Advice on colour checkers

I’m thinking about buying a colour checker to create a profile in darktable for my camera. However, the tried-and-true ones are quite expensive where I live and, being an amateur, I cannot justify the costs. I have been thinking about getting one of the cheap chinese ones, but have no idea if they are worth anything.
Does anyone have experience with these cheap colour checkers? I know they will not be as precise as others, but do they get the job done?

The precision you need is between the target and and the reference file with the spectral values for each patch. The vendor measures their target’s patches to produce that file. If a vendor is saying their target matches the X-Rite Colorchecker spectral values, they’re probably BS-ing you…


Given how expensive these are, doesn’t anyone rent them?

The other thing I have to offer is that you want a matte finish. It’s hard enough to remove glare from a matte target, next to impossible with anything else.

Regarding rentals:

$24US for 7-day rental.

color calibration based on a color checker is just supported for calibrite/xrite and datacolor targets.
That’s more reliable than generating a custom camera profile since that’s also just valid in illumination situations comparable to the target shot.
If your sensor isn’t damaged and behaving quite different to the camera standard profile such one-size-fits-it-all icc file doesn’t give more precision in general

1 Like

Thanks. Is it possible to put this in plain English? I have an x-rite colour checker and have previously created camera profiles (since lost in a Darktable update) for my cameras following the process described in one of Aurelien’s videos. Are you saying it’s not worth doing? Thanks

DT just got support for the Spyder Checkr Photo, which I have seen for as little as 50€ on sale.

generating camera icc profiles for each illumination situation is the gold standard, but needs more effort than the colorcalibration way.
Using a one size fits it all all icc profile generated in a D50 or D65 illumination in different illumination situations doesn’t have the precision expected when using custom profiles instead of the standard profile darktable uses if the latter isn’t completely off.
So it’s up to you to decide, how much effort you’ll spend to get a result and decide which measurement and calculation inaccuracy you can accept…
a lot of different shades of grey and no simple black white :wink:

1 Like

Generally, the profile information included with dt or RT was made with a ColorChecker or snarfed from an Adobe DCP, so your effort to duplicate that isn’t usually going to give you significant gain in what I’ll euphemistically call “color quality”.

These nine-number matrices are usually anchored to a D65 white point, so they’re best with daylight-illuminated images. Making a custom profile for a given non-D65 lighting situation may be prudent, someone with color reproduction experience might be able to tell you how beneficial that might be. This, by the way, is the basis for the Adobe DCP workflow, where the “dual-illuminant profile” actually contains two matrices, usually D65 and StdA (tungtsten and its ilk), and a postulated color temperature is used to interpolate a matrix for processing a given image.

Stop here if all you wanted was an answer to your “why should I…” question. :laughing:

The situation where I’ve personally seen material gain from making a custom profile is in handling extreme hues. The matrices don’t contain enough information to make good gradations of things like red flowers or LED blues in sRGB, so those hues get squashed into a couple of cartoonish gradations. The way to improve on that is to replace the matrix with a LUT, where various algorithms can be used to define good lookups for challenging ranges.

Thing is, a 24-patch target shot doesn’t provide enough information to build such a LUT. Instead, you need spectral sensitivity data for your camera, measured across the visible spectrum in 5 to 10nm increments. Getting such data can be a chore; if someone has already measured your camera and publishes the data, Yay!! If not, measuring the camera traditionally requires a lab setup, with north of $2000US worth of components. I’ve done it with a cheaper setup, about $230US, but my methods might be considered dodgy by lab rats…

It’s worth considering a LUT profile for more than just extreme hue handling. For instance, you can make any color temperature camera profile from the same dataset, no need for two target shots in different lighting. Making a profile from spectral data also requires a “training set”, a set of spectral measurements of known colors with which to anchor the LUT. The ColorChecker dataset is commonly used, but you can swap out a larger or more targeted dataset, like Munsell, or the Lippman 2000 skin tone set.


Due to an accident, I actually bough two color checker classic minis. I lost it, bought a new one, and just yesterday, found the old one again. I also bought one of Wolf Faust ITU8.7/2 targets a few years ago that I found too much trouble to ever use.

So, if anyone of you has a use for either of them, and lives in Europe, you can have them if you pay for postage.

BTW, I found the color checkers very useful for recreating my cameras’ film simulations using darktable-chart.

The bigger chart would theoretically work even better, but the semigloss surface is actually kind of a pain to work with (you’ll need to construct a black box light trap to get a good profile).

1 Like

…checks watch, quietly slips out of the room…

1 Like

Thanks for the replies, the issue is surely more complicated than I had anticipated.
Given the cost of a good (and supported) colour checker where I live, I think I’ll keep working as usual until I find a good value.