6x Entangled Curves [0,1] (Wavelength)

This is the wavelength twin of the node behind the pack's headline workflow - the six-interleaved-prompts example with the AnimateDiff refinement pass, frame interpolation, and upscaling. If you saw that demo and wondered how six prompts could coexist, the answer is this family of nodes, and six is the version the author himself chose to show off.

It's the 6x member of the "Entangled Curves" family in dmarx's ComfyUI-Keyframed, parameterized by wavelength. Six phase-locked sine curves sum to ~1 at every frame. Each swings between ~0 and ~1/3, so the lead prompt carries about a third of the conditioning while the other five hold the blend. The lead rotates through all six prompts over whatever wavelength you set.

How it works

Six SinusoidalCurves with phases a sixth of a turn apart, each with amplitude 1/6 plus a small floor so none fully zeroes - fully zeroed conditions create "sharp edges" in the traversal, per the README. Balanced phases make the six sum to ~1 at every instant. With the default wavelength of 12, one full rotation of the lead across all six prompts takes 12 frames; raise it to 24 or 36 to stretch the cycle and give each prompt a longer moment at the front.

One input, wavelength (FLOAT, default 12, stepped at 0.5). Six KEYFRAMED_CURVE outputs.

Wiring it up

Standard family wiring: each curve output into its own Apply Curve To Conditioning node with its prompt, merge with KfConditioningAdd, then into the sampler. Six weighted conditionings per frame is heavy - the README's showcase pairs it with a refinement pass and VFI, and that's a sensible pattern to copy if your results come out mushy.

Installing it

Part of ComfyUI-Keyframed. ComfyUI Manager: search "ComfyUI-Keyframed". Or:

cd ComfyUI/custom_nodes
git clone https://github.com/dmarx/ComfyUI-Keyframed

Restart ComfyUI. No model downloads; the only dependency is the keyframed library, auto-installed on first startup (first load can pause on pip; pip install keyframed if it fails). It's under keyframed → entangled [0-1]. If you want the exact look from the README's most famous example, this - or its frequency twin - is the node to start from.