SD3 ControlNet Canny
5.0
0 reviewsDescription
Hi everyone, ControlNet for SD3 is available on Comfy UI! Please read the instructions below:
1- In order to use the native 'ControlNetApplySD3' node, you need to have the latest Comfy UI, so update your Comfy UI.
2- Right now, there is 3 known ControlNet models, created by Instant-X team: Canny, Pose and Tile. This example is for Canny, but you can use the others as well.
Canny: https://huggingface.co/InstantX/SD3-Controlnet-Canny/blob/main/diffusion_pytorch_model.safetensors
Pose: https://huggingface.co/InstantX/SD3-Controlnet-Pose/blob/main/diffusion_pytorch_model.safetensors
Tile: https://huggingface.co/InstantX/SD3-Controlnet-Tile/blob/main/diffusion_pytorch_model.safetensors
Thanks to the ComfyOrg & Instant-X teams!
Important Note: Make sure you are using 1024x1024 latent. Other resolutions may generate a lot of artifcats. Here is the explanation from the Instant-X Team:
Due to the fact that only 1024*1024 pixel resolution was used during the training phase, the inference performs best at this size, with other sizes yielding suboptimal results. We will initiate multi-resolution training in the future, and at that time, we will open-source the new weights.
P.s: you might need to adjust your 'strength' and 'threshold' parameters for getting the best results.
Discussion
(No comments yet)
Loading...
Reviews
No reviews yet
Versions (1)
- latest (a year ago)
Node Details
Primitive Nodes (1)
ControlNetApplySD3 (1)
Custom Nodes (12)
ComfyUI
- CheckpointLoaderSimple (1)
- CLIPTextEncode (2)
- ControlNetLoader (1)
- Canny (1)
- PreviewImage (2)
- VAEDecode (1)
- LoadImage (1)
- KSampler (1)
- SDXLEmptyLatentSizePicker+ (1)
- Image Save (1)
Model Details
Checkpoints (1)
sd3_medium_incl_clips_t5xxlfp8.safetensors
LoRAs (0)