LTFloat_Step_002

LTFloat_Step_002 is the 0.02 step in the Latent Tools helper family - coarse enough to get through a sweep without babysitting it, fine enough that you can still see the relationship between parameter and output. It's a calibrated constant: type a value, get the same number back as a FLOAT and a STRING, with the widget stepping by 0.02.

The family's job is parameter sweeps with a fixed seed. At 0.02 per image, a 50-run batch sweeps a whole unit - that's a genuinely useful range for σ, μ, or even a denoise sweep. Consecutive outputs will look visibly different, which is fine: this is the "explore the space" step, not the "render a smooth animation" step. When you find something interesting, drop to a finer sibling and home in.

How it works

Be clear about the mechanism so the name doesn't mislead you: the node does not auto-increment. It takes the value you give it, rounds to five decimals, and returns it as a FLOAT and a STRING. The step is the widget's spinner increment and the label. You bump the value between runs; the node keeps it consistent.

The inputs and outputs

• value - FLOAT, default 0, range ±1,000,000, step 0.02.
• FLOAT - the value.
• STRING - the value as text, ready for a filename prefix so every output is labeled with its parameter.

Typical wiring: FLOAT into a mean/std/cfg input on the relevant node, STRING into a Save Image filename, seed locked, one variable changed per run.

Installing it

Part of xl0's Latent Tools pack. ComfyUI Manager → search Latent Tools → install → restart, or:

cd ComfyUI/custom_nodes
git clone https://github.com/xl0/latent-tools

Restart, and it's under LatentTools. The pack's only dependency is lovely-tensors (Manager installs it; manual cloners may need pip install lovely-tensors). No model downloads.