How to Color-Balance Field Photos for Accurate RGB Analysis in Photoshop?

[kian_4445]

I’m working in a lab studying how fruit color influences herbivory, and I need accurate RGB values from fruit photos to use as standardized data points. Each photo includes a color checker alongside the fruit specimen, but the photos were taken in the field under varying lighting conditions, causing slight differences in exposure and contrast. Photos are .jpgs

How can I color-balance these photos in Photoshop (or Lightroom, if necessary) to ensure the RGB values of the fruit are as scientifically accurate as possible? Is there a way to set RGB values of colors in the image to true the values on the card?  I understand that exposure and contrast can impact these values, so I’m looking for a way to standardize these variables across all the photos.

I know that it would have been ideal to make sure the orginal photos were shot in the same light/contrast environement, but I unfortunetly have no control over that. So looking to make do with what I can in post processing.

Once the colors are corrected, I plan to determine fruit color by selecting a representative area of the fruit using the rectangular marquee tool and applying a blur > average filter.

Could someone provide a detailed, step-by-step explanation for this process? I’m new to this and would really appreciate any guidance.

Note: Examples of the photos are attached for reference.

[D Fosse]

The only way to get accurate color measurements is to use a spectrophotometer on the fruit itself. Like they do in the hardware store when you bring in a sample and need paint to match 😉

In a photo, your best bet is to use one of these:

The 24 color patches have standardized Lab values (that can be converted to any RGB color space). If working in RGB, you need to define a color space, sRGB, Adobe RGB etc. In practice, you'll mainly use the neutral patches to set the gray balance and tone curve.

If you're going to fine tune by eye, you absolutely need a good, calibrated and profiled monitor.

For this to have any practical value, you need absolutely consistent lighting. A portable flash unit with some sort of diffusion should work. Also note that angles need to be consistent.

[AlanGilbertson]

You have a tricky task ahead. "Fix it in post" is a very bad workflow. However, given that the shots have already been done, can't be redone, and are what they are, you'll have to rely on the reference card to at least get the color balance and general contrast range consistent. With that accomplished, you can try to match the exposures, but I see from your first image that the color card is exposed very differently from the fruit. I don't think those images are worth any effort trying to salvage them.

Here's how to get the most basic steps accomplished for each usable (evenly lit) image:

1.  Add a Curves adjustment layer, using the icon at the bottom of the Layers panel:
   
2.  There are three eyedropper icons in the Curves layer's Properties panel. Select the top one, which sets the BLACK level.
   
3.  In the Control panel at the top of the UI, change the Sample Size dropdown to 11 by 11 Average:
   
4.  Click in the middle of the black panel of your reference card.
5.  Select the bottom eyedropper tool, which sets the WHITE level, and click in the middle of the white swatch on the card:
   
6.  Select the middle eyedropper icon, which sets NEUTRAL grey, and click in the middle of the third swatch counting from the black (beside the last K of "DKK":
   
7.  Save the PSD and repeat the process for each image.

At this point, for any images that were taken with the card and the fruit in the same plane, so that they received the same illlumination, you will have consistent color and contrast and can read off the RGB values you need if all the images are from the same camera using the same color profile. (If not, I don't want to say all bets are off, but you are unlikely to be comparing like with like.)

Where the amount of light reaching the card is different from the light reaching the samples, you have no way to compensate accurately for exposure, and I would consider these unsable for this kind of application. This is also true if any of your photographs are in mixed light (e.g., the camera used a strobe, but there is also sunlight reaching the samples from a different angle).

You may not have scientifically accurate color and contrast from your usable images, but you will have a baseline that is consistent. To be quite fair here, fully accurate colorimetry is very hard to obtain in the field, if it can be done at all. If your aim is to obtain a set of absolute values, you will need to photograph the specimens in tightly controlled conditions, with calibrated equipment. The cost of that equipment vs. the degree of accuracy is something you will have to work out.

[NB, colourmanagement]

@kian_4445 I'd start by choosing a neutral patch on that colour strip you have and using Photoshop's greybalance. eyedropper in levels or curves to try neutralise the grey balance of the image.

This might help IF the card really does have a spectrally neutral gray patch 

Does that help?

["Colorchecker" is a GretagMacbeth / Xrite trademark name for the patched sample tool that @D Fosse shows in his post]

I hope this helps
neil barstow, colourmanagement net - adobe forum volunteer - co-author: 'getting colour right'
google "neil barstow colourmanagement" for lots of free articles on colour management
Help others by clicking "Correct Answer" if the question is answered.
Found the answer elsewhere? Share it here. "Upvote" is for useful posts.