9

Yesterday I went out to take some pictures of a field of poppies next to my home. There seems to be something wrong with the color in most of the pictures. The first one of the following 3 pictures is more or less ok, while the other two look bad (besides being out of focus). I'm not sure what it is. I think they're not saturated, is it the white balance?

Right picture

Wrong color 1

Wrong color 2

I'm using a Nikon D3100.

  • The first picture was shot with automatic, and AWB. 1/200, F7.1, ISO 100, SD.
  • Pictures two and three were shot with shutter priority, AWB. 1/125, F10, ISO 200, SD.
  • The red channel is blown (i.e. it has reached the saturation point). The typical visible sign of this is that red areas in the picture look flat, with little noticeable colour variation. If you have the RAW file, check if the red is blown in that too. If it does, underexpose next time to compensate. – Szabolcs May 21 '18 at 11:24
  • @Szabolcs And if the red is not blown in the RAW file? Is it a white balance problem then? – antonro May 21 '18 at 12:08
  • In the future, just using a card to set the white balance is a really simple solution. – user3067860 May 21 '18 at 21:27
  • 1
    This looks more like overexposure than a white balance problem. – bwDraco May 22 '18 at 7:09
25

What is going on?

Comparison

I compared both pictures of the field (left out the one with the tractor, as it suffers from the same problem as the other over-exposed picture, IMHO) in After effects. The image above is a composition of all that I did: First, the composition of both your original images that I made in AE (white canvas added only here), then both vectorscope readouts (each picture separately) and then, the waveform parade of both (separated by the black border in the composition, which is shown as thick line at 0).

What we can see:

  • The left picture is much brighter,
  • The left picture is clipping a lot in the red channel, while the right one is not clipping as much,
  • The greenish colors have a wider range of tones (stretched closer to yellow) in the left picture.

A bit of trying around showed me that with a gamma correction of 1 / 2.2 = 0.4545, the left picture gets quite close to the right picture (or, if wanted, a gamma correction of 2.2 gets the right one close to the left one):

corrected


Why is that?

Looks like overexposure and its effects on JPEG to me. If you saved the images as RAW, you could try to lower the exposure. It is hard to give tips or guess why this happened without knowing how you accomplished the shots: If you exposed manually, then you simply got it wrong a bit. If it was an auto-exposure, then your camera got it wrong somehow. The white balance seems to be slightly off (too blue), but that is not the main problem here.

A wild guess: You were shooting in M mode. You set your exposure settings when it was a bit cloudy - the first shot got properly exposed. Then, the sun came out a bit more, but you did not notice and therefore, your shots got slightly overexposed.


Setting things (roughly) right in GIMP:

my GIMP try

Beware: This is the first picture I ever edited with GIMP - and I'm bad at editing non-RAWs, anyway. So what you see here is merely what you can accomplish as a newbie - and in 5 minutes time, without any talent. Also, please understand that over-exposed parts of a JPEG cannot be recovered (easily) - you can easily see that the flower's petals are blown out.

What I did:

  • The red/orange is so saturated in your last example that most detail of the flower petals is lost. – Michael C May 21 '18 at 8:41
  • @MichaelClark You're right, of course. However, although the color correction might have had an impact on that, it's mainly because (as seen above) the petals are blown out anyway and I did nothing to recover them. They were already/still blown out after using the curve tool. – flolilo May 21 '18 at 10:23
  • 2
    upvote purely for using a vectorscope :) – James Snell May 21 '18 at 12:35
  • @JamesSnell While I too love to use waveforms and vectorscopes, it makes me sad to see that you don't like the rest of my answer. :-( ;-) – flolilo May 21 '18 at 15:01
  • 1
    @JamesSnell I was just kidding, but thank you ;-) Yeah, well, vectorscopes are a bit difficult to read, especially when the subject isn't a color chart/SMPTE color bars. Though I still find them more practical in real world than histograms (waveform for the win!) – flolilo May 22 '18 at 10:33
5

The first one appears, to my eyes, to be slightly oversaturated, particularly in the red color band (which is the more narrow color we call "red", rather than the entire red color channel that include "magenta" and parts of "orange"). Many digital cameras tend to do this to very red objects in a photo when "Auto White Balance" and "Vivid", "Portrait", or "Standard" picture styles are selected and used to convert the raw sensor data to a viewable image.

There's also the issue that many cameras place grass and other "green" plants more in the yellow color band than the green band when set to the aforementioned raw processing modes (AWB and Vivid or Standard PS).

The other two appear to be slightly overexposed, which tends to reduce color saturation and also can shift colors if "Auto White Balance" is used.

The solution is to take control of the color of your photos by setting a specific color temperature (such as 5200K) and specific white balance correction settings¹ (such as Green +3, Blue +1) either in camera or in post processing using raw image files. If that isn't enough, the next step is to adjust each color band (Red, Orange, Yellow, Green, Aqua, Blue, Purple, Magenta) independently using an HSL (Hue-Saturation-Luminance) or HSV/HSB tool (where the 'V' or 'B' stands for 'Value' or 'Brightness' instead of 'Luminance') in post processing.

  • @MikeW Several of those links didn't even have an answer from me. It's obvious from the way many of the questions on this site are asked that new users do not bother to "to read more about [white balance] (or whatever else it is they are asking about) they can find it here on the site and be guided by the site's metrics" unless they are provided links in an answer to their question. – Michael C May 22 '18 at 3:40
  • In my limited experience, I found that 1-2 "further reading" links are a good thing, but if you slap 200 (I'm exaggerating, obviously ;-) ) links below your answer, people will feel overwhelmed and will not even bother to look at one of them, no matter whether they want further information or not. They'll simply decide that it's too complicated. (As stated: Personal experience, not an empirically proven theory.) – flolilo May 22 '18 at 14:35

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.