LTFloat_Step_00005

LTFloat_Step_00005 is the Latent Tools step-helper whose value widget increments by 0.0005 - a mid-fine step between the 0.0001 and 0.001 siblings. It's a constant: you type a number, it hands you the same number as a FLOAT and a STRING, and the step size tells you the intended use.

And the intended use is the same as every member of this family: sweeping a parameter across a batch of images with the seed held fixed, so each run changes exactly one thing. 0.0005 per image is a good compromise - fine enough that a noise sweep looks smooth, coarse enough that a 100-image run actually covers ground (half a unit of σ over a hundred frames, versus a fifth of a unit at 0.0001). If you want visible-but-gentle change per image without committing to the slowest step in the pack, this is the one.

How it works

No magic, and the honest version matters here: the node does not increment automatically. It returns the value you give it, rounded to five decimals. The step is the widget's spinner increment and the label for the constant. The incrementing is your job - bump the value between runs, change nothing else, compare.

The inputs and outputs

• value - FLOAT, default 0, range ±1,000,000, step 0.0005.
• FLOAT - the value, rounded to five decimals.
• STRING - the value as text. This is the sleeper feature of the whole family: wire it into a filename prefix and every image you generate is labeled with the exact parameter that made it.

Feed the FLOAT into a mean or std input on LTGaussianLatent (or a cfg/denoise input) and the STRING into a Save Image filename, and you've got a self-documenting parameter sweep with no extra bookkeeping.

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.