Qwen2VL Flux ControlNet
Qwen2VL-Flux was released a while ago. It comes with a standalone ControlNet model that works with Flux Dev. Fully compatible with ComfyUI.
Fuente del modelo
- Tipo
- UNet
- Tamaño del archivo
- 6,15 GB
- Versión
- vBF16
- Actualizado
- 7 oct 2026
- Fuente
- Hugging Face
Descripción de la fuente
Qwen2VL-Flux was released a while ago. It comes with a standalone ControlNet model that works with Flux Dev. Fully compatible with ComfyUI.
I works well with Depth and Canny and kinda works with Tile and Realistic Lineart. You can also combine Depth and Canny.
Usually works well with strength 0.6-0.8 depending on the image. You might need to run Flux at FP8 to avoid OOM.
There's a sample workflow in this repo that you can try on ComfyUI.
The original project can be found here:
https://huggingface.co/Djrango/Qwen2vl-Flux
The model in my repo is simply the weights from https://huggingface.co/Djrango/Qwen2vl-Flux/tree/main/controlnet
All credit belongs to the original creator of the model Pengqi Lu.
license: other license_name: flux-1-dev-non-commercial-license license_link: https://huggingface.co/black-forest-labs/FLUX.1-dev/blob/main/LICENSE.md
Fuentes
1 fuente
Fuentes
1 fuenteVerificado 7 oct
Artefactos del modelo
1 artefacto
Artefactos del modelo
1 artefactoqwen2vl_flux_controlnet_bf16.safetensors
safetensors · BF16 (unverified) · 6,15 GB · SHA-256 b2263e803925…0bfc · Hugging Face
Download fileExtractos de fuentes
2 extractos
Extractos de fuentes
2 extractosQwen2VL-Flux was released a while ago. It comes with a standalone ControlNet model that works with Flux Dev. Fully compatible with ComfyUI. I works well with Depth and Canny and kinda works with Tile and Realistic Lineart. You can also combine Depth and Canny. Usually works well with strength 0.6-0.8 depending on the image. You might need to run Flux at FP8 to avoid OOM. There's a sample workflow in this repo that you can try on ComfyUI. The original project can be found here: https://huggingface.co/Djrango/Qwen2vl-Flux The model in my repo is simply the weights from https://huggingface.co/Djrango/Qwen2vl-Flux/tree/main/controlnet All credit belongs to the original creator of the model Pengqi Lu. --- license: other license_name: flux-1-dev-non-commercial-license license_link: https://huggingface.co/black-forest-labs/FLUX.1-dev/blob/main/LICENSE.md ---
Source context: 0 downloads · 21 likes · Repo Nap/Qwen2VL-Flux-ControlNet
