mirror of https://github.com/nod-ai/AMD-SHARK-Studio.git synced 2026-02-19 11:56:43 -05:00

Files

Phaneesh Barwaria 604c0f81b9 Update model_wrappers.py

remove exception

2023-02-24 19:55:48 +05:30

stable_diffusion

Update model_wrappers.py

2023-02-24 19:55:48 +05:30

__init__.py

[SD] Reorganize the stable diffusion model. (#806 )

2023-01-31 14:42:41 -08:00

README.md

[WIP] Add ControlNet to SD pipeline

2023-02-22 15:06:27 +00:00

README.md

[WIP] ControlNet integration:

Since this is dependent on diffusers and currently there's a WIP draft PR, we'd need to clone takuma104/diffusers, checkout controlnet branch and build diffusers as mentioned here.
NOTE: Run pip uninstall diffusers before building it from scratch in the previous step.
Currently we've included ControlNet (Canny feature) as part of our SharkStableDiffusion pipeline.
To test it you'd first need to download control_sd15_canny.pth and pass it through convert_controlnet_to_diffusers.py.
You then need to modify this line to point to the extracted diffusers checkpoint directory.
Download sample image and provide the path here.
Finally run

python apps/stable_diffusion/scripts/txt2img.py --precision=fp32 --device=cuda --prompt="bird" --max_length=64 --import_mlir --enable_stack_trace --no-use_tuned --hf_model_id="runwayml/stable-diffusion-v1-5" --scheduler="DDIM"