IAMCCS BoardMaker Dialogue Foley
Storyboard a two-person dialogue scene with the sound design built into the plan
- cine_linx
- cine_linx
The BoardMaker got a dialogue-and-foley sibling, and it's the one to reach for when your shot is two people talking. IAMCCS_BoardMaker_DialogueFoley builds the same kind of beat board, but each beat carries dialogue - who's speaking, for how long, what they say, how they sound - plus the foley bed underneath. It's a plan for an audio-driven scene, not just a visual one, which fits the LTX-2 era of audio-guided generation nicely.
The heart is dialogue_data, another JSON blob the UI edits, but the schema is richer. Each line has: speaker (A or B), seconds, ref (which reference image for that speaker), label, framing (medium close-up, reverse angle…), dialogue (the actual line, e.g. the pack's sample Italian thriller dialogue), voice (direction like "low tense voice, almost whispered"), local_prompt, and foley (the sound under that beat: breathing, cloth rustle, room tone). The default ships a four-line over-the-shoulder scene so you can see how speaker, framing and ref interleave.
Around it, the scene-level inputs:
global_prompt,scene_context(the situation the whole scene lives in),foley_prompt(the master foley direction - note the default ends with "no music", which is the pack being explicit).frame_rate,image_width,image_height- default 1280x736, a taller canvas than the plain BoardMaker's.default_force(0.65),guide_policy(every_checked_row/safe_core_guides/prompt_only).woosh_chunk_seconds(8) andwoosh_overlap_seconds(1) - chunking settings for generating the audio/visual bed in pieces with a little overlap, the pack's standard approach for long content.
Output: a single cine_linx (the pack's supernode connector), which carries the whole board - dialogue, foley, framing, timing - into the downstream Cine nodes that turn it into conditioning and generation. That's the "one connector, one coherent plan" philosophy: you get everything you need in one object.
Why it's worth knowing: dialogue scenes are where chunked generation falls apart - two speakers, cut points, eyelines, and audio sync all need to stay aligned, and doing that with ad-hoc nodes is miserable. Baking speaker, ref, dialogue, and foley into one timed plan is genuinely the right structure, and the sample data teaches the schema better than any doc. It pairs with IAMCCS_CineAudioInfo and IAMCCS_CineAudioPromptDirector on the generation side.
Installing: part of IAMCCS/IAMCCS-nodes - ComfyUI Manager → search "IAMCCS", or cd ComfyUI/custom_nodes && git clone https://github.com/IAMCCS/IAMCCS-nodes.git, restart. No models, no extra deps.
The honest caveat: it's a planning node - the cine_linx output is data, not frames or audio. You need the Cine/Shotboard execution side to consume it, and the execution side of IAMCCS's dialogue pipeline is where the fiddly parts live. And keep your ref numbers pointing at actual reference images you've loaded, or the scene will swap faces between lines - that's the #1 beginner failure with ref-indexed dialogue boards.
Inputs (13)
| Name | Type | Default | Description |
|---|---|---|---|
| board_name | STRING | dialogue_foley_board | — |
| global_prompt | STRING | Two men in a tense field and reverse-field dialogue scene, stable identity, coherent eyelines, cinematic realism, restrained camera movement. | — |
| scene_context | STRING | A quiet interior at night. The two men speak in low voices while something unseen waits beyond the room. | — |
| foley_prompt | STRING | realistic cinematic foley, tense room tone, male breathing, cloth movement, subtle chair creaks, distant door pressure, no music | — |
| dialogue_data | STRING | { "lines": [ { "speaker": "A", "seconds": 4.0, "ref": 1, "label": "campo_A", "framing": "medium close-up", "dialogue": "Non guardare la finestra. Se capisce che l'abbiamo visto, siamo finiti.", "voice": "low tense voice, almost whispered", "local_prompt": "man A in field shot, tense medium close-up, restrained fear, coherent eyeline", "foley": "soft breathing, cloth rustle, distant room tone" }, { "speaker": "B", "seconds": 3.5, "ref": 2, "label": "controcampo_B", "framing": "reverse angle", "dialogue": "L'ha gia capito. Sta aspettando che uno di noi faccia il primo passo.", "voice": "controlled low voice, guarded", "local_prompt": "man B in reverse field shot, listening then answering, controlled fear", "foley": "small chair creak, quiet breath, low room pressure" }, { "speaker": "A", "seconds": 3.5, "ref": 1, "label": "ritorno_A", "framing": "close-up", "dialogue": "Allora il primo passo lo fai tu, lentamente, verso l'uscita.", "voice": "firm whisper, urgent but contained", "local_prompt": "return to man A close-up, slight push-in, decision in the eyes", "foley": "subtle foot shift, jacket movement, controlled breath" }, { "speaker": "B", "seconds": 4.0, "ref": 2, "label": "chiusura_B", "framing": "reverse close-up", "dialogue": "No. Se esco da solo, tu resti qui a spiegargli dove siamo.", "voice": "dry, bitter, quietly direct", "local_prompt": "man B reverse close-up, hard pause before the last words, tense eye line", "foley": "distant door tension, faint wood creak, breath held" } ] } | Edited by the IAMCCS Dialogue/Foley BoardMaker UI. JSON lines: speaker, seconds, ref, label, dialogue, voice, local_prompt, foley. |
| frame_rate | FLOAT | 241–120 | — |
| image_width | INT | 128064–8192 | — |
| image_height | INT | 73664–8192 | — |
| default_force | FLOAT | 0.650–1 | — |
| guide_policy | COMBO | every_checked_row | 3 options: every_checked_row, safe_core_guides, prompt_only |
| woosh_chunk_seconds | FLOAT | 8.01–8 | — |
| woosh_overlap_seconds | FLOAT | 1.00–4 | — |
| cine_linxopt | IAMCCS_SUPERNODE_LINX | — |
Outputs (1)
| Name | Type | Description |
|---|---|---|
| cine_linx | IAMCCS_SUPERNODE_LINX | — |