Instructions to prepare environment with conda.txt

Download the models from here
https://huggingface.co/runwayml/stable-diffusion-v1-5/tree/main (normal full)
https://huggingface.co/runwayml/stable-diffusion-v1-5/tree/fp16 (normal half)
https://huggingface.co/runwayml/stable-diffusion-inpainting/tree/main (inpainting full)
https://huggingface.co/runwayml/stable-diffusion-inpainting/tree/fp16 (inpainiting half)


If you need to convert models, use an anaconda powershell promt and the diffusers env.
Then move to the diffusers-rs dir and set the LIBTORCH to the libtorch dir (uncompressed from libtorch-win-shared-with-deps-1.13.0+cu117.zip).

conda activate diffusers

cd C:\Users\Siriux\Workspace\External\diffusers-rs

$Env:LIBTORCH = "C:\Users\Siriux\Workspace\External\libtorch"
$Env:Path += ";C:\Users\Siriux\Workspace\External\libtorch\lib"
$Env:RUST_BACKTRACE=1;
___________

To convert the models first use: python convert_to_npz.py
Then:
cargo run --example tensor-tools cp ./data/clip.npz ./data/clip.ot
cargo run --example tensor-tools cp ./data/vae_v1_5.npz ./data/vae_v1_5.ot
cargo run --example tensor-tools cp ./data/unet_v1_5.npz ./data/unet_v1_5.ot

cargo run --example tensor-tools cp ./data/clip_v1_5_fp16.npz ./data/clip_v1_5_fp16.ot
cargo run --example tensor-tools cp ./data/vae_v1_5_fp16.npz ./data/vae_v1_5_fp16.ot
cargo run --example tensor-tools cp ./data/unet_v1_5_fp16.npz ./data/unet_v1_5_fp16.ot