Exposure/Black Level vs Brightness/Contrast

[Dazz1]

I am using Darktable at the moment, but I think the question is more general than that. I see a similar question about 9 years ago by IanB but the links in the answers don't work anymore.

So, manipulating the image using exposure/black levels, is very similar, but not quite the same, as using brightness/contrast. I am preferring to use exp/black these days, but I realise I really don't know the difference. I know that combining exp/black adjustments with the shadow/highlight tool adjustments, gets me there for nearly any image. I don't end up using brightness/contrast hardly at all.

[arthurking83]

When you use the term exposure/black levels I assume you're referring to exposure compensation and black levels?

On the proviso that you do mean exposure compensation, then as I remember it:

exposure comp: you literally increase the value of every pixel value in the image by a linear amount across the board. But it's not just increasing the RGB values(for example as would happen in a jpg image), exposure compensation assumes that the program reads raw file data and re manipulates it into new values accordingly. So it's a digital post increase in how you'd have exposed the image at the time of exposure.
Hopefully you're confused by that explanation .. as I certainly am!
The true effect of exposure compensation is almost exactly like increasing ISO value after the fact.

Black levels: are what they sound like. In your program you should have a setting or option to set black point. That is at what RGB values you want black to render 'true black'.
True black is obviously RGB 0, 0, 0, but for a computer screen this may not strictly be the right way to do it. So you have a black point setting to play with. Most raw converters have this feature.
If your program is set to use black at 0, 0, 0 and it sets RGB values in that manner .. when you print an image, apparently 0, 0, 0 values in RGB mode(the normal mode) are not entirely accurate. So your print may have small blotchy anomalies in large deep black areas in the print.
So it's normal to set black point at something like 1,1,1 or even 5, 5, 5 as what you'd see on your screen as black. That is you see black, but the data in the image file is not RGB 0,0,0.
So as you set blacks with your slider, it's supposed to basically move this reference point which ever way you need too to adjust the black.
Can't remember up to what values it work on, but it's used to control the rendering of the lower dark tones.

Brightness: is not like exposure compensation. It's supposed to work more so on the mid to higher tones, but not so much on the far highlight tones(the explanation immediately below won't make sense, but bear with me on that bit, till you read the bit below that too tho)
Each program has their own way of doing things, but usually using brightness will make an image brighter at the higher end of the tone scale.

eg. lets say your tone values are 0-64 for darks, 64-192 for the mid tones and 192-255 for the highlight tones(just roughly speaking to clarify the point).
Say we have an image with certain pixel values, such as 22,22,22 for a single pixel, well call that dark tone(or shadow tone) RGB 22 .. and so on. So ...
With a single value brightness increase adjustment(lets say 10%), you may see an increase in a dark pixel with the value of 22 to 24, but you will see a mid tone increase say from 128 to 140 and a highlight increase from 240 to 264(or past the 255 threshold).
So you can see even tho the single upward value brightness increase seems like a linear increase, relative to each of the tone scales that 10% increase isn't really linear. dark tones only go up by an RGB value of 2, whereas highlight tones increase by 24 RGB value points across the board.

So we see that brightness increases usually affect mid to highlight tones, more so than low value shadow tones.
With respect to the confusing explanation above, I think that in the real world using brightness sliders affects the mid tones a bit more than I've described above. It's all a matter of how the programmer has implemented it in the software, so it varies in all software.

The contrast tool is used to compress the spread of pixel values across the entire tone spectrum in the image. So if you have a widely spaced spread of pixel values clumped in separate groupings, decreasing contrast will bunch the pixel values to make them as broadly and evenly spread as possible.
You need to be able to visualise a histogram graph for the next explanation!
eg. say on the image above where you had a large group of dark tones in the 22 value range(say from 0-64 centred on the 22 range) peaking at the top of the graph range, but you had hardly any pixels in the mid tone value range(a mm or so above the X axis) and again you had a high number of pixels in the highlight tone range again. imagine the histogram with two peaks one on the LHS and one on the RHS, with a large valley in the middle of the peaks.
lower the contrast slider and what it will try to do is to compress the widely spaced tone values to group them all closer together. I say it tries too, success is dependent on the initial image and the software's ability to re render those pixels again.

Important point is to watch the histogram as you edit.

eg. use exposure compensation and watch the histogram simply move leftwards if you decrease compensation and move to the right as you increase compensation. That is, the shape of the histogram stays the same, just transposed one way or the other.
But if you change brightness, the histogram doesn't so much shift as much as it's morphed into a new one as the brightness is altered. Increase brightness and histogram shape changes and moves(to the right).

You can(theoretically) achieve the same results using either method, but the important point to note on that topic is that we humans have very insensitive eyes when it comes to noting subtle differences in colour and brightness. The computer is literal in it's ability.
That histogram doesn't lie.
One may think they have achieved a similar result using either method, but if the histogram doesn't show that, then it's not similar, out eyes aren't sensitive enough to know the difference.

[Dazz1]

Thanks for the lengthy explanation Arthur.

Firstly, yes, it is exposure compensation (or correction as darktable calls it) , +/-3EV allowed on the control.

When reading your explanation something clicked when you brought up the histogram. I opened up darktable and watched the histogram and saw pretty much as you described happening (variations according to how it was programmed aside) For example I would describe brightness as moving the mid-tones to the right while compressing the highlights. This compares to the exposure comp which clips the highlights.

Contrast was the most revealing. It makes the lows lower and highs higher ( expands the scale), but clips the highs. Black level expands the scale but fixes it at the high end, so you lose nothing there at all. This explains my preference for using it.

[Tannin]

If you think about it, only two ways of changing exposure are possible.

First, it helps to conceptualise exposure as a graph of light levels, dark on the left-hand side, bright on the right-hand side. A line upwards to the right. Only two adjustments are possible.

(a) You can push the whole exposure line up or down. (Simply turning up the brightness overall.)

(b) You can change the slope of the line. (Changing the contrast.)

Everything else - all those complex and subtle adjustments photo-editing software can do - are essentially just combinations of those two basic actions.

(You may add (c) changing the shape of the line if you wish; or you may prefer to think of that as simply an advanced form of (b). Either visualisation is correct, use whichever one you like.)

[Dazz1]

If you think about it, only two ways of changing exposure are possible.

First, it helps to conceptualise exposure as a graph of light levels, dark on the left-hand side, bright on the right-hand side. A line upwards to the right. Only two adjustments are possible.

(a) You can push the whole exposure line up or down. (Simply turning up the brightness overall.)

(b) You can change the slope of the line. (Changing the contrast.)

Everything else - all those complex and subtle adjustments photo-editing software can do - are essentially just combinations of those two basic actions.

(You may add (c) changing the shape of the line if you wish; or you may prefer to think of that as simply an advanced form of (b). Either visualisation is correct, use whichever one you like.)

Yes, and understanding it properly means I can choose the best way, and not feel I am doing it incorrectly.. Thanks

[ameerat42]

The picture should indicate that. For instance:
- Is the picture too dark/light overall?
- Are only the shadows too deep?
- Is the contrast flat?
- Have I fed the dog?

[arthurking83]

.....

Contrast was the most revealing. It makes the lows lower and highs higher ( expands the scale), but clips the highs. Black level expands the scale but fixes it at the high end, so you lose nothing there at all. This explains my preference for using it.

Yep, that's what happens when you increase contrast.
But try decreasing contrast, and it makes the darks higher and the high tones lower, hence compresses the overall tone graph(histogram).

Of course this is all dependent on how the image was captured in the first place too.
For example, you may have a high contrast scene, where it's very bright out(full sun, middle of summer middle of the day kind'a thing), but you trying to capture very bright highlights AND quite dark shadows.
The way this kind of scene is captured to begin with is important.
eg. use a polariser to try to lower the very bright highlights. Should be remembered that a polariser is not to 'blue the sky' more for a landscape scene, it's purpose is to lower glare from strongly reflected light sources(in a manner of speaking).
So the polariser 'purpose' is more like an additional(but physical) tone compressing device.

Other thing to keep in mind is to set up the start point of the image for processing. The start point is not a technical term, lets just use it to create one tho.
So for this start point, what you would initially do to start your processing is to set a Picture Style type contrast curve on the image.

eg. if your image was captured with very strong contrast(as above with the bright highlights and dark shadows) your first 'processing' choice would be the Picture Style contrast curve. (You'd go the other way if the image was captured in low contrast situations)
Once the contrast curve is set to help process the tonal range now available, then you delve into the exposure/brightness/contrast adjustments to help make the image render to your liking.

eg. many times I find I captured a very high contrast scene, and the first thing I do is to cycle through the various Picture Control(Nikon term) styles to help with controlling the tone challenge. From there I then choose exposure comps and all those other edit processes.

Think of it as a process tree, a hierarchy of the process steps required to 'better the image' in a logical way.
Remember that you can't really reset the exposure of the image in a true sense.
It's exposure is a fixed thing. Think Shutter/Aperture values. With exposure compensation, you manipulate the gain value that relates to ISO value set(the third variable in the exposure triangle). That's all exposure compensation does .. tries to reset ISO(higher lower).
The next element that the CAMERA does with respect to exposure of the image(at the time of image capture) is set a tone curve.
You know this tone curve as Picture Style(Canon) in the cameras jargon(I know it as Picture Control in Nikon environment).
Your software will call it something else .. like Camera Preset, (I use DxO's software sometimes, and they simply call it 'Presets'). Canons' DPP calls it Picture Style, Nikon's software calls Picture Controls(kind'a makes sense).

Many people dismiss this preset/tone curve feature as a gimmick. It's not.

The important editing tools in your software are(in order of preference) whitebalance, tone curve(preset/picture style) exposure compensation and noise reduction(in a raw converter).
This doesn't imply that you need to set them for every image tho!
All I'm saying is that if you have an image that is marginal, in that it needs a lot of processing due to the way it was captured, then you'd go through the edit step process in that order, and skip any of the steps if they don't require any tweaks.

[Dazz1]

darktable does have a tool called "Tone Curve" with some simple presets. Basically low/med/high contrast. I will have to explore.

[ameerat42]

(Not having DT and going on what Photoshop has) I'd say that corresponds to the Curves function of PS.
It's the graph that Tannin was talking about. You can change it: raise-lower, tilt, twist, etc, and the tones
with undergo corresponding contortions.

- - - Updated - - -

---And I'm blaming mostly you for getting me interested in having a look at it

- - - Updated - - -

This makes the Windows version appear the poor relative

[Dazz1]

(Not having DT and going on what Photoshop has) I'd say that corresponds to the Curves function of PS.
It's the graph that Tannin was talking about. You can change it: raise-lower, tilt, twist, etc, and the tones
with undergo corresponding contortions.

- - - Updated - - -

---And I'm blaming mostly you for getting me interested in having a look at it

- - - Updated - - -

This makes the Windows version appear the poor relative

Had a quick look. It seems, other than attaching a camera, and printing, two things I don't do anyway, it seems the rest is direct port and should be very similar.

I am finding it VERY powerful (under linux), and don't use much else. Gimp sits abandoned. I don't even use Digikam to organise my stuff. Digikam is actually better than darktable, but darktable is good enough, so why bother with extra software.

[arthurking83]

darktable does have a tool called "Tone Curve" with some simple presets. ....

Does it have 'camera profiles' available either within the program or for download?
That is, these camera profiles will usually be specific to your camera .. make Canon, model XXX or whatever.

It's those profiles that try to emulate the Canon Picture Style tone curves.

They are not the same thing as camera support(ie. sensor specific support).