Undo/redo fixes: this feature works as expected now

Build changes

• Upgraded to torch 2.0
• Exclude unneeded libraries for smaller build
• Previous builds relied on a separate aibuilder release for windows (aibuilderwindows), now both linux and windows use the same install path
• Upgraded bitsandbytes
• Upgraded torch, torchvision, torchaudio to 2.0 (along with various other library updates)
• Moved docker builds to the cloud (before i was building docker images locally and pushing to the cloud which would then use the images to build the code and deploy…. now everything happens in the cloud)
• Modified included / excluded libraries (no longer including tkinter files in windows… annoying locale binaries are gone)
• Fixed pypi build
• Updated installation instructions
• Moved installation instructions from README to wiki
• Added a manual to the wiki

Extra info

LoRA and extension support development has begun and is nearly complete, I hope to get these changes into the next release. Additionally I have started adding a feature to allow widgets to be added to the canvas. Currently there is a blank button in the toolbar which will add a word balloon to the canvas. This button was released by mistake and will be removed in the next release, but will be re-added when the feature is complete.