颜色调整

ColorAdjustment (颜色调整) is the third interactive image node in LAOGOU-666's Comfyui_LG_Tools pack, and it's the same pitch as the crop and resize nodes: the picture appears inside the node, you adjust it visually, the graph waits, and the adjusted image continues on its way. It's the closest thing to a little Photoshop color-grade panel that just lives in your graph.

Where does it belong in a workflow? Anywhere you'd otherwise chain several color-correction nodes and rerun until the numbers look right. A finished generation, a reference photo you want warmed up before it becomes a ControlNet condition, a composite you want to tint before it hits a save node - that's its territory. If you know exactly which numeric adjustment you want, ComfyUI's stock color nodes are fine. If you want to see the change as you make it, this is the one you reach for.

How it works

Same pattern as its siblings: the node sends the first frame of your image to the frontend as base64 and blocks on an event. The frontend gives you the color tools and posts the adjusted RGBA pixel data back to /color_adjustment/apply, which rebuilds the tensor and hands it downstream.

Here's the catch that separates this node from the others: the timeout is five seconds. Not thirty, not fifteen - five. The crop and resize nodes give you time to think; this one assumes you already know what you're doing. Confirm your adjustment quickly or the node passes the original through and the graph marches on. In practice, that means "get the color roughly right, then hit apply" rather than leisurely tweaking.

The inputs that matter

• image (IMAGE) - your input. As with the other interactive nodes, only the first frame of a batch is shown and edited.

One input, one output: a single IMAGE with your adjustments baked in. It's an output node, so it runs even with nothing downstream - handy if it's the last stop before a save.

Installing it

It ships in Comfyui_LG_Tools, so install the pack:

cd ComfyUI/custom_nodes
git clone https://github.com/LAOGOU-666/Comfyui_LG_Tools.git
pip install -r requirements.txt

Or use ComfyUI Manager and search "Comfyui_LG_Tools", then restart. Find it under Add Node → 🎈LAOGOU → Image. The pack installs opencv-contrib-python and transparent-background along with it.

Common issues

• It resets to the original. Five seconds is brutal. Either you ran out of time or you're executing from an API client where no frontend ever answers. Be ready to click before you queue.
• "Why is my image unchanged?" If you're running a batch, only frame one is previewed and edited; the rest of the batch comes through untouched.
• Chinese labels. 颜色调整 is the name in the menu; the pack is Chinese-first with no translation. You'll identify it by the color tools in the node body more reliably than by the name.

Is it a must-have? Not when you know exactly what you want. But when you're finessing a look and want immediate visual feedback, the five-second window is a small price for not having to guess at hue/saturation numbers blind.