IAMCCS Cine Shotboard V4 Guide

IAMCCS_CineDirectorGuideB is the guide stage that runs the shot data through IC-LoRA and motion parts. IAMCCS_CineShotboardV4Guide is the other face of the same job - it's the Shotboard V4 guide stage proper, built around the core GUIDE_DATA input, with an optional MOTION_GUIDE_DATA lane on top. If DirectorGuideB is the "director-compatible" variant, this is the plain-V4 one: shotboard guide data in, sampler-ready conditioning out.

What it takes

Required: positive/negative, a vae, a latent, and guide_data (type GUIDE_DATA) - the shot metadata from the V4 planner. Optional but where the action is:

• motion_guide_data (type MOTION_GUIDE_DATA) - the motion lane. This is the input DirectorGuideB doesn't have; it's what makes this the fuller V4 guide.
• model + ic_lora_name/ic_lora_strength + image_attention_strength - attach an IC-LoRA for reference-image identity and control how hard the sampler attends to it.
• auto_snap_ic_grid (default true) - snaps the reference-image grid to the guide's expected layout automatically. Turn it off if your grid is already exact.
• retake_mode - marks this as a retake so the guide can behave differently for regenerated shots.
• latent_downscale_factor, scale_by/upscale_method, crop, and the use_tiled_encode/tile_size/tile_overlap group - resolution and VRAM handling for the encode.

Outputs: positive, negative, latent, the patched model, and latent_downscale_factor. Same shape as the DirectorGuideB, minus the report string.

Why two guide stages?

The Shotboard V4 backend ships a family of guide variants for the same contract - plain V4, director-compatible, motion-parts. If you load a workflow and see one vs the other, it usually reflects which generation of the pipeline the graph was built against, not a meaningful quality difference. For fresh builds, pick whichever your planner's output pairs with; the GUIDE_DATA/MOTION_GUIDE_DATA types are the pack's own, so the pairings are fixed by the pipeline, not free choice.

Install

Part of IAMCCS-nodes:

• ComfyUI Manager → search "IAMCCS" → install → restart.
• Or:

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

Restart ComfyUI. Requirements: ComfyUI ≥ 0.3.0, Python ≥ 3.12, Torch ≥ 2.8. Deeply tied to the pack's Shotboard V4 planner nodes upstream - it won't do much in a standalone graph.