Implementation of CoreML + Live Render Preview #370
Conversation
With this implementation, we can select a CoreML model for the Text To Image mode, which increases performance (currently, the model must be manually added to the .diffusonbee/coreml_models folder and the compiled executable in the src/Debug folder).
|
Is this repo still being updated? |
|
@divamgupta Ayeeee |
|
@AEtheve Haven't seen the developer active anywhere in like a month. Any chance you could explain how to compile this application with your modifications? I'm only mildly versed in Electron and failed miserably when attempting to bundle it up. Thanks for adding this implementation. |
|
Hello, I will probably add more information in the README. In the meantime, you can contact me on Discord in a private message. |
|
What is your discord handle! I'd love to add you to get help on how to run this. What is the compiled executable I put into src/Debug? I have the coreML models. |
|
My Discord is Alexis Ethève#4161. In the src/Debug folder, you need to put the compiled executable from the backends/swift_backend directory. |
|
Thanks for the PR, but i think we should use the coreml backend once it supports multiple sizes, loading of custom models, img2img, inpainting. |
With this implementation, we can select a CoreML model for the Text To Image mode, which increases performance (currently, the model must be manually added to the .diffusonbee/coreml_models folder and the compiled executable in the src/Debug folder).