Lag Chop | Akatz

Lag Chop applies a lag effect to a list of floats: the output value at each frame is pulled toward the input, but only partway, so the signal trails behind the original. Crank it and the output becomes a sluggish, echoey version of the input - a value that keeps drifting toward where the source used to be.

It's a one-knob smoothing filter, really, and it earns its place in the pack's audio toolkit. Feed it a raw, jittery amplitude list from Schedule Audio Framesync and it turns the nervous flicker into a slower, heavier pulse - the difference between a light that stutters with every transient and a light that swells and settles. The same trick applies to any per-frame float stream: brightness scores, keyframe outputs, anything you want to feel weightier.

The name is a nod to "chop" as in the DJ move - the effect is like a manual stutter where the output lags and catches up. It's deliberately simple: one input list, one factor, one output.

How it works

lag_factor controls how much of the new value the output absorbs each frame - the source computes output = output + (input − output) × lag_factor per frame, effectively an exponential moving average. A factor of 1.0 passes the input through instantly (no lag); toward 0 the output barely moves and trails far behind. The node validates the factor must be in (0, 1]; outside that you get an error rather than weird behavior. Note it's a per-element lag over the list position, so "trailing" here means each output element is a blend of its position's input and the accumulated history.

Inputs and output

• float_list - the input FLOAT list.
• lag_factor - 0 to 1, default 0.5. The only knob.

Output is output_list, same length, with the lag applied.

Installing it

Part of akatz-ai/ComfyUI-AKatz-Nodes. Install via ComfyUI Manager (search "AKatz") or:

cd ComfyUI/custom_nodes
git clone https://github.com/akatz-ai/ComfyUI-AKatz-Nodes
cd ComfyUI-AKatz-Nodes
pip install -r requirements.txt

Restart ComfyUI. No model downloads; requirements are numpy, torch, opencv-python and pydub.

Common issues

The trap is dialing lag_factor too low and wondering why your animation barely moves - the output is hugging its history and only creeps toward the input. Start at 0.5 and go down in small steps. Also, lag is an approach, not a delay: there's no hold-and-jump here, so it won't create the kind of discrete echo a true sample-and-hold would. For that flavor, feed a lagged list into Binary Amplitude Gate to get a gated, stuttered on/off pattern that feels like a chop.