Feature Summary
Allow ADetailer to be used with sd-server
Detailed Description
It would be great if ADetailer could work with the sd-server. There already seems to be arguments for the server. But apparently it's not wired into the code to do anything currently so it never actually runs.
Since I plan on running the software with llama-swap to switch models. I would assume that if i run the sd-server with adetailer models active it will just run automatically each time.
Alternatives you considered
I'm not sure if it was not added yet because of it possibly being toggleable option or api request,
But it should work similar to the CLI example:
./build/bin/sd-cli \
-m ~/Downloads/stable-diffusion-webui/models/Stable-diffusion/realisticVisionV60B1_v60B1VAE.safetensors \
-p "a portrait of a woman, detailed face, realistic photography" \
-n "blurry, distorted face, bad anatomy" \
-W 512 \
-H 512 \
--steps 20 \
--cfg-scale 7 \
--backend "diffusion=vulkan1,te=vulkan1,vae=vulkan1" \
--auto-fit on \
--ad-model adetailer/face_yolov8n.safetensors \
--ad-prompt "[PROMPT], detailed realistic face" \
--ad-negative-prompt "blurry, distorted face, bad anatomy" \
--extra-ad-args "confidence=0.3,denoising_strength=0.4,inpaint_width=512,inpaint_height=512" \
-o adetailer/test/full_generation.png
But the server version currently seems unused:
./build/bin/sd-server \
--listen-ip 127.0.0.1 \
--listen-port 1234 \
-m ~/Downloads/stable-diffusion-webui/models/Stable-diffusion/realisticVisionV60B1_v60B1VAE.safetensors \
--backend "diffusion=vulkan1,te=vulkan1,vae=vulkan1" \
--auto-fit on \
-W 512 \
-H 512 \
--steps 20 \
--cfg-scale 7 \
--ad-model ~/Downloads/stable-diffusion.cpp/adetailer/face_yolov8n.safetensors \
--ad-prompt "[PROMPT], detailed realistic face" \
--ad-negative-prompt "blurry, distorted face, bad anatomy" \
--extra-ad-args "confidence=0.3,denoising_strength=0.4,inpaint_width=512,inpaint_height=512"
I did have an AI check the code wondering why it did not work. But apparently it's just not in the server at all even though the arguments are already there.
Feature Summary
Allow ADetailer to be used with sd-server
Detailed Description
It would be great if ADetailer could work with the sd-server. There already seems to be arguments for the server. But apparently it's not wired into the code to do anything currently so it never actually runs.
Since I plan on running the software with llama-swap to switch models. I would assume that if i run the sd-server with adetailer models active it will just run automatically each time.
Alternatives you considered
I'm not sure if it was not added yet because of it possibly being toggleable option or api request,
But it should work similar to the CLI example:
But the server version currently seems unused:
I did have an AI check the code wondering why it did not work. But apparently it's just not in the server at all even though the arguments are already there.