IAMCCS CineInfo V2V Backend Router

Video-to-video is a different beast from planning a fresh LTX shot: you start from source footage and drive a backend with poses, faces, references, and masks. IAMCCS CineInfo V2V Backend Router is the bridge between the Cine shotboard world and that V2V world. It takes a cine_linx plan, collects whatever media you can supply - source images, audio, pose videos, face videos, masks - and packs it into three typed buses that the pack's V2V backends consume: media_bus, control_bus, and preview_bus. It's the node you reach for when a shotboard-planned sequence has to survive contact with a pose-transfer or face-swap backend.

The pack's V2V/LTX-2.3 backend work is one of its newest areas - the README notes the LTX-2.3 nodes for v2v and audio+img2vid, with instructions hosted on the author's Patreon. This router is the entry point for that pipeline.

How it works

It reads the cine_linx payload, keys off its backend_mode (the plan says whether this is pose transfer, face-driven, etc.), and routes your optional media into the matching bus. For example, in pose_transfer mode it wires up the pose source image/video paths from the payload and keeps the media bus coherent. The nine optional inputs give you the raw material:

• source_images / source_audio / result_image - the base media.
• pose_video_mask_image / reference_image_mask_image / character_mask - mask and reference inputs for controlled transfer.
• face_video / pose_video / background_video - the drive signals for face and pose backends.

Everything it can't source from the payload falls back to what you connect; anything left unconnected just isn't in the bus. The report output tells you what mode was detected and what got routed.

Outputs

• media_bus (IAMCCS_V2V_MEDIA_BUS) - base media for the backend.
• control_bus (IAMCCS_V2V_CONTROL_BUS) - pose/face/mask control signals.
• preview_bus (IAMCCS_V2V_PREVIEW_BUS) - preview media for UI/review.
• report - what got routed.

These are custom bus types, so they only plug into other IAMCCS V2V nodes - you can't feed them to arbitrary ComfyUI nodes.

Installing it

Part of IAMCCS-nodes:

cd ComfyUI/custom_nodes
git clone https://github.com/IAMCCS/IAMCCS-nodes.git

Restart ComfyUI, or install via ComfyUI Manager by searching "IAMCCS". Baseline: ComfyUI ≥ 0.3.0, Python ≥ 3.12, Torch ≥ 2.8. The V2V backends themselves use models you load separately; the router carries no weights.

The honest take

This is the most "ecosystem-locked" node of the twenty: it's only useful inside the IAMCCS V2V pipeline, and the newer LTX-2.3 V2V backend instructions are Patreon-gated rather than in the README. If you're not running that pipeline, skip it. If you are, the thing to know is that backend_mode in the cine_linx decides what the router does with your inputs - so check the report to see which mode was detected before you assume your face_video ended up where you think. The typed buses are the pack's way of keeping the graph tidy, but they're also the reason this node won't compose with the broader ComfyUI ecosystem. Plan accordingly.