colorcal.app
runs entirely in your browser

Correct an LED wall with a camera you already own.

Show a test pattern, photograph it, and drop the frame in here. colorcal compares what the display should have shown against what your camera actually saw, then tells you in plain language what to change. About ten minutes, no probe required.

Free. No account. Nothing to install.

Your workflow

Which of these is your job?

A workflow here is three answers: where the measurement comes from, what you are correcting, and who the result has to look right to. That last answer changes the math more than the other two put together. One combination is built and working today. The rest are on the map so you can see the shape of the tool rather than guess at it.

Why the camera path shipped first

When the job is to look right on camera, the camera is the observer. We measure in exactly the space that matters, so the answer is a measurement rather than an estimate.

Matching for a person's eyes is a harder problem, and using a camera to do it is a stand-in for the real thing. Cameras and eyes do not respond to the colors in light the same way, and LED walls emit very narrow slices of the spectrum, which is the case where that difference shows up worst. Two walls that read as identical to a sensor can look plainly different to somebody standing in front of them. We would rather ship the path we can defend than have every path sound equally confident.

The built path

Three steps

  1. Show a pattern on the wall

    Play a reference chart full screen on the wall, the same way you would play any other content through your processor. The charts live on our sibling site testpattern.app and are free to download. See which patterns to use.

  2. Shoot it

    One frame per chart, straight on, with exposure and white balance locked so the camera stops helpfully re-correcting between shots. Almost every bad result traces back to this step, so it has its own guide.

  3. Get your numbers

    Drop the frames in and colorcal reads the patches, compares them to what the chart is supposed to be, and hands back the adjustments to make: black level, white balance, and per-channel gain, described in the terms your processor uses.

Honest limits

This is not a probe replacement

A camera is not a spectroradiometer. It sees through its own sensor response and its own lens, and colorcal cannot know either of those exactly. So this tool does not claim absolute colorimetric accuracy, and you should not use its output to certify a wall or to sign off a reference display.

What it does claim is more useful than it sounds: most walls in the field have never been corrected at all. If yours is one of them, a camera match will close most of the gap, for free, in the time it takes to set up. If you already run a probe-based workflow, keep running it.

You do not need to know what a transfer function is to use any of this. Where a term is unavoidable, the page explains it on the spot. If you do know the material, nothing is hidden from you either: the measured values are shown next to the recommendation, so you can see what the tool saw and decide for yourself.

Privacy

Your frames never leave the machine

colorcal is a static web page. There is no upload step, no server that sees your images, no account, and no storage. Your frames are decoded and measured by JavaScript running in your own browser tab, and they are gone when you close it.

That is not a slogan about respecting privacy. It is a practical fact that matters if you are on a job under NDA, on a set where nothing gets posted, or on a network that will not let you upload anything anyway. You can load this page once, disconnect, and it still works.

Before you shoot

Two pages worth reading first