How can I make this image appear OK on an average monitor?

Question

I have taken a portrait picture which has to look good. It does look good on the Dell 2410 I use for editing. On the garden variety Fujitsu I use as a secondary monitor, it is terrible. In the face, the highlights are yellowly-blown with a reddish halo and the shadows are greenish. Half of the auburn hair looks flat black, and the remainder is tinted orange.

I know that the Fujitsu isn't meant for picture work, but usually, when I look at a random portrait from the web, or at other portraits taken with the same camera and lens, I don't see such an extreme difference. There is usually loss of detail, but not this overcontrasted oversaturated monstrosity. I am not happy that it looks good on the Dell, because I expect most people who look at the picture to do so on an average monitor. Any ideas how to make it look good?

I have no idea what causes the problem. It isn't overzealous editing, as the problem is already present in the NEF original. It has much more colour noise than my lens usually produces at ISO 400 (I don't know why it had such a bad day, maybe the green background), but reducing it with noise filters/scripts didn't help any, so this probably isn't the cause. Any hints are welcome.

If you want to see it for itself, I uploaded it on http://www.flickr.com/photos/28989244@N05/5160991505/sizes/l/in/photostream/. It was shot in AdobeRGB and converted in sRGB with ufraw. I use GIMP for editing, don't have access to expensive software like Photoshop and Lightroom. But if you know how to repair it in Photoshop, describe it and I will try to replicate the workflow in GIMP if possible.

Edit: Matt's answer wasn't full, but got me thinking in the right direction. I started from scratch. Just colour balance didn't do it, but sharply reducing the contrast got rid of the glowing patches on cheeks and brow, then changing the tone and saturation for each primary, once for a skin area selection and once for the rest, did produce something which looks presentable on both monitors.

Answer 1

My first instinct was a colour profile issue but then you said you converted to sRGB. To me the image has too much green in general, looks great on the foliage in the background but not in the skintones. The shadows will always have a green tint with this sort of image as any areas not directly lit will be receiving green light reflected from the nearby leaves. BTW I'm viewing your image on an average monitor right now!

I would suggest you have another go at the white balance on you NEC, cool it a fraction and err on the magenta side. Apart from that there is no magic solution, as you have no control over the monitor - or ambient light conditions of an arbitrary web user! There is no "average monitor" profile, all non calibrated displays will be off by a random amount!

The only thing you can do is reduce the saturation - this will at least limit how far off the colours can appear!

Answer 2

I'm not sure if your monitor is to blame here. I think there is a (green) color cast in your photo, your white-balance is likely off. I dropped it in Photoshop Elements, and using Auto Levels or Auto Smart Fix I was able to achieve a more realistic skin tone. It seems Gimp has an Auto menu, try "Auto White Balance" for kicks.

On the left, yours, then Auto Color, then Auto Smart Fix. There is a compromise to be found in there.

I don't think you mentioned that your monitors were calibrated, but that's really an important first step. Some decent colorimeters can be bought for cheap, I use an inexpensive HueyPro.

As far as White Balance is concerned, you can either trust Gimp, Photoshop or Lightroom to guesstimate a better WB and fix the color cast, or you can help out by picking what you think should have been gray in your photo using a WB picker (in Lightroom at least). Another option, you can buy an inexpensive set of grey cards and take a picture of the card under the same light conditions. This will give you the proper color temperature to use when you look at that photo.

Answer 3

First off, I think you did the right thing by converting the image to sRGB. Most computer screens are designed to output in the sRGB color space. Additionally, few web browsers provide proper color management, so when they render images, they usually assume sRGB. An image from a different color space will usually render incorrectly on the average viewers screen due to these factors.

That said, you could have a few issues. First, I would make sure your image is properly color balanced (or white-balanced.) White balance is a key factor in achieving a color-correct image in a color-corrected workflow. Conversions between color space sometimes work around the white point. Luckily, AdobeRGB and sRGB both use the same white point (6500k, often abbreviated D65), so conversions between these two color spaces usually does not cause any color shift.

An important factor about color space conversions is the intent. There are a variety of intents, including Absolute Colorimetric (which shifts white point if there is a difference between the two color spaces), Relative Colorimetric, Perceptual, and Saturation intent. The two most commonly used are Relative and Perceptual. Relative attempts to maintain as much color accuracy relative to the original space as possible, however it will shift colors somewhat to compress them into a narrower gamut (since AdobeRGB is a wider gamut, Relative intent will "compress" colors into the sRGB space). Relative intent may introduce some color shifts when converting, since it is relative to the original space. Perceptual intent attempts to maintain as much color accuracy relative to how human eyesight perceives colors in relation to each other. While it will not preserve original colors relative to their original color space, it will usually produce an image that looks correct to a human observer.

Color correction and color spaces aside, I think your image is just not properly color balanced. The kind of greenish/yellow tinting you see can be corrected with post-processing tools. I am not sure exactly what Gimp is capable of, or how accurate or reliable its ICM is, but I do believe it has color balance controls. Here is an example of your image that I have corrected using the Auto Tone and Auto Color correction tools of Photoshop:

Here is a split sample of her face, showing the original image to the left, the image with simple tone correction on the right, and both tone and color correction in the center.

If the fully corrected image from auto correction tools does not meet your needs, then you can always do some fine-tuning adjustments after the fact to perfect the results. Use the curves and levels tools, and possibly some slight saturation adjustments.

Answer 4

If you shot in raw, you have a mosaic file which is just a bunch of intensities: it has no colour space in the sense being discussed here. Any camera setting which says "colour space" is to tell the camera what to do when it does on-board JPEG conversions. It's not "shot in sRGB and then converted to sRGB", it's shot in raw and then converted. There's a big difference, part of which is directly relevant.

The interesting thing about that shot is that it's got two colours which look pretty much complimentary. So I think you'll find that colour balance is tricky. If you're trying to colour balance on the non-linear (jpeg, after conversion) image then it's especially so. There the balance control is extremely sensitive. That's because you're working with non-linear data. So the trick is to do the colour balance on the raw file, in the raw converter, if you can. It's much easier there as you have much finer control.