Image Amplify

Ever run an image comparison and gotten back a difference map that's just... black? That's not a bug. It's the whole reason this node exists.

Difference maps are floats where "identical" is 0.0 and "visibly different" is often 0.03. When you display that as an image, everything rounds to black and you can't tell a tiny VAE drift from a big change. Image Amplify fixes that the dumbest possible way: it multiplies every pixel by an integer (default 8, up to 64) and clips to [0, 1]. That's the entire node - in the source it's literally one line of torch: torch.clamp(image * float(scale), 0.0, 1.0).

Why it's in this pack

Image Amplify ships in easygoing0114/ComfyUI-easygoing-nodes, the "Easygoing Nodes" pack - a grab-bag of practical workflow improvements from a Japanese ComfyUI author (model-loading helpers, prompt metadata embedding, HDR LAB color, image difference analysis). Image Amplify is the designated sidekick to the pack's Image Difference Checker: that node outputs color and brightness difference maps whose values cluster near zero, and this node is what you chain after it to make those faint signals actually legible. The README's own sample workflow wires them together, and it's the natural pair for evaluating VAE reconstruction accuracy or A/B-ing an upscaler.

The two inputs, and the honest caveat

The info_schema is refreshingly small - this is not a node you'll fiddle with:

• image - the IMAGE to amplify (any near-black analysis map, not just the difference checker's).
• scale - the integer multiplier, default 8, min 1, max 64. Every pixel value gets multiplied before the clamp.

One thing worth understanding before you crank scale to 64: the clamp makes it lossy. Any pixel at or above 1/scale saturates to pure white, so you're trading absolute fidelity for visibility. A scale of 8 maps your 0.03 differences up to 0.24 - clearly visible - while anything over 0.125 clips. If you want a linear brightness lift on a normal image, this is the wrong tool; a gamma or brightness node does that without blowing out highlights the way a linear multiply does. Use Image Amplify for analysis maps, not for "make my render brighter."

The output is a single image tensor, ready to plug into any preview or Save Image node.

Install and gotchas

Install is the standard two-step, and the pack is in the ComfyUI Registry:

cd ComfyUI/custom_nodes
git clone https://github.com/easygoing0114/ComfyUI-easygoing-nodes.git

...then restart ComfyUI. Or just search "Easygoing" in ComfyUI Manager. No extra Python dependencies - the node only uses numpy, torch, and PIL, all already in a stock ComfyUI, so there's nothing model-y or heavy to download.

The one real trap: v2.0.0 of the pack is written against the V3 node schema (io.ComfyNode / define_schema), so it needs a ComfyUI build that supports the V3 node API (comfy_api.latest). If the nodes don't appear after install, your ComfyUI is too old - update it. Also note the v2.1.2 change renamed the difference checker's grayscale_diff_map output to brightness_diff_map; if you have old saved workflows, reconnect that link.

Honest verdict: this is a node you could write in thirty seconds, and that's fine. It's a convenience leaf - deterministic, instant, no model - that exists to make the difference checker's output legible. Grab it because it's bundled with the checker, not because it's clever.