4

I'm using Darktable 2.4.2 with Canon 6D. The problem is that when I import a raw image, and apply the Canon 6D base curve, the histogram is totally different with respect to the histogram visualized on the camera. I attach the image to explain better. I tried to calibrate the base curve using basecurve internal tool with no results. My photos appear too bright. Can you help me?

enter image description here

  • Just out of curiosity, if you load a JPEG version of an image into darktable, how closely do the histograms match? This would be more of an apples<>apples comparison of the histograms themselves. – junkyardsparkle Apr 27 '18 at 0:43
  • 1
    Loading JPEG, histogram is the same until the base curve is applied. After, is quiet different. – Gabriele Fatigati Apr 28 '18 at 15:35
  • So, that tells you that the histograms give more or less the same display for the same input (normally you wouldn't apply a base curve to a JPEG). As suggested in the answers below, then, creating a base curve that matches the camera JPEG should give you similar histograms. – junkyardsparkle Apr 28 '18 at 22:32
4

First thing to do is to check the "mode" of your histogram. Darktable has three ways to visualise the histogram. One is the 'normal' one, which is linear in intensity, the second is logarithmic in intensity, and the third is a "waveform" display (I never figured out what the use of that one is). To see which mode is displayed, hover your mouse over the histogram. Four squares will appear: left to right display mode, red channel, green channel, and blue channel. Clicking on those squares cycles through the states.

So try clicking (perhaps a few times) on the left square until it shows a straight line (log mode will show a curve, waveform mode some kind of squiggles).

Then, there are a few basecurves for Canon provided. Perhaps one of the alternatives gives a better results? Nothing obliges you to use the camera basecurve. The manual even suggests not using that module at all, if you want to take full control through the "tonecurve" or "zone system" modules.

The basecurves are supplied as an easy way to get close to your camera's rendering of the raw files. Some, however, rise very fast, and can easily lead to clipping in the final image (on which the histogram is based). Try using the "cubic spline" to see the image without any curve applied.

  • Yeah, the two histograms appear to be showing roughly the same information, the horizontal axis is just "stretched" differently. Both histograms show clipping in the highlights. – Michael C Apr 24 '18 at 22:19
  • Actually it's the vertical axis that is "stretched" differently (log instead of linear, which is why the lowest intensities seem a lot higher). And yes, both seem clipped (probably the sky). – remco Apr 25 '18 at 5:36
  • When you stretch the horizontal axis based on a lower logarithmic power, it bunches up the areas that are close in intensity and eventually some of them get stacked on each other where before they were in discrete bands next to each other. That's why the histogram on the camera looks taller than the one on the computer. It's because there are not 256 distinct bands on Canon's in-camera histograms. The scale lines on the in-camera histogram are two stop apart, so the distance between each two lines is twice the difference in intensity as the previous set of lines. – Michael C Apr 25 '18 at 6:18
  • The one on the computer seems to be more linear on the horizontal axis. The vertical axis is just the number of pixels at a given intensity on the horizontal axis. The horizontal axis is the one that moves from lowest intensity on the left to brightest intensity on the right. – Michael C Apr 25 '18 at 6:21
  • I just notice something else that would influence the histogram: the camera is in AdobeRGB mode, DT by default produces sRGB. That's apart from the DT histogram showing only the blue channel. I guess we'll have to wait for OP to respond/clarify. – remco Apr 25 '18 at 6:52
2

In the photograph, Darktable's history stack indicates Curva Base [Base Curve] has been applied to the digital file...twice. Usually the first time is a default base curve based on the camera model. Without custom configuration, Darktable applies a tone curve by default to RAW images. Because there are two applications, the Base Curve is definitely not the default, though it may have also been applied automatically. Anyway, any tone curve (gamma adjustment) applied by Darktable, even the default, may differ from the manufacturer's built in tone curve (gamma adjustment) because gamma adjustment is a matter of taste.

It's also worth considering that the single histogram from the camera combines red, green, and blue channels in an arbitrary way to provide information at a glance. With most images, the darkest shadows and brightest highlights will not consist of all channels equally (unless the shadows/highlights are blown to black/white). Interestingly, the Darktable histogram seems to be just for the blue channel. Blue channel extending well to the right is common for daylit images containing sky. Often in such images, the blue channel will also extend well to the left because shadows are illuminated by ambient blue light from the sky.

  • 1
    History stack can show any module several times, all that means is that you changed parameters several times. Only the topmost module of a set with identical names will be used (use "compress histoory stack to get rid of the extra mentions). And the normal DT histogram shows all colors, unless you tell it not to (the 4 squares I talked about) – remco Apr 25 '18 at 5:36
  • @remco I agree. The relevant part of my answer is that the history stack shows that the Darktable histogram reflects changes made by the Base Curve tool within Darktable. This is one reason the histograms don't match. – ben Apr 25 '18 at 16:00
  • Minor quibble: just disabling and re-enabling the base curve is enough to put it in the history stack a second time, without changing anything. – junkyardsparkle Apr 27 '18 at 0:35
  • Minorer quibble: blue colored shadows wouldn't extend the blue histogram channel to the left, yellowish ones would. – junkyardsparkle Apr 27 '18 at 0:36

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.