ComfyUI-BCVideoNodes

Video nodes for ComfyUI: a Wan Animate preprocess built from small nodes that are usable in any video pipeline (wholebody pose with ViTPose-H or Sapiens2, SAM 3.1 person tracking, face crops, pose and mask checks), a SCAIL-2 preprocess, three samplers that turn a reference image plus a driving video of any length into a Wan Animate or SCAIL-2 video of exactly that length, and the video nodes around them: load a video at the model's generation size, fit a reference image to it, save the result, compare two videos in the node.

Node	Id	Category
Pose Detection	BCVPoseDetection	BCVideoNodes
Pose Config	BCVPoseConfig	BCVideoNodes
SAM 3.1 Multiplex Video Track	BCVSAM3VideoTrack	BCVideoNodes
SAM 3.1 Multiplex Config	BCVSAM3Config	BCVideoNodes
Face Crop	BCVFaceCrop	BCVideoNodes
Pose Guard	BCVPoseGuard	BCVideoNodes
Mask Guard	BCVMaskGuard	BCVideoNodes
WanAnimate Preprocess	BCVWanAnimatePreprocess	BCVideoNodes/Wan/Animate
WanAnimate Preprocess Guard	BCVWanAnimatePreprocessGuard	BCVideoNodes/Wan/Animate
Wan Animate Long Video Sampler	BCVWanAnimateLongVideoSampler	BCVideoNodes/Wan/Animate
Wan Animate 2 Long Video Sampler	BCVWanAnimate2LongVideoSampler	BCVideoNodes/Wan/Animate
SCAIL-2 Colored Mask	BCVSCAIL2ColoredMask	BCVideoNodes/SCAIL
SCAIL-2 Preprocess	BCVSCAIL2Preprocess	BCVideoNodes/SCAIL
SCAIL-2 Preprocess Guard	BCVSCAIL2PreprocessGuard	BCVideoNodes/SCAIL
SCAIL-2 Long Video Sampler	BCVSCAIL2LongVideoSampler	BCVideoNodes/SCAIL
Load Video	BCVLoadVideo	BCVideoNodes/Video
Get Video Info	BCVGetVideoInfo	BCVideoNodes/Video
Load Reference Image	BCVLoadReferenceImage	BCVideoNodes/Video
Conform Video	BCVConformVideo	BCVideoNodes/Video
Save Video	BCVSaveVideo	BCVideoNodes/Video
Video Comparer	BCVVideoComparer	BCVideoNodes/Video

Install

Clone into ComfyUI/custom_nodes/, install the requirements into ComfyUI's Python and restart:

pip install -r requirements.txt

The only runtime dependency beyond ComfyUI is opencv-python; torch, numpy, scipy, safetensors, tqdm and PyAV (av, the video nodes' decoder and encoder) come with ComfyUI, so the video nodes add no requirement. The previews are plain JavaScript in web/js/, which ComfyUI serves; nothing to build. onnx and huggingface_hub are needed only for the offline conversion and upload scripts (pip install .[dev]).

Documentation

Video nodes

• Video nodes: Load Video, Get Video Info, Load Reference Image, Conform Video, Save Video, Video Comparer, and the player in the node.

Preprocess nodes

Feed them frames already at the generation size (Load Video loads them so); the pose images, masks and boxes come out at the size of the frames that went in, the pose images at Pose Detection's width x height when both are connected.

• Pose and Face Crop: Pose Detection, Pose Config, Face Crop.
• SAM 3.1 Multiplex: SAM 3.1 Multiplex Video Track and its modes, SAM 3.1 Multiplex Config, and the input precedence of the preprocess nodes.
• Guards: Pose Guard and Mask Guard, their fails, warnings and metrics, and the reference check.
• WanAnimate Preprocess: WanAnimate Preprocess and WanAnimate Preprocess Guard.
• SCAIL-2 Preprocess: SCAIL-2 Colored Mask, SCAIL-2 Preprocess, SCAIL-2 Preprocess Guard.

Long video samplers

• Long video samplers: the three samplers, their wiring and defaults, frames_per_chunk by VRAM, the length math, tail padding and the color anchor.

Reference

• Unused outputs: which outputs come out empty when nothing is connected to them.
• Measured against the earlier workflow: time, RAM and VRAM of one Wan 2.2 Animate replacement workflow, this pack against the earlier packs.
• Models: what is downloaded, from where, into which folder.
• Development: the tests.

Roadmap

• SCAIL-2 pose-driven mode: not planned. The pack runs SCAIL-2's end-to-end mode: the raw driving video is the pose input. The pose-driven mode drives the model with a rendered 3D skeleton instead. zai-org/SCAIL-Pose marks end-to-end "(Recommended)", "More accurate and easier than pose-driven for most cases", and keeps pose-driven for extremely challenging inputs. It would be added only if a need for it appears.
• Multi-person (all nodes): multi-person pose and mask are on the roadmap, to be implemented and tested later. Today everything is optimised for one person; future multi-person changes will take effect in multi-person mode only.

Licences

The code is MIT (LICENSE), except the files vendored from the Alibaba Wan team's WanAnimate preprocess, which are Apache-2.0 and keep their copyright header (libs/pose_utils/LICENSE):

• libs/pose_utils/pose2d_utils.py
• libs/pose_utils/human_visualization.py
• pipelines/face.py
• models/common/wrapper.py
• models/vitpose/wrapper.py
• models/yolo/wrapper.py
• models/vitpose/decode.py

The model weights keep their own licences:

Model	File	Licence
ViTPose-H wholebody	vitpose_h_wholebody_fp16.safetensors	Apache-2.0
YOLOv10x	yolov10x_fp32.safetensors	AGPL-3.0
SAM 3.1	sam3.1_multiplex_fp16.safetensors	Meta's SAM License
Sapiens2 pose	sapiens2_pose_*.safetensors	Sapiens2 License (Meta)

The person detector's weights (YOLOv10x) are AGPL-3.0. Running them locally is unaffected; offering a service over a network that runs them (a hosted workflow, a SaaS) brings AGPL-3.0's network clause into play, which requires making the corresponding source available to that service's users. With bboxes connected, Pose Detection neither downloads, loads nor runs the detector.

SAM 3.1 is fetched by ComfyUI from Comfy-Org/sam3.1 under Meta's SAM License; this package does not redistribute it.

The Sapiens2 pose weights are Meta's, under the Sapiens2 License; the converted files are fetched from beycanai/sapiens2-convrot, whose LICENSE.md carries it.