Image Gaussian Blur

Exactly what it says: a Gaussian blur with a single strength dial. There's no mystery here and no cleverness - it softens an image, and the amount is one number. It ships inside RES4LYF because a sampler pack this involved ends up needing small image-prep helpers, and rather than send you off to another pack, it bundles its own.

Where it earns its place is on the plumbing, not the final render. Blurring a mask before it drives a region gives you a feathered edge instead of a hard seam. Blurring a guide or reference image knocks out fine detail so a composition guide steers layout without dragging texture along with it. Blurring a source before you composite it softens a hard cut. It's the kind of node you don't think about until a mask edge is showing and you realize you needed to feather it.

How it works

It's a standard separable Gaussian convolution over the image tensor. Larger radius, more spread, softer result. Because it's a real blur and not a downscale-upscale trick, it degrades gracefully - you can push it hard without the blocky artifacts a cheap approximation would give you.

The inputs that matter

Only two, and one of them is the image:

• images (IMAGE) - what you're blurring. It'll take a batch.
• size (INT, default 6) - the blur radius. This is the whole control. 6 is a gentle softening; push it into the double digits for a heavy, dreamy smear. There's no upper cap in practice, so if the image goes to mush you've gone too far - back it off.

The single output is image (IMAGE).

Installing it

ComfyUI Manager: search RES4LYF, install, restart. Or by hand:

cd ComfyUI/custom_nodes
git clone https://github.com/ClownsharkBatwing/RES4LYF/
cd RES4LYF
pip install -r requirements.txt

Run pip install in your venv or the portable Python, restart, hard-refresh (F5). No downloads for this one.

Where people get tripped up

The main gotcha is scale. A size of 6 does something very different to a 512px thumbnail than to a 4K render - blur radius is in pixels, so the same number reads as heavy on a small image and barely-there on a big one. If you're switching resolutions, expect to re-tune size rather than reuse it.

Second, remember the type on the output is IMAGE. If you want to blur a mask, you'll usually convert the mask to an image, blur it, and convert back - feeding a mask-typed input a blurred image only works where the downstream node accepts an image as a mask source. And if the node's missing from your list, that's not a blur problem, that's RES4LYF not being installed.