IAMCCS Cine Shotboard Lite
The on-ramp to IAMCCS's shotboards — FLF guides, none of the PromptRelay drama
- multi_input
- cine_linx
If you've ever stared at an LTX-2.3 workflow and wished you could just tell the thing "here's frame one, here's where the camera lands, here's the beat at 4 seconds," this is the node you're looking for. IAMCCS Cine Shotboard Lite is the simplest member of the pack's shotboard family: it turns a table of timeline rows into timed FLF (first-frame/last-frame) image guides, and it deliberately leaves out the fancy stuff - no PromptRelay relay prompts, no camera-move columns, no multi-track audio. That omission is the feature. It's the node to start with before you climb the ladder to the Pro planners.
How it works. You give it a global_prompt (the through-line for the whole shot) and a timeline_data table. Each row carries a time, an image reference, a force strength, a label and a note - that's it. The node expands those rows into guide frames that get injected into the generation at their timestamps, at the strength each row specifies. The whole result is packed into a single cine_linx output, which is the pack's LINX contract: a dict that carries all the plan forward so a backend node can actually execute it. You don't hand-wire a dozen strings around anymore; one socket in, backend out.
The inputs that matter. A beginner can get a lot done with three of them:
timeline_data- edited by the Lite UI. Rows contain time, image reference, force, label and note. This is your shot list.duration_secondsandframe_rate- defaults 8 and 24. These anchor the timeline.default_force- the guide strength applied to rows that don't set their own (0.22 default, 0–1). Too high and the model gets glued to the reference frames; too low and it wanders.
guide_policy has two options: every_checked_row (default) fires every checked row as a guide, while safe_core_guides only uses the ones the planner considers safe. image_paths lets you point at internal reference images, and you can also feed a batch straight in through the optional multi_input. Output is the single cine_linx, which you wire into the Cine FLF backend path (or CineInfo for inspection).
Install. Grab the whole pack once, either from ComfyUI Manager (search "IAMCCS") or:
cd ComfyUI/custom_nodes
git clone https://github.com/IAMCCS/IAMCCS-nodes
Restart ComfyUI. No model files ship with the pack - the Cine path expects an LTX-2.3 setup (model, video VAE, audio VAE) and ComfyUI ≥ 0.3.0 with Python ≥ 3.12. The Lite node itself is pure planning: no heavy dependencies beyond what the pack already needs.
Where people get burned. Remember LTX needs frame counts of 8n+1, so 8 seconds at 24fps isn't 192 frames - the pipeline rounds up for you, and a couple of padded frames show up at the tail. That's expected, not a bug. Also, this is a planner node, not a generator: if you wire cine_linx into nothing, you get a dict and a shrug. And keep the pack updated - the author ships fast, and stale saved workflows occasionally show old widget ordering after an update. If your node looks scrambled, delete and re-add it.
Inputs (12)
| Name | Type | Default | Description |
|---|---|---|---|
| global_prompt | STRING | one continuous cinematic FLF shot with coherent motion through the image guides | — |
| timeline_data | STRING | Edited by the Cine Shotboard Lite UI. FLF-only rows contain time, image reference, force, label and note. | |
| duration_seconds | FLOAT | 8.000.01–36000 | — |
| frame_rate | INT | 241–120 | — |
| guide_policy | COMBO | every_checked_row | 2 options: every_checked_row, safe_core_guides |
| min_guide_gap_seconds | FLOAT | 0.000–60 | — |
| max_guides | INT | 121–50 | — |
| default_force | FLOAT | 0.220–1 | — |
| image_paths | STRING | Internal reference images for the Lite shotboard. | |
| image_width | INT | 76864–8192 | — |
| image_height | INT | 43264–8192 | — |
| multi_inputopt | IMAGE | — |
Outputs (1)
| Name | Type | Description |
|---|---|---|
| cine_linx | IAMCCS_SUPERNODE_LINX | — |