AI Image Editing
https://spacecruft.org/deepcrayon/image-edit
- Python 100%
| firered | ||
| gradio | ||
| .gitignore | ||
| LICENSE-apache.txt | ||
| README.md | ||
Image Edit
Editing images with AI.
Just use ComfyUI with FireRed-Image-Edit 1.1.
Ignore below....
XXX
Right now, using gradio with FireRed-Image-Edit-1.1 served up by sglang.
Web Interface
There are many options for image editing. Quick, dirty, and easy is gradio.
Install
Thusly:
git clone https://spacecruft.org/deepcrayon/image-edit
cd image-edit/gradio
python -m venv venv
source venv/bin/activate
pip install -U setuptools pip wheel
pip install -r requirements.txt
deactivate
sudo cp gradio.service /etc/systemd/system/
sudo systemctl daemon-reload
sudo systemctl start gradio.service
sudo systemctl status gradio.service
sudo systemctl enable gradio.service
Inference
There are a lot of edit models too. Right now FireRed-Image-Edit-1.1 is one of the best.
Install
# git clone https://spacecruft.org/deepcrayon/image-edit
# cd image-edit/firered
cd ../firered
# For B200 GPU
export TORCH_CUDA_ARCH_LIST="10.0"
python -m venv venv
source venv/bin/activate
pip install -U setuptools pip wheel
pip install "sglang[all] @ git+https://github.com/sgl-project/sglang.git@64950d8f9735e0e0f652092b695d21e3ca34f310#subdirectory=python"
pip install accelerate
pip install --no-build-isolation flash-attn
pip install nvidia-cudnn-cu12==9.16.0.29
# update registry for FireRed model:
sed -i 's|hf_model_paths=\["Qwen/Qwen-Image-Edit-2509"\]|hf_model_paths=["Qwen/Qwen-Image-Edit-2509", "FireRedTeam/FireRed-Image-Edit-1.1"]|' ./venv/lib/python3.13/site-packages/sglang/multimodal_gen/registry.py
# Patch a bug with HEAD
sed -i '/^ setattr(encoder_config\.arch_config, key, value)$/a\ for _prop in ("image_token_id", "video_token_id", "vision_start_token_id", "vision_end_token_id", "vision_token_id"):\
if hasattr(diffusers_pretrained_config, _prop):\
setattr(encoder_config.arch_config, _prop, getattr(diffusers_pretrained_config, _prop))' \
./venv/lib/python3.13/site-packages/sglang/multimodal_gen/runtime/loader/component_loaders/text_encoder_loader.py
sudo mkdir -p /etc/sglang/
sudo cp firered_cache_dit_config.yaml /etc/sglang/firered_cache_dit_config.yaml
sudo cp sglang-firered-image-edit-1-1.service
sudo systemctl daemon-reload
sudo systemctl start sglang-firered-image-edit-1-1.service
sudo systemctl status sglang-firered-image-edit-1-1.service
sudo systemctl enable sglang-firered-image-edit-1-1.service
Prompt
Add something like:
Ultra sharp, 4K quality.
Web Proxy
Set up via apache. Some config snippets:
# 1. Redirect /gradio to /gradio/ (trailing slash)
RewriteRule ^/gradio$ /gradio/ [R=301,L]
# 2. WebSocket proxy
RewriteCond %{HTTP:Upgrade} =websocket [NC]
RewriteRule ^/gradio/(.*) ws://10.68.68.249:7860/$1 [P,L]
# 3. HTTP proxy
ProxyPass /gradio/ http://10.68.68.249:7860/
ProxyPassReverse /gradio/ http://10.68.68.249:7860/
# Block cloudflare serving (meh, still does DNS):
LoadModule substitute_module modules/mod_substitute.so
<Location /gradio>
AddOutputFilterByType SUBSTITUTE text/html
Substitute "s|https://cdnjs.cloudflare.com/ajax/libs/iframe-resizer/4.3.1/iframeResizer.contentWindow.min.js|data:text/javascript,|i"
</Location>
RequestHeader set X-Forwarded-Proto "https"
RequestHeader set X-Forwarded-Host "proxy.deepcrayon.org"
License
Apache 2.0.
Copyright © 2026 Jeff Moe.