IAMCCS Cine Director Guide B

The IAMCCS Shotboard V4 pipeline is a full multi-timeline production system - storyboard, motion plans, audio lanes - and none of it generates a frame until something converts that planning data into actual model conditioning. IAMCCS_CineDirectorGuideB is that conversion point: it's the guide stage that takes shotboard GUIDE_DATA, control frames, and a motion render plan, and produces the positive, negative, latent, and patched model that feed a Wan sampler. If you're working with the pack's Shotboard V4 workflows, this is a stage you'll see in the middle of the graph. If you're not, this node has nothing to consume - it's deeply coupled to the pack's own data types.

What it takes

Required: positive/negative conditioning, a vae, a latent, control_frames (the per-shot image sequence), motion_render_plan_json (the motion plan), plus default_strength (0.75) and default_attention_strength (0.65) that scale how hard the guide influences the result. Optional but load-bearing:

• guide_data (type GUIDE_DATA) - the shotboard's shot metadata. This is the pack-specific handoff from the planner side.
• model - if connected, an IC-LoRA gets applied (via ic_lora_name + ic_lora_strength, and image_attention_strength for how strongly the reference image is attended).
• latent_downscale_factor, crop (disabled/center), use_tiled_encode + tile_size/tile_overlap - resolution and VRAM management for the latent encode.
• scale_by + upscale_method - if the latent or guide needs scaling to match.

Outputs: the edited positive, negative, latent, the optionally patched model, a pass-through latent_downscale_factor, and a report. Wire the conditioning and latent into the sampler and the shot renders.

The honest scope

This is not a beginner node, and it's not a general-purpose conditioning tool. It's a specific stage in the Shotboard V4 / cine_linx authoring model - the class docstring says it keeps IAMCCS's shotboard/motionParts contract while executing the guide-stage job. If your workflow is the pack's cinematic pipeline, this is where the shot becomes samplable. If you grabbed it expecting a generic "apply reference image" node, you'll find the GUIDE_DATA input refuses to cooperate, because the whole design assumes shotboard planning upstream.

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. Model-side you'll need whatever Wan base your Shotboard workflow targets plus the pack's own planner/guide nodes upstream.