Puzzled Over Blown Highlights

[dje]

I understand the concept of blown highlights but I'm puzzled by how they are treated in camera and in processing software. In the two shots below, there are some areas of cloud that are blown. The first shot shows what I'll call "normal exposure" whereas the second shot shows what happens when the brightness slider is wound down to -100 in PSE.

My question is this:

If the blown areas are given a value of 255,255,255 (white) in the jpeg file, why dont they turn grey in the second shot (eg a value of say 128,128,128) ? Doesn't the brightness control just reduce every pixel in the image by a certain amount ? Are these pixels given some sort of special tag which says to the processng software "leave these as they are" ?

Incidentally, these blown highlights were able to be recovered from the RAW file but I have not done so in the test images displayed here.

Cheers Dave

[devoh]

I believe it is because each pixel is made up of Red,Green,Blue values.. and you can blow out a single channel or multiple channels. So not all blow highlights will be white... Well at least that is my understanding.

I can easily blow out my red channel on my camera taking a picture of our red roses.. but it does not make the rose white.

-Dave

[vladimir]

Doesn't the brightness control just reduce every pixel in the image by a certain amount ? Are these pixels given some sort of special tag which says to the processng software "leave these as they are" ?

I think older versions of Photoshop did this. Mine (CS4) has "Use Legacy" dial, which you can check to make it work like this. (Does PSE have the "Use Legacy" option?) By default, though, it's now "smarter" than that. It works on mid-tones, [gradually] leaving darks and lights alone, so the blown highlights are not affected not because they are tagged, but because they are bright.

[FrankMi]

Hi Dave. According to what I have read, Brightness focuses its attention on the middle tones. When you make a positive Brightness adjustment, the brighter pixels are compressed and the darker tones are expanded. A negative Brightness adjustment reverses the effect on the photographs tones. The highlights are expanded and the shadows are compressed.

If you set the Exposure first, then, if necessary, Recover before you adjust Brightness, the Brightness slider is more likely to reduce contrast in either the highlights or the shadows and the image will have more ‘pop’. When you compress tones in part of the tonal range, such as compressing the highlights with a positive Brightness setting, you lose contrast among those tones. The highlights will be brighter, but they will also have less contrast than they otherwise might if Exposure was used instead. Likewise, when you make a negative Brightness adjustment, the photograph will darken but your shadows will appear flatter than if you used the Exposure slider with a negative value. Hope this helps!

[Colin Southern]

Hi Dave. According to what I have read, Brightness focuses its attention on the middle tones. When you make a positive Brightness adjustment, the brighter pixels are compressed and the darker tones are expanded. A negative Brightness adjustment reverses the effect on the photographs tones. The highlights are expanded and the shadows are compressed.

Hi Frank,

No - brightness moves everything by the same amount. The effect of that is most noticeable in the midtones though. It's a gamma adjustment that compresses/expands etc.

[vladimir]

... brightness moves everything by the same amount. The effect of that is most noticeable in the midtones though. It's a gamma adjustment that compresses/expands etc.

Respectfully, from here:

In normal mode, Brightness/Contrast applies proportionate (nonlinear) adjustments to image layer, as with Levels and Curves adjustments. When Use Legacy is selected, Brightness/Contrast simply shifts all pixel values higher or lower when adjusting brightness. Since this can cause clipping or loss of image detail in highlight or shadow areas, using Brightness/Contrast in Legacy mode is not recommended for photographic images (but can be useful for editing masks or scientific imagery).

[Colin Southern]

Respectfully, from here:

In normal mode, Brightness/Contrast applies proportionate (nonlinear) adjustments to image layer, as with Levels and Curves adjustments. When Use Legacy is selected, Brightness/Contrast simply shifts all pixel values higher or lower when adjusting brightness. Since this can cause clipping or loss of image detail in highlight or shadow areas, using Brightness/Contrast in Legacy mode is not recommended for photographic images (but can be useful for editing masks or scientific imagery).

Thanks Vladimir (and apologies to Frank) - someone changed the rules when I wasn't looking!

At least I'm in good company! ...

Understanding brightness and gamma

Brightness vs. Exposure

I might add that in practice, I often use the exposure slider to push the highlights as hard as I can (typically up to a stop) with many types of images (especially portraiture), and then bring the skin detail back using the brightness slider.

[dje]

Many thanks everybody for your comments. They have most definitely help me crystalise my thoughts on this question. As I see it, the situation is probably something like this

• any pixel with at least one colour value of 255 is regarded as blown (the software has no way of knowing whether the value is exactly 255 or greater)
• the brightness control in the current software does not change blown pixel values (PSE 9 doesn't appear to have any legacy method by the way). Presumably they are not changed because there's a very good chance that the colour makeup is wrong anyway.
• if you work with a RAW file in ACR, the exposure slider can recover some highlights that would be blown in a jpeg simply because of the extra resolution of the 11 or 12 bit A-D.

I think I'll go and have a beer now

Cheers Dave

[Colin Southern]

if you work with a RAW file in ACR, the exposure slider can recover some highlights that would be blown in a jpeg simply because of the extra resolution of the 11 or 12 bit A-D.

I'm not sure what the default behaviour for the exposure slider is for ACR when dealing with blown channels, but they do of course have the seperate recovery slider for these purposes. The best solution is to just not blow the channel in the first place; most cameras are capturing arounds 12 stops of dsynamic range, and yet we typically only see about the last 6 on our monitors. So it's far far far better to under-expose slightly to protect the highlights, and then use (primarily) the fill light slider to reveal some of the low-tone detail (accompanied by exposure and brightness sliders too).

[Donald]

Sorry folks, I'm just putting this post in to correct a glitch that's happened elsewhere on the forum that's diverting people away from this thread. Just ignore me.

[vladimir]

I think I'll go and have a beer now
Cheers Dave

Me too!
I wasn't 100% sure if my first speculation (about Brightness not touching clipped pixels because they are clipped) was correct or not, until I staged this experiment: I opened a new document in Photoshop, filled in with white (so all pixels are "clipped") and then moved Brightness slider. Nothing changed. Then I turned on Legacy Mode, and moved the slider again. White turned grey...
Now I can relax... And have some beer

[dje]

Thanks Vladimir for your interest. This confirms that the brightness slider normally does not change full white pixels.

Cheers Dave

[Hero]

I wasn't 100% sure if my first speculation (about Brightness not touching clipped pixels because they are clipped) was correct or not, until I staged this experiment: I opened a new document in Photoshop, filled in with white (so all pixels are "clipped") and then moved Brightness slider. Nothing changed. Then I turned on Legacy Mode, and moved the slider again. White turned grey...

And suddenly you got me curious as to what happens if you use values of say 254. I wonder if it really is only the blown highlights it skips or whether PS uses a stretched s-curve for brightness controls (in short the closer the pixel-value comes to 128, the more it will be affected by the brightness-control)

[Colin Southern]

Me too!
I wasn't 100% sure if my first speculation (about Brightness not touching clipped pixels because they are clipped) was correct or not, until I staged this experiment: I opened a new document in Photoshop, filled in with white (so all pixels are "clipped") and then moved Brightness slider. Nothing changed. Then I turned on Legacy Mode, and moved the slider again. White turned grey...
Now I can relax... And have some beer

Hi Vladimir,

On a side note, if you're using the brightness control in Photoshop instead of in ACR then you're not maximising the image potential (doing it in ACR preserves a lot more data).

[vladimir]

Hi Vladimir,

On a side note, if you're using the brightness control in Photoshop instead of in ACR then you're not maximising the image potential (doing it in ACR preserves a lot more data).

Thanks Colin,
I almost never use brightness control in PS or ACR (Lightroom in my case). Last time I used it (and that's when I appreciated the Legacy mode) was when I was working on the mask to select hair [of almost same tonality as a background]. And as far as I can remember - it was also the first time I used it.
But the point about preserving data by ACR and not by PS is debatable. I am very stingy when it comes to data. Hate loosing my bits of information. That's why I use 16-bit mode of editing in PS whenever possible. And the question then - do I loose any data in PS when I am in 16-bit mode? I know - I loose a lot of disk space, yes. But I don't think I loose any of my precious bits. ... And as far as non-destructive nature of the edits - again, if they are done in an adjustment layer - they are very much non-destructive, right?

PS: Disclaimer: I am NOT trying to discourage anybody from using ACR controls. 99% of my shots never see PS at all. ACR/Lightroom is great, and this argument is strictly theoretical

[Colin Southern]

But the point about preserving data by ACR and not by PS is debatable. I am very stingy when it comes to data. Hate loosing my bits of information. That's why I use 16-bit mode of editing in PS whenever possible. And the question then - do I loose any data in PS when I am in 16-bit mode? I know - I loose a lot of disk space, yes. But I don't think I loose any of my precious bits. ... And as far as non-destructive nature of the edits - again, if they are done in an adjustment layer - they are very much non-destructive, right?

Hi Vladimir,

Not really debateable I'm afraid - the image is still in linear gamma when it's in ACR, but when it's passed through to PS gamma conversion is applied, and a lot of information is discarded. This is fine & normal, but you don't want to be compressing / expanding tonal ranges any more than absolutely necessary past this point. Additionally, ACR applies all changes as a single adjustment to the conversion, whereas once you're in PS, the edits are cumulative.

In reality there is of course a safety margin, but best practice is generally considered to be "anything we CAN do in ACR we SHOULD do in ACR).

[dje]

And suddenly you got me curious as to what happens if you use values of say 254. I wonder if it really is only the blown highlights it skips or whether PS uses a stretched s-curve for brightness controls (in short the closer the pixel-value comes to 128, the more it will be affected by the brightness-control)

Hero I tried a similar test to Vladimir but with a value of 254 and the point on the histogram did move with the brightness slider. Then I changed the value to 200 and the point moved further along the histogram with the same amount of brightness slider movement.

Cheers Dave

[vladimir]

And suddenly you got me curious as to what happens if you use values of say 254. I wonder if it really is only the blown highlights it skips or whether PS uses a stretched s-curve for brightness controls (in short the closer the pixel-value comes to 128, the more it will be affected by the brightness-control)

Yep, good question, Hero,
Here are some measurements.
First number - initial pixel value
-- Second number - pixel value after [max] -150 brightness adjustment in Normal mode
---- Third number - pixel value after [max] -100 brightness adjustment in Legacy mode

255 - 255 - 174
254 - 217 - 172
253 - 201 - 171
252 - 190 - 170
....................
245 - 157 - 162
....................
235 - 135 - 150
....................
138 - 63 - 22
....................
128 - 58 - 4
....................
118 - 53 - 0

So, what we have is more or less linear behavior for Legacy mode, and very much non-linear behavior for Normal mode. I wouldn't try to make much sense out of the actual values, because that would lead us to very difficult (at least for me) to understand topics like gamma, rendering intents, color spaces etc... The shape of the curve seems to be more like stretched N than like stretched S.

[vladimir]

Hi Vladimir,

Not really debateable I'm afraid - the image is still in linear gamma when it's in ACR, but when it's passed through to PS gamma conversion is applied, and a lot of information is discarded. This is fine & normal, but you don't want to be compressing / expanding tonal ranges any more than absolutely necessary past this point. Additionally, ACR applies all changes as a single adjustment to the conversion, whereas once you're in PS, the edits are cumulative.

In reality there is of course a safety margin, but best practice is generally considered to be "anything we CAN do in ACR we SHOULD do in ACR).

Thanks Colin,
This is a really good advice. As I already mentioned - understanding gamma is one of my weak points, but the advice is simple enough for me to follow.