LayerColor: ColorTemperature
ColorTemperature
- image
- image
This is the warm/cool slider, and it's probably the most intuitive color node in the whole pack. LayerColor: ColorTemperature does one thing: it shifts the image toward warm (orange) or cool (blue), the way a camera's white-balance dial does. One knob, obvious result. When a render comes out too cold and clinical, or you want that golden-hour warmth, this is the fastest way there.
There's exactly one control that matters:
- temperature - runs -100 to 100, default 0. Positive warms the image (pushes it orange/amber), negative cools it (pushes it blue). Zero is a no-op.
Feed it an image, get an image back. That's the entire node.
The reason it feels nicer than fixing warmth by hand in the RGB node is that it moves along a single perceptual warm-cool axis instead of making you balance two channels against each other. In RGB you'd nudge red up and blue down and eyeball the ratio; here you just turn one dial and it does the coordinated shift for you. For the common "this needs to feel warmer/cooler" instinct, that's the whole game.
Where it fits: it's a white-balance and mood tool. Correcting a scene shot (or generated) under the wrong light temperature, matching two images so they sit in the same lighting, or deliberately grading toward cozy-warm or cold-and-moody. It's a natural neighbor to the other LayerColor nodes - do your temperature here, your channel work in RGB, your exposure in the Exposure node, and you've built a tidy little grade chain.
Don't overthink this one. It's not trying to be a full grading suite, and it shouldn't be your tool for a specific color cast that isn't on the warm-cool axis (a green fluorescent cast, say - that's a ColorBalance or RGB job). It's also easy to overcook: much past ±40 and skin tones go orange or corpse-blue, so keep it gentle for anything meant to look natural. As a subtle finishing nudge, though, it's exactly right, and it's the node I'd reach for first when an image just feels a little off-temperature.
Install is the pack standard. The easy way is ComfyUI Manager - search ComfyUI Layer Style (or "LayerStyle"), install, restart. Manual: cd ComfyUI/custom_nodes && git clone https://github.com/chflame163/ComfyUI_LayerStyle, then pip install -r requirements.txt from inside the folder using ComfyUI's Python, and restart. On a hosted ComfyUI it's usually already baked in.
Worth knowing about the pack as a whole: Layer Style has a reputation for import failures. You install it, restart, and the entire node set is red or shows up missing. It's almost never this specific node - it's a dependency clash. The pack pulls a heavy dependency stack (transformers, onnxruntime, opencv), and one mismatched version - a stray transformers or tensorflow in the environment being the classic - brings the whole pack down on import. The fix is dull but works: open the ComfyUI startup log, find the traceback (it names the offending package), reinstall requirements against ComfyUI's own Python, and use Manager's Try Fix. If ColorTemperature is missing, every LayerStyle node is missing with it, and that shared import failure is what you're actually fixing.
Inputs (2)
| Name | Type | Default | Description |
|---|---|---|---|
| image | IMAGE | — | |
| temperature | FLOAT | 0-100–100 | — |
Outputs (1)
| Name | Type | Description |
|---|---|---|
| image | IMAGE | — |