We’d like to think that white balance is a science, right? And that the Temperature and Tint values that form the basis of white balance adjustments are universal constants that we can expect to be consistent across all software applications?
That’s not how it is. If you’ve ever suspected that different RAW converters interpret white balance settings differently, here’s the proof.
All three produced broadly similar results using the camera’s own As Shot setting and their own manual Daylight white balance presets, but all three disagreed not only about what those Daylight settings should be, but even what the camera captured in the first place.
What happened with the As Shot settings
My sample image was taken with a Sony A6000, which normally gives reasonable results using auto white balance, but with a Laowa 9mm ultra-wide lens fitted, the while balance came out very ‘cold’ for some reason.
Now you would expect each of these RAW converters to accurately report the white balance Temperature and Tint values recorded by the camera in its EXIF data, wouldn’t you?
That’s not what happened.
In each case, I left the software’s white balance settings at ‘As Shot’, and although the colours looked similar, the three programs reported three different sets of values from this RAW file:
Adobe Camera Raw: Temperature: 4050K, Tint: +10
Capture One: Temperature: 4336K, Tint: 4.3
DxO PhotoLab: Temperature: 4321K, Tint: -30
Now it’s difficult to make exact comparisons with the Tint values because the three programs use different scales – but even here it’s clear that Adobe Camera Raw and Capture One are applying a positive Tint value (more magenta), and PhotoLab is applying a negative adjustment (more green).
Worse than that, these programs don’t agree about the recorded Temperature value, and you would surely expect this to be an absolute value recorded by the camera?
What happened with the Daylight preset
So the next step was to apply the Daylight white balance preset in all three programs to see if they would agree about the Temperature and Tint values this time.
Well, they all produced broadly similar colours, correcting the blue colour cast in the original image, but they did it with different white balance adjustments:
Adobe Camera Raw: Temperature: 5500K, Tint: +10
Capture One: Temperature: 5295K, Tint: 0.1
DxO PhotoLab: Temperature: 5200K, Tint: 0
This is interesting.
DxO PhotoLab is the only one to reproduce what most photographers would consider a classic, ’neutral’ daylight colour response, i.e. a colour temperature of 5200K and no Tint adjustment.
For some reason, Adobe Camera Raw thinks ‘daylight’ is 5500K with a +10 magenta shift.
Capture One appears to be a law unto itself. Its Daylight setting has very precise values that led me to investigate its response to other camera RAW files, and it soon became apparent that it’s different every time. The other two programs are at least standardising on what they think the ‘Daylight’ Temperature and Tint values should be, but Capture One appears to be changing the values camera by camera as a kind of camera-specific correction that gets you to the same place but is subverting the whole point of white balance adjustments.
All RAW converters use camera ‘profiles’ as a basic processing step ahead of any manual adjustments, and Capture One should be correcting camera characteristics here, not with the white balance settings. I love Capture One, but this has been a disappointment.
This test has been a bit of an eye-opener. I’m somewhat surprised and disappointed that white balance Temperature and Tint values are not the consistent and absolute settings I’d assumed them to be (I don’t suppose I’m alone here!)
The three programs I tested all gave similar outcomes (well, more or less), but DxO PhotoLab is the only one I’d consider gave classically correct Temperature and Tint values. Adobe Camera Raw reported different values in the RAW file and, like Capture One, appears to juggle the values applied with the ‘As Shot’ setting to achieve ‘correct’ colour – but at least its ‘Daylight’ setting was consistent, if a little odd. I did not like the way Capture One changes its ‘Daylight’ values camera by camera.
Some remarks about colour
In my regular day job I test a lot of different cameras, not just software applications, so I get to see how a lot of different RAW files (and RAW converters behave). The results I got in these comparisons confirm those I’ve seen many times before, so here’s how I’d characterise the colour response of these three programs:
Adobe Camera Raw: Generally poor. I find the presets too ‘warm’ and blue skies often show a slight shift toward violet, while mid-tones can sometimes be reddish.
Capture One: Generally good, though blue skies can shift towards cyan or green as they lighten towards the horizon (I’d rather have ACR’s violet shift).
DxO PhotoLab: Spot on. It’s slower and more awkward to use than the other two, but every time I use it I’m reminded how good it is – and in the ‘Daylight’ comparison I carried out for this test, its result is the best by some margin.
In fairness to Adobe Camera Raw and Capture One, it is possible to adjust the default response to individual cameras, but I don’t think that’s something you should have to do.
The overall conclusion is that white balance Temperature and Tint values are interpreted and applied differently by different RAW converters. You can achieve ‘correct’ colour using manual adjustments or white balance presets in each program, but you cannot manually apply Temperature and Tint values measured separately or in another program an expect the same results.