IAMCCS Cine Shotboard Timeline Pro

Timeline Pro is the original all-in-one shotboard, and it's still the best way to understand what this whole Cine family is doing. The docstring sums it up: a "filmmaker-facing shotboard that drives FLF guides and PromptRelay timing from one table." Where the later Pro planners are planner-only (they hand a plan to a backend), Timeline Pro is the whole single-generation package - it takes your conditioning, your VAE, and your latent, plus the shot table, and returns the conditioned outputs ready for sampling. If you want one node to see the whole concept working, this is it.

How it works. timeline_data holds the rows the UI edits - time, image reference, force, label, camera move, transition, note, guide enable. From those, the node builds two things simultaneously: FLF guide images to inject at their timestamps, and per-segment local prompts that switch at segment boundaries (that's the PromptRelay timing). global_prompt is the through-line; local_prompts and segment_lengths are the per-beat text. The output spread hands you everything at once.

The inputs that matter:

• guide_policy - safe_core_guides (default) fires only a few well-spaced guides; every_checked_row and prompt_only are the alternatives.
• min_guide_gap_seconds (default 1.75) and max_guides (default 5) - the spacing and cap that keep guides from overwhelming the model.
• default_force (0.22) - per-row guide strength fallback.
• promptrelay_epsilon (0.001) - the timing tolerance for the per-segment prompt switches.
• ltx_round_mode - up_8n_plus_1 by default, which rounds your frame count up to LTX's 8n+1 requirement.
• tail_safety_frames (default 1) - extra tail margin, the same family of idea as the Tail Trim Policy node.

You feed positive/negative conditioning, a vae, and a latent in; reference images come from image_paths or the optional multi_input (the tooltip notes that if image_paths is empty, it uses the connected multi_input from the Cine Reference Board).

The outputs. positive and negative (updated conditioning with the guides baked in), latent, plus the planning spread you can wire anywhere: flf_timeline, global_prompt, local_prompts, segment_lengths, max_frames, promptrelay_epsilon, report, multi_output, image_1, duration_seconds_int, and frame_rate_int. Most graphs only need the first three; the rest are there so you don't lose visibility into what the planner decided.

Install. Pack-wide: ComfyUI Manager (search "IAMCCS"), or git clone https://github.com/IAMCCS/IAMCCS-nodes into custom_nodes, restart. ComfyUI ≥ 0.3.0, Python ≥ 3.12, no bundled models - the conditioning/latent path needs your LTX-2.3 stack (model, video VAE, audio VAE if you're doing audio).

Gotchas. This node computes conditioning and latent and emits the plan - which is exactly the graph-cycle problem the later Planner Pro was split out to avoid. If you find yourself needing the plan upstream of a PromptRelay node, use Planner Pro instead. Otherwise the usual house rules apply: keep the pack updated, and remember the 8n+1 tail - the clip comes back a few frames longer than your duration_seconds, which is normal until a tail-trim step handles it.