ImageEffectsAdjustment

ImageEffectsAdjustment is a one-stop color-and-light tuning node: brightness, contrast, saturation, hue, gamma, sharpness, and per-channel red/green/blue multipliers, all on a single images input. It's the "Photoshop Adjustments menu in one box" of the Allor pack, and for batch work it's genuinely handy - every image in the tensor gets identical treatment, so you can color-grade a whole animation strip with one node instead of fifty.

Mechanically it's a wrapper around torchvision's functional adjustments (the same primitives PIL's ImageEnhance uses). Each parameter only applies when it's off its neutral value, so a no-op pass costs almost nothing. The neutral defaults are the thing to learn, because they're not all the same number:

• brightness, contrast, saturation, gamma, sharpness - neutral at 1.0. Below 1 darkens/dulls, above 1 brightens/intensifies.
• hue - neutral at 0.5. The node internally subtracts 0.5 and feeds torchvision's hue rotation, so 0.5 = no change, 0 = full shift one way, 1 = full shift the other. People get burned on this one constantly because it's the odd default out.
• red, green, blue - neutral at 1.0, and these are plain multipliers applied per channel after everything else. They're your color-balance controls: crank blue a touch to warm a scene down, reduce green to take the cast out of foliage.

The operation is per-image, keeps the batch shape, and - worth checking before you assume it - it preserves an existing alpha channel. Feed it RGBA and the transparency rides along untouched, which is exactly what you want in a cutout pipeline. Output is the adjusted IMAGE tensor.

Installing it

It's part of the Allor plugin:

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

Restart ComfyUI, or install "Allor" via ComfyUI Manager. Pack gotchas: requirements.txt installs rembg and onnx for the segmentation nodes even though this node needs only torchvision, and Allor's repo was rebased to strip all images from git history (a ~344x size cut) - if auto-updates start failing, the docs at nourepide.github.io/ComfyUI-Allor-Doc have troubleshooting pages for exactly that.

Real talk

This is the node I'd actually reach for over chaining four separate adjust nodes, purely because everything lives in one place and the batch behavior is consistent. The main traps are the hue default (0.5 means "do nothing," and it will absolutely catch you) and the temptation to overdo gamma - a little goes a long way. One honest caveat: it's a global adjustment node, so don't expect it to do the masked, localized grading you'd get from a proper color-grading setup. For a fast, uniform grade on a batch, though, it's hard to beat.