ComfyUI nodes for ControlNext-SVD v2
These nodes include my wrapper for the original diffusers pipeline, as well as work in progress native ComfyUI implementation. For the diffusers wrapper models should be downloaded automatically, for the native version you can get the unet a/here.
Nodes (6)
Turning the sampler's latents into frames you can actually watch (ControlNext Decode)
The optional scheduler that buys you 10-step ControlNeXt (ControlNext Diffusers Scheduler)
Turning your footage into the skeleton frames ControlNeXt actually wants (ControlNext GetPoses)
The node that actually turns pose frames into video (ControlNext Sampler)
ControlNeXt as a model patch, not a black box (ControlNext SVD Apply)
The one-stop loader that downloads a few gigs and builds the whole pipeline ((Down)Load ControlNeXt)
ComfyUI nodes for ControlNext-SVD v2
These nodes include my wrapper for the original diffusers pipeline, as well as work in progress native ComfyUI implementation.
For the diffusers wrapper models should be downloaded automatically, for the native version you can get the unet here:
https://huggingface.co/Kijai/ControlNeXt-SVD-V2-Comfy/blob/main/controlnext-svd_v2-unet-fp16_converted.safetensors
Diffusers wrapper:
https://github.com/user-attachments/assets/9bc06d4c-b29a-45f5-8a67-211dd5d0f555
ComfyUI native:
https://github.com/user-attachments/assets/7ed23e44-3652-4ccb-8a48-5f0c703ed8b9
Original repo:
https://github.com/dvlab-research/ControlNeXt/tree/main/ControlNeXt-SVD-v2