AI Runner with Stable Diffusion | AI Art Editor and Custom Chatbots
v1.11.1
Introducing AI Runner v1.11.0! This latest version brings a host of exciting improvements and new features to enhance your experience. Let’s take a look at the highlights:
The drawing tool has undergone significant enhancements, providing smoother brush strokes with improved rendering and anti-aliasing. You now have the ability to adjust the opacity of brush strokes and easily create straight lines by holding shift and clicking two points.
Design improvements include a revamped layout for the embeddings list, a dynamic brush cursor that adjusts in size, and a more visible border for the active grid area. Additionally, the strength slider now maintains its correct initial value.
LoRA is now built-in and enabled by default in AI Runner. Experience its seamless integration and improved functionality.
Several miscellaneous updates have been implemented, such as the option to toggle sending new images to a new layer, upgraded libraries, and the addition of unit tests. I’ve also made it easier to work with custom paths for models, LoRA, and Textual Embedding, eliminating the need to restart the software when changing these paths.
To optimize memory usage, I’ve improved model handling and memory settings for faster inference speed. The memory management window now includes helpful labels and tooltips.
As for bug fixes, I’ve resolved crashes when switching between models and fixed issues with exporting to JPEG, pasting and copying images, draw history undo/redo, image filters, and xformers. Moreover, Triton has been added to the Ubuntu build.
These are just the highlights of the numerous improvements and bug fixes brought by AI Runner v1.11.0. Take a look at the full changelog below
Changelog
Drawing tool improvements
- Improved rendering of brush strokes
- Added anti-aliasing to brush strokes
- Added ability to change opacity of brush strokes
- Clicking without dragging now adds a single brush stroke
- Added ability to draw a strait line by holding shift and clicking two points
Design improvements
- Improve layout of embeddings list
- Added a brush cursor that changes size with brush
- Improved active grid area border so that it doesn’t blend in with background color
- Fixed the strength slider initial value
LoRA
- LoRA is now built-in to AI Runner and enabled by default
- LoRA now work as expected
Misc
- Ability to toggle sending new images to a new layer (ability to work with single layer if desired)
- Removed extensions
- Upgraded multiple libraries
- Removed extensions until the feature matures
- Adds unit tests
- Various improvements to code backend
- Added support for weighted prompting using Compel
- Added ability to set model, LoRA and Textual Embedding custom paths
- Removed need to restart when changing model, LoRA and Textual Embedding paths
Memory usage improvements
- Improved handling of models and memory settings for inference speed boost
- Added labels and tool tips to memory management window
Bug fixes
- Application no longer crashes when switching between models
- Unable to export to JPEG fixed
- Fixes pasting image
- Fixes copying image
- Fixes draw history undo / redo
- Image filters work again - some were broken
- Fixes xformers
- Adds Triton to Ubuntu build
Comments
Log in with itch.io to leave a comment.
Hey, just trying it for the first time, coming from a1111, this looks amazing.
![]()
But I have a question after watching the LoRA tutorial video. After I made a path to LoRA and embeddings, it doesn't show as a slider like it did on the video, i can't select any of them too.
That is correct - in the latest version the sliders and checkboxes have been removed.
The LoRA in your folder are automatically applied to the model and can be triggered with the correct LoRA keyword.
For example, if you add the Ghibli LoRA, it will automatically be applied to your model, but it won’t trigger until you use the keyword ghibli style
I will be updating the tutorial videos soon.
Eventually I will re-add the checkboxes and sliders, but they were causing issues and had to be removed for now.
Ah i see, no big deal, can their weight still be adjusted in the prompt?
Currently no - but I will look into adding this feature. LoRA is a bit messy with the diffusers library at the moment and I had to hack around the code in order to get it to work so this is the first phase of that feature. I will definitely be improving it over time though.