IAMCCS Cine FLF Engine Simple Dyno

The Cine FLF Engine Simple handles the common case; the Dyno version is for when the common case isn't holding. From the node's own description: "Dyno keeps append_keyframe image locking and also adds legacy guide attention entries for motion-heavy shots." In plain English: it does everything the Simple engine does - reference images encoded and locked into the latent at scheduled frames - and on top of that it injects guide-attention entries into the conditioning, which is the mechanism that keeps a fast-moving shot from drifting off its reference frames mid-motion.

If your action shots keep losing the character between keyframes, this is the engine to reach for.

How it works

Identical interface and inputs to the Simple engine: positive, negative, vae, latent, multi_input (batched reference images), num_images, insert_mode, frame_rate, plus timeline_data (overrides) and manual_keyframes in the frame | ref | strength format. The difference is the second pass: after encoding and locking the guide frames, it calls into ComfyUI's LTX internals to append guide-attention entries to the conditioning at each guide's strength. That's the "dyno" - dynamic guide attention that reinforces the references during sampling rather than only placing latents.

One dependency caveat: it imports _append_guide_attention_entry from ComfyUI's comfy_extras. If your ComfyUI version doesn't expose it (it's an LTX-era internal), the node catches the failure and falls back to the plain engine behavior - the import is wrapped in a try/except and the method degrades gracefully. That means it'll never hard-crash on an old ComfyUI, but on a too-old version you're silently running the Simple engine. Keep ComfyUI current if you want the Dyno feature to actually be there.

Installing it

Part of IAMCCS-nodes:

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

Restart ComfyUI, or install via ComfyUI Manager by searching "IAMCCS". Baseline: ComfyUI ≥ 0.3.0, Python ≥ 3.12, Torch ≥ 2.8.

The honest take

Don't use Dyno everywhere. The extra attention entries add cost and can make motion feel stiffer on calm shots - it's a tool for the shots that need it, not a default. Start with the Simple engine, and when a specific motion-heavy take drifts, swap in Dyno for that branch. The graceful fallback is nice for compatibility but it's also the kind of silent degradation you should know about: the only way to tell whether Dyno actually engaged is the conditioning output or a version check. If you're building a long pipeline, that's a reason to pin your ComfyUI version rather than letting it float.