◎ Radiance False Color
See your exposure in a 7-zone false color overlay
- image
- false_color
- original
Professional cameras and monitors have a false-color exposure mode that paints an image in a rainbow of meaning: magenta for mid-gray, green for skin tones, red for clipped highlights. It's how cinematographers know what's actually exposed without trusting a screen that lies. Radiance False Color brings that to ComfyUI - it takes an image and bakes a 7-zone false-color exposure overlay onto it, so you can see at a glance which parts of a frame are under, over, or correctly exposed.
What makes this version interesting is that it's built for HDR data, not just 8-bit. There's an is_linear flag because the zone classification only means something if it's applied in the right space: on a scene-linear (HDR) image, the zones map to real stops of light; on an sRGB-encoded PNG, you need it flagged as such or the mapping is nonsense. That's the difference between a toy and a tool - the same overlay is only trustworthy if it's analyzing the right math.
The controls
- image - the frame to analyze.
- is_linear - true = scene-linear HDR input, false = sRGB-encoded. Get this right or the overlay will lie to you.
- blend - 0 to 1. 0 = original image, 1 = pure false color. You'll usually preview at 1 to read the zones, then drop it toward 0 to check against the actual image.
- exposure_offset - −6 to +6 stops. Shifts the analysis before zone classification. This is the useful knob: if your key subject is sitting two stops under where you want it, dial in the offset and the overlay tells you where it should be, not just where it is.
Outputs: false_color - the overlay as a regular IMAGE (which is the point: it's baked, so you can feed it to a save, a compare, or a viewer without special handling) - and original, the image passed through untouched.
When to reach for it
Exposure QC is the main gig: run your frame through, confirm the skin tones land in the right zone, confirm highlights aren't crushed to red. For a pipeline it's the analysis half of a loop - you can't easily automate "is this exposed right" with most ComfyUI output, but a baked overlay lets a QC step (human or automated) look at a stable, interpretable image. Compared to the scopes in the same pack (waveform, vectorscope), false color is the one that tells you where in the frame the problem is, not just that one exists.
The honest note
It's an analysis node, so it doesn't fix anything on its own - it shows you the truth and expects you to act on it. And like several Radiance nodes, it's more valuable to people already working in a float/scene-linear pipeline; if all your images are 8-bit PNGs the overlay still works, but the stop math is doing less for you. The pack's launch reception was mixed on its grander claims, but this node is a straightforward, well-scoped utility that does exactly what it says.
Install
In fxtdstudios/radiance - Manager search "Radiance", or:
cd ComfyUI/custom_nodes
git clone https://github.com/fxtdstudios/radiance.git
cd radiance
pip install -r requirements_linux.txt # or windows/mac_silicon
Restart after; Linux needs libopenexr-dev before pip, and the heavy dependency set (OpenEXR, OpenColorIO, colour-science, transformers) makes first boot slow.
Inputs (4)
| Name | Type | Default | Description |
|---|---|---|---|
| image | IMAGE | — | |
| is_linear | BOOLEAN | true | True = input is scene-linear (HDR). False = sRGB-encoded (PNG). |
| blend | FLOAT | 1.000–1 | 0 = original image, 1 = pure false-color. |
| exposure_offset | FLOAT | 0.00-6–6 | Shift analysis by N stops before zone classification. |
Outputs (2)
| Name | Type | Description |
|---|---|---|
| false_color | IMAGE | False-color exposure visualization (7-zone industry palette). |
| original | IMAGE | Original image passthrough. |