MOSS Image Color-Depth Control (ControlNet) 🕹️
https://civarchive.com/models/3010285/moss-color-and-depth-control?modelVersionId=3416153
How to use it:
Node: LC MOSS Control 🕹️ in my LC MaskMaker pack (v0.22.1 or newer). Manager, or [Here](https://github.com/lonecatone23/ComfyUI_LC_MaskMaker_nodes)
Needs a ComfyUI with native MOSS Image (newer than 0.41.3).
Wire it between the model and the sampler: model, vae (Flux.2), and a depth_map and / or color_map.
💡 LC Depth Anything V2 (same pack) makes the depth map from any photo. Near = white, which is what this wants.
```
ComfyUI/
└── models/
└── model_patches/
└── MOSS/
└── moss-color-depth-control.safetensors
```
Settings ⚙️
depth_end / color_end: when the map lets go.
1.0 for maps made from a real photo.
3D mannequins and painted maps: color 0.4 - 0.5, and depth 0.15 (0.25 with the 8-step turbo LoRA) or no depth at all.
⚠️ Keep the depth on longer with a mannequin and you get a mannequin. It shows up in the picture, holding hands with your character.
strength: the adapter's LoRA. Leave it at 1.
Color map: flat colors, one per object. The colors are only IDs, black = nothing. Same size as the picture.
Prompt: the map gives the pose and layout, the prompt gives everything else (who, clothes, light, place).
Works on SFT and Base, with or without the 8-step turbo LoRA.
🚀 Works with LC Speed Boost (v0.22.1 and up): the maps follow the picture when it grows.
Credits:
💡 darask0 also made the same adapter for Krea 2 [Here](https://huggingface.co/darask0/Krea-2-color-depth-control) (LC MOSS Control runs the MOSS one only, for now)
Follow me on Instagram [Here](https://www.instagram.com/synth.studio.models/)
Buy me a ☕ [Here](https://ko-fi.com/lonecatone)
If you enjoy it, please 👍 like, 💬 comment, and go give darask0 a ❤️ on Hugging Face 😉