What Does the Exposure Slider Actually Do?

[wernerg]

I have been noticing color changes using the Lr exposure slider so I tried to do a systematic test and comparison with Capture NX2. The following links are all jpgs from the same raw NEF that was shot approximately one stop underexposed. (The links don't seem to highlight until you put the cursor over them. Edit: they do highlight in the final post, sorry)

Lightroom, As Shot, Adobe Standard Calibration for Nikon D90 .

Capture NX2, As Shot, Standard Camera Calibration for Nikon D90 . and using Nikon's Adobe Wide color space, which appears to be close to ProPhoto. However, all the outputs were converted to sRGB jpgs.

Lightroom, As Shot, same calibration, exposure slider moved to right until one color saturates, 3.35 stops.

Capture NX2, As Shot, same calibration, exposure slider moved to right until one color saturates, .75 stops.

The two calibrations are close enough alike that I don't think the differences are material to this test. In Lr, the colors are obliterated long before any color saturates.

1. What is the exposure slider in Lr? I did a search on the forum and saw a comment that it is a "whites" adjustment and compared it to something in PS. I don't have PS to compare to.


2. If the colors are blown long before saturation is indicated, what is the point of the saturation indicator? How can I adjust exposure without getting color changes?

The Capture NX2 results are about what I would expect. The Lightroom results are confusing and disturbing if this test is actually valid. You can put these default jpgs into Lr and get similar-looking results but with different numbers.

 

[Mark Sirota]

How are you determining when a color saturates? If you're looking at the clipping indicator in the histogram, that's not as accurate as zooming in to 1:1 and holding down the alt/option key while you adjust the Exposure slider. But still, that's showing clipping in ProPhoto, which is huge. In some colors it might not clip there, but might get clipped later when you export as sRGB. To see that, you need soft proofing, or a raw converter that can work in sRGB.

 

[wernerg]

How are you determining when a color saturates? If you're looking at the clipping indicator in the histogram, that's not as accurate as zooming in to 1:1 and holding down the alt/option key while you adjust the Exposure slider. But still, that's showing clipping in ProPhoto, which is huge. In some colors it might not clip there, but might get clipped later when you export as sRGB. To see that, you need soft proofing, or a raw converter that can work in sRGB.

I was looking at the clipping indicator. I tried the 100% method, guessing at where clipping would occur and got a different number but the same obliterated colors. This method said that the first color to clip was green and at 2.7 stops rather than 3.35 stops, yet it is clear from the image, which looks the same as the third image I posted, that the reds saturated much earlier. What is disturbing is that I have been assuming that colors remain "true" until at least one color saturates. I can also watch the behaviour of the colors in the histogram as I move the slider and it is clear that the colors do not move together, the reds are overtaken by the yellows as the slider moves to the right. That explains why the color turns yellow but it doesn't explain why it behaves like that. And I really don't understand the big numbers. Capture NX2 says it is only .75 stops underexposed using Adobe wide color space, and that's consistent with what the camera metering indicated.

If the exposure slider is not really an exposure adjustment, then what is it, and where is the real exposure adjustment? I was able to get much better results by sliding the white point of the tone curve to the left until the clipping indicator lit up. It correctly indicates that red is the first color to saturate and the image looks like the image I get out of Capture NX2 with +.75 stops (the fourth image). Of course, that also increase the slope of the tone curve which increases contrast. But it does seem like a safer method to increase exposure albeit by unknown amounts and with contrast changes.

 

[Mark Sirota]

I don't know for certain, but I think the exposure slider is a true exposure adjustment. I think what you're seeing is clipping from the color space conversion.

Do you have Bridge and Camera Raw? What if you use Camera Raw, setting the color space to sRGB while making the exposure adjustment? (This is one of the differences between Camera Raw and Lightroom's Develop module -- in Camera Raw, you can set the output color space.)

 

[wernerg]

But I'm seeing all this on my monitor in Lightroom's Development module, a 24" calibrated top-of-the-line NEC, the posted jpgs are just vehicles for communicating what I see on the monitor. It's the lack of clipping that's the issue. How can I change the exposure of a .75 stop underexposed raw file by 3+ stops and not see clipping even though the colors (on the monitor, in Lightroom ProPhoto) are obliterated?

No, I don't have those other programs, I only have Lr and Capture NX2.

I've been doing some googling while this has been posted and from what I read the exposure slider in Lr is not an exposure adjustment at all, it anchors the blacks and pushes the highlights, with the highest tones moving the most (I interpret that to mean that colors of different brightness will not move together), while the brightness slider is the opposite, holding back the highlights to protect them while pushing midtones. It appears that in the end its all guess work, if it looks good, do it, if it doesn't look good, don't do it. There was also a discussion by Martin Evening that was several years old that suggested that using the tone curve was a better way to make brightness adjustments that either exposure or brightness sliders, although he has since modified that opinion. I'm not sure that he should have. I just found the same thing, noted in an earlier post above.

 

[Mark Sirota]

How can I change the exposure of a .75 stop underexposed raw file by 3+ stops and not see clipping even though the colors (on the monitor, in Lightroom ProPhoto) are obliterated?

Perhaps because your monitor is incapable of displaying the colors? They're not clipped in the file, but they're clipped on your monitor?

 

[RikkFlohr]

Exposure pushes blacks too. A simple test of the slider +4 on an image with no clipped blacks can tell you this.
Brightness protects Blacks and Highlights both.

This is the second time I have had to struggle not to put words into Martin's mouth today but advice from several years ago (when LR had a different rendering engine) and was subsequently modified (your words) is probably less valid today.

 

[wernerg]

Perhaps because your monitor is incapable of displaying the colors? They're not clipped in the file, but they're clipped on your monitor?

That's certainly possible. My monitor does not have the full ProPhoto colorspace. Could that really be the problem? Do you really need a full ProPhoto color space monitor for the highlight clipping indicator to be useful? Was the raw file really 3+ stops underexposed? Does the first link show that?

RikkFlohr
Exposure pushes blacks too. A simple test of the slider +4 on an image with no clipped blacks can tell you this.
Brightness protects Blacks and Highlights both.

This is the second time I have had to struggle not to put words into Martin's mouth today but advice from several years ago (when LR had a different rendering engine) and was subsequently modified (your words) is probably less valid today.

I tried your test on a raw file that had no blacks or whites. The highlights and most midtones piled up on the right, as expected. But they left a long tail of blue that led almost but not quite back toward the darks to where they started.

I re-read Evening's explanation for how to use exposure, blacks, brightness sliders in his Lr3 book again. Basically use exposure to set highlights, blacks to set blacks, brightness to set midtones. I don't know why I can't do that with this image. Interestingly I can move the exposure .75 stops in Lr and get the same image I get in Capture NX2 moved .75 stops, but that's because I know .75 is the right number. Lr is telling me there is lots more room to go higher.

 

[RikkFlohr]

There is only room to go higher if your scene warrants it. The examples you post have very few tones I would consider highlights.It looks like an overcast day in a naturally shadowy area. The subject doesn't contain huge tonal value differential. Exposure is valuable to make White in your image white by the numbers-but in the absence of white in the scene it is to compensate for inadequacies in camera metering or the dynamic range of the scene.

In your scene you have a very low contrast range and kicking exposure past the point of matching neutral gray to 50% values in Lightroom is going to make for a garish image as you have found. I suspect if you had a gray card in this image for reference and matched it you might still be disappointed.

Sliding Exposure to the point of just barely avoiding clipping a color channel is great for holding white and the brightest highlights (provided they exist) but useful for little else.

 

[wernerg]

There is only room to go higher if your scene warrants it. The examples you post have very few tones I would consider highlights.It looks like an overcast day in a naturally shadowy area. The subject doesn't contain huge tonal value differential. Exposure is valuable to make White in your image white by the numbers-but in the absence of white in the scene it is to compensate for inadequacies in camera metering or the dynamic range of the scene.

In your scene you have a very low contrast range and kicking exposure past the point of matching neutral gray to 50% values in Lightroom is going to make for a garish image as you have found. I suspect if you had a gray card in this image for reference and matched it you might still be disappointed.

Sliding Exposure to the point of just barely avoiding clipping a color channel is great for holding white and the brightest highlights (provided they exist) but useful for little else.

Thank you Mr. Flohr!!! Now I understand why this has been an occasional thing, and mostly with flower pictures. But I never connected it to image content when there are often no whites or bright highlights. I have no issue with Lr being different than Capture NX2. I'm just trying to understand what those differences are.

Thank you again for your persistence and help.

Werner

 

[MarkNicholas]

Interesting. That answers one or 2 questions nigling in the back of my mind. It always troubled me if the histogram did not reach the extremeties after making adjustments but now I know why.

 

[wernerg]

Interesting. That answers one or 2 questions nigling in the back of my mind. It always troubled me if the histogram did not reach the extremeties after making adjustments but now I know why.

Mark, I found this on Luminous Landscape. note the first post by Michael Frye (MFRYE)

Extract;

In some cases boosting the midtones like this [exposure slider] is okay, but in many cases it flattens highlight contrast and washes out colors.

The "some cases" he refers to are those where there is distance between image's brightest highlights and white point of the histogram.

 

[Bruce J]

Perhaps because your monitor is incapable of displaying the colors? They're not clipped in the file, but they're clipped on your monitor?

We've been around this before on the forum, but I can't find the thread at the moment. I don't think this has much to do with the OP's monitor. In fact, the 'exposure' slider in LR does not adjust exposure as a photographer would think of it. Its primary function is to adjust the 3/4 tones (high values, but not the highlights). Look at the histogram while hovering over the exposure slider and you can see the shaded area which is where the primary adjustment is made. The amount of the adjustment tails off toward the lower tones and toward the highlights. All of the tonal adjustment sliders do the same; they are all more-or-less bell-shaped curves (most of the effect is in the middle of the shaded area; less towards the ends). If you want a true exposure adjustment, you will have to do it in curves. Raising the entire curve straight up to increase exposure would affect all parts of the image equally.

I struggled with this for the first several years that I used LR until I finally spent a day experimenting w/ a low-contrast image. I wanted to increase the exposure w/o increasing the contrast. The only effective way I found was to use a curve, as I mentioned. I only wish Adobe had picked a different name for the 'exposure' slider, as it has confused many users. I also wish they would include a true exposure slider; boy, would that make my life easier. That's my $.02; hope it's useful, but I suggest you do your own testing.

 

[RikkFlohr]

Here is a quick illustration and exercise in three images. The test image is a 1000 step gradient from 100% Black to 80% Black. The three images show: No Exposure Adjustment, 2 Stops of Exposure Adjustment and 4 Stops Exposure Adjustment. Remember that F Stops are doubling or halving of the amount of light.



When you increase or decrease exposures you are pushing tonal values into larger or smaller spaces. That is what creates that long trail that Werner is seeing. Notice that the clipped blacks are pulled from being clipped to a value of 2% in a 2 Stop Exposure Increase and to 6% RGB values in the 4 Stop exposure increase. I am not sure it is fair to say that the highlighted area on the interactive histogram controls primarily the area beneath which is highlighted. Even if there are no pixels in the range highlighted on mouse over, you can affect both ends of the histogram. Recovery will affect your shadows and Blacks will affect your highlights.

Degree of reactivity, of course depends upon the image content and original in-camera exposure.

It is a good exercise to prepare some gradient test files of various styles and push and pull them with the Tone tools. It helps you understand what is happening.

 

[wernerg]

I saw a similar test to Rikk's somewhere on the web while I was researching this topic yesterday and it is very instructive. However, it would be even more instructive to do the test with a color checker image to see what happens to colors when you increase the exposure slider by 2 and 4 stops. After you posted your response yesterday I went back to look at lots of old raw files in Lr. As you pointed out, if the image has a reasonably full range of tones from black to white, the histogram/exposure slider combo do what most people would expect. But I have lots of flower pictures taken in umbrella-shaded light that do not conform to that model and hence pushing the midtones with the exposure slider results in washed out colors long before any saturation occurs. I think Michael Frye's conclusions in the link I posted above are consistent with your analysis and what I am seeing in my low contrast images. Also, as Frye points out, if you set the white point with the tone curve, the image responds like a typical levels adjustment.

 

[Mark Sirota]

Look at the histogram while hovering over the exposure slider and you can see the shaded area which is where the primary adjustment is made..

Don't be fooled by that -- it's showing you the control area of the histogram. In other words, rather than dragging the slider, you could click and drag the histogram in that area to make the same adjustment. It's not telling you what part of the histogram it is affecting -- as Rikk demonstrated, it affects the entire histogram.

 

[Bruce J]

When you increase or decrease exposures you are pushing tonal values into larger or smaller spaces. That is what creates that long trail that Werner is seeing. Notice that the clipped blacks are pulled from being clipped to a value of 2% in a 2 Stop Exposure Increase and to 6% RGB values in the 4 Stop exposure increase.

Yes, that is what creates the blacks trail. However, look what's happening to the high end of your gradient. While the low end is going from 2% to 6%, the high end is going from 50% to 80%, which is a much larger increase. As you demonstrate, the effect is not linear. If you had included a full range of values in your test image, you would have seen the effect tail off somewhat at the high end also. I'm not claiming that the effect is limited to the shaded area of the histogram, just that it's largest in the shaded area. It actually tails off all the way to both ends; it has to to keep from creating artificial breaks in the histogram (and truly ugly images).

I am not sure it is fair to say that the highlighted area on the interactive histogram controls primarily the area beneath which is highlighted. Even if there are no pixels in the range highlighted on mouse over, you can affect both ends of the histogram. Recovery will affect your shadows and Blacks will affect your highlights.

Yes, I agree that all of the tonal sliders have some affect on all tone ranges; again, that's necessary to prevent gaps in the histogram. Nevertheless, the effect is still largest in the shaded ranges, even if there are no pixels there. It's just that you can't see it in the image (or on the histogram) if there are no pixels in that range.

Degree of reactivity, of course depends upon the image content and original in-camera exposure.

It is a good exercise to prepare some gradient test files of various styles and push and pull them with the Tone tools. It helps you understand what is happening.

I'm not sure what you mean by the degree of reactivity, but I don't think the effects of the tonal sliders change between images, whether because of different content or exposure. A certain amount of slider movement will always move pixels from 30% to 50%, in every image. I don't think LR cares whether there are pixels with those tones in the image or not; it always does the same thing. Which sliders you have to move and how much to get a pleasing image certainly depend on content and exposure; perhaps that's what you meant by degree of reactivity.

I certainly agree that doing the tests yourself, both on gradients and actual images, is very educational. The best way I know to get an understanding of what those controls do. So, here's a test for you: make a gradient image with values from, say 30% to 70% gray. This represents a low contrast image exposed on auto-exposure. Now, you want to make a high-key image out of it; that is, increase the exposure without increasing the contrast. Maybe you want to wind up with values of 50% to 90%, or 55% to 95%. That's what I think an 'exposure' slider should do. That is, after all, what would happen if you increased the exposure in camera. You can't do it w/ any of the LR tonal sliders because they don't act on the histogram linearly. The only way I know to do it in LR is with the curves control. Cheers,