How to Match Colours Between Two Photos (Without Guessing)
Matching two photos by eye is unreliable, because eyes adapt. Look at a warm frame for thirty seconds and it stops looking warm — which is precisely when you make the wrong call.
The fix is to work in a fixed order and lean on measurements at each step. The order matters because each stage changes what the next one sees.
Pick the reference first
Decide which frame is correct and do not touch it. Matching two images to each other simultaneously means chasing a moving target, and you will converge on something neither of you wanted.
Usually the reference is whichever frame has the best light. If neither does, correct one properly first and make that the reference.
Match exposure before anything else
Find a region present in both frames that should be the same brightness — a patch of sky, a wall, a subject's forehead — and match its luminance. Everything downstream depends on this, because contrast and colour perception both shift with brightness.
Use a measurement rather than judgement. A pixel readout or a histogram tells you the two patches read 148 and 172 far more reliably than your eye, which is busy adapting.
Then white balance
Find something that should be neutral in both frames and make it neutral in both. If R, G and B read within a few points of each other on a grey patch, the white balance is right regardless of how the frame feels.
If there is no neutral object, use skin, which occupies a narrow and well-documented hue range. Skin that reads too magenta or too green is a white balance error, not a grading choice.
Then contrast, then hue
With exposure and white balance matched, compare the two histograms. If one has crushed blacks and the other does not, that is a black-point difference, not a colour one — fix it with blacks and shadows rather than by chasing colour.
Only now address hue. At this point most of the perceived mismatch will already be gone, and what remains is usually one or two HSL bands — often greens, which vary enormously between lighting conditions and are the most common residual offender.
- ·1. Exposure — match a common region's luminance
- ·2. White balance — make neutrals neutral in both
- ·3. Black point and contrast — match the histogram shape
- ·4. HSL — fix the one or two bands still visibly off
- ·5. Only then, a shared grade on top
Matching a whole set
For a shoot, correct every frame to neutral independently, then apply one shared grade across all of them. That is the reverse of the instinct — which is to find a look on one frame and paste it everywhere — and it is why the instinct produces sets that almost match.
The shared grade is only consistent if what it lands on is consistent. Correction is per-frame; grading is per-set.
Frequently asked questions
Why do my photos look matched on my screen but not on my phone?
Screen calibration, and often screen white point. If your monitor runs warm you will correct against it and everything will read cool everywhere else. Checking on a second device is a rough but genuinely useful sanity check.
Can I just copy settings from one photo to another?
Only for grading moves. Copying exposure and white balance across frames shot in different light produces exactly the mismatch you were trying to fix — those are per-frame corrections. Copy the HSL, curves and split toning; leave the corrections alone.
What if the two photos were shot in completely different light?
Accept a partial match. Golden hour and overcast midday cannot be made identical without a grade so heavy it damages both. Match exposure and neutrals, get the skin tones consistent, and let the ambient difference stand — it usually reads as intentional.
Is there an automatic way to match two photos?
Automatic matching solves the global part — exposure, white balance, overall saturation — by measuring both images and fitting the difference. It is a good starting point, and it is measurement rather than guesswork, but per-hue work usually still needs a person.
RawCut's pixel probe reads exact RGB and luminance values off any point in the frame, and the live histogram shows the tonal distribution shifting as you work.
Measure instead of guessing