Lr2

[jeeper]

Originally Posted by ronocnikral

"It’s always best to nail an exposure"

...That's what I was thinking. What's wrong with just exposing properly? Why over/under expose...?

[Imagebuffet]

Originally Posted by jeeper

...That's what I was thinking. What's wrong with just exposing properly? Why over/under expose...?

Sometimes, the dynamic range of a scene is too great, and one must choose what will be in correct exposure. Other times, one might wish to emphasize a region with the exposure.

I don't know about "always" best to nail the exposure. In my amateur realm, I have seen a wide variety of opinion of what "correct" exposure is.

[jeeper]

Well, it's up to you to decide what "correct" is. If the dynamic range of the scene is beyond the latitude of whatever medium you are using, meter the shadows, meter the highlights - then decide which you're willing to loose.

...That, or HDR (just don't make it look like a cartoon, lol).

[jeffkohn]

I don't know why the article says that JPEG is not linear.

Because they're not. They're encoded in a gamma 2.2 color space. The gamma encoding happens as part of the process of converting a raw file into an RGB image, either in-camera if you're shooting JPEG, or by the raw converter if shooting raw.

jeffkohn added 13 Minutes and 57 Seconds later...Double Post Merged Below

Originally Posted by jeeper

...That's what I was thinking. What's wrong with just exposing properly? Why over/under expose...?

The correct exposure is the exposure that the photographer wants. As to why you might want overexpose, one reason is to get cleaner files.

A lot of people get confused about what ETTR is and when it should be used, there's quite a bit of misinformation out there on this topic. ETTR is for when you've got extra headroom because the contrast of the scene is smaller than the dynamic range of the camera. Let's say your scene has a contrast range of 5 stops, but you know your camera can easily handle 6-7 stops. The ETTR approach would be to over-expose by 1-1.5 stops when shooting, and then shift the exposure slider back by the same amount in the RAW converter to get a correctly exposed image. The benefit of this will be less noise in the shadows and darker mid-tones than if you had just used the 'correct' exposure in-camera.

When you have a very high contrast scene and you expose so that your highlights are just short of clipping, that's not necessarily ETTR. You could still have too-dark mid-tones and blocked-up shadows, in which case shifting the exposure back to the left in the RAW converter would make no sense. That's exposing for the highlights, which is a different approach to exposure.

[jeeper]

Sorry, but I don't really see the difference.

If you under-expose by one stop, then push it one stop in PP - what's the difference between that and just doing it right the first time?

I think there might be a little confusion about what I'm saying...

I am NOT saying to just mindlessly zero out the meter. I AM saying that you should know what you want the end result to be, meter, and expose accordingly.

That underexposing, and then pulling it back creates less noise is news to me...
I tend to think that the less PP you do, the better - as far as noise goes.

"Correct" could vary from person to person - but as long as you get what you wanted, you succeeded.


BTW - what is ETTR? I'm sure I know the theory, I just don't know what the letters stand for...

EDIT

Nevermind - Ettr= Expose to the right, right?

I'm of the school of thought that you should just expose for what you want to expose for...

[Imagebuffet]

Originally Posted by jeeper

Sorry, but I don't really see the difference.

If you under-expose by one stop, then push it one stop in PP - what's the difference between that and just doing it right the first time?

I think there might be a little confusion about what I'm saying...

I am NOT saying to just mindlessly zero out the meter. I AM saying that you should know what you want the end result to be, meter, and expose accordingly.

That underexposing, and then pulling it back creates less noise is news to me...
I tend to think that the less PP you do, the better - as far as noise goes.

"Correct" could vary from person to person - but as long as you get what you wanted, you succeeded.

This goes more to the physical nature of electronics than to the nature of photography per se. A CMOS sensor is a collection of transistors that are sensitive to light. One thing about transistors is that they have a leakage current. If the signal they are supposed to produce is large, it drowns out the noise signal, but when it is small, the noise becomes a significant part of the total signal. There are other factors that produce noise, too; shot noise and thermal noise are the two I recall from my electronics classes. Anyway, the bottom line is, one can generally drown out noise by having a large signal, and this is particularly important before amplification stages (because the amps would amplify the noise along with the signal). Hence, overexposure (large signal) is apt to produce less noise than underexposure. At least, that's my take on it. I studied the general components used in digital cameras in my Laser Electro-Optic degree, especially sensors and lenses, but I never took a specific class in camera engineering.

PP itself probably won't create noise in a digital file, because the error correction on digital bits is pretty good. Running a file through a program that intentionally degrades the image (e.g., JPEG compression) will add noise to the file, but simply editing the file shouldn't.

[jeffkohn]

Originally Posted by jeeper

Sorry, but I don't really see the difference.

If you under-expose by one stop, then push it one stop in PP - what's the difference between that and just doing it right the first time?

Under-exposing in-camera and then pushing it back up in the raw converter doesn't make much sense, I was talking about the opposite. Assuming you have the headroom, overexposing in-camera and then pulling the exposure back in the raw converter can minimize noise and give you a cleaner file.

I think there might be a little confusion about what I'm saying...

I am NOT saying to just mindlessly zero out the meter. I AM saying that you should know what you want the end result to be, meter, and expose accordingly.

I don't disagree with that, it's just that sometimes the end result that I want is pushed to the right right so that I can pull it back in post.

That underexposing, and then pulling it back creates less noise is news to me...

No, that would increase noise. But doing the opposite...

BTW - what is ETTR? I'm sure I know the theory, I just don't know what the letters stand for...

Sorry, I typed out 'expose to the right' in my first post, but got lazy in subsequent posts and just used the acronym.

[jeeper]

Originally Posted by jeffkohn

I don't disagree with that, it's just that sometimes the end result that I want is pushed to the right right so that I can pull it back in post.

So, why not just do it all in the camera? Maybe it's just me, but the less I have to do in post, the better...

[Imagebuffet]

Originally Posted by jeffkohn

Because they're not. They're encoded in a gamma 2.2 color space. The gamma encoding happens as part of the process of converting a raw file into an RGB image, either in-camera if you're shooting JPEG, or by the raw converter if shooting raw.

My grasp of color spaces is very weak. I do not know what a gamma 2.2 color space is, or how it is achieved. I have read the general algorithm for making a JPEG, and I recall something about a non-linear mathematical conversion taken from adjacent pixels. I imagine that is what is causing the non-linearity in a JPEG image.

Originally Posted by jeffkohn

A lot of people get confused about what ETTR is and when it should be used, there's quite a bit of misinformation out there on this topic.

I've never read any of it, and I know nothing of it.

Edit: Oh, "Expose To The Right." OK, I've heard of that, anyway.

Originally Posted by jeffkohn

ETTR is for when you've got extra headroom because the contrast of the scene is smaller than the dynamic range of the camera. Let's say your scene has a contrast range of 5 stops,

I don't know how I would know this.

Edit: I suppose I might do some math with a light meter, but that would be awkward. I mostly do landscapes.

Originally Posted by jeffkohn

but you know your camera can easily handle 6-7 stops.

I do know that the old Canon 10D can handle 8.5 stops, and the 50D can handle about 10.5 stops. Some Canon cameras can handle 11 or more stops of dynamic range.

[jeffkohn]

Originally Posted by Imagebuffet

My grasp of color spaces is very weak. I do not know what a gamma 2.2 color space is, or how it is achieved. I have read the general algorithm for making a JPEG, and I recall something about a non-linear mathematical conversion taken from adjacent pixels. I imagine that is what is causing the non-linearity in a JPEG image.

It has nothing to do with JPEG compression. TIFF's, PSD's, PNG's, etc are also non-linear. Maybe this will help explain. Basically gamma encoding is a way to compensate for the fact that the displays we use are non-linear, and the way our eyes see is also non-linear.

I don't know how I would know this.

Either by spot-metering elements of the scene, or taking a test shot and looking at the histogram distribution.

Edit: I suppose I might do some math with a light meter, but that would be awkward. I mostly do landscapes.

I also shoot mostly landscapes. ETTR is not that often useful, unless shooting in overcast light or other low-contrast situations. For scenic type shots (especially when the sky is included in the image), the contrast range is usually too great for ETTR.

I do know that the old Canon 10D can handle 8.5 stops, and the 50D can handle about 10.5 stops. Some Canon cameras can handle 11 or more stops of dynamic range.

I think the claims of 10, 11, or even 12 stops of dynamic range are kind of silly unless you're only interested in shooting test charts. For real world images, the amount of useful dynamic range in even the best DSLR's is about 8-9 stops IMHO. Beyond that, any detail you think you might see in the shadows is probably indistinguishable from noise. (The Fuji DSLR's with the dual-sensitivity sensors are the only real exception).

jeffkohn added 4 Minutes and 21 Seconds later...Double Post Merged Below

Originally Posted by jeeper

Maybe it's just me, but the less I have to do in post, the better...

That's fine. Different people have different preferences and priorities. Part of it has to do with image volume, too. My workflow allows me to get what I feel are the best results possible from my images; but if I were trying to make a living shooting weddings or portraits, my workflow would put me in the poor-house.

[ronocnikral]

Originally Posted by jeffkohn

...it's just that sometimes the end result that I want is pushed to the right right so that I can pull it back in post.

this is the key here. to me, "nailing" the exposure is getting the exact desired result straight out of the camera. if you want to expose to the right for the reasons above, then that is control. if you expose to the right all the time b/c you heard someone say you should, that is not control and results will vary greatly.

there are tools and techniques for everything, but doing them mindlessly without knowing what you are doing is not sound photography technique.

thanks for sharing all the information.

[Cheermom]

Originally Posted by Cajungaltx

okay... i just found some information... that actually says we are wrong (for shooting underexposed v. overexposed) in how we are handling it.

LINKY HERE:


Raw is different than JPEG

As many of these photographers graduated to shooting raw, many of them continued to intentionally underexpose to protect highlights. In fact I’ve heard respected photography teachers tell their students to do this. However, raw files are very different than JPEG files. They not only contain a huge amount of information, but due to the linear nature of a raw file, much of it’s capture bandwidth is dedicated to the highlights regions. In fact, with a standard raw capture 50% of the file’s information is used to record the brightest stop. To better understand this, take a look at the graph below.
The linear nature of a raw file means that as each stop gets darker, 50% less information describes it.

The linear nature of a raw file means that as each stop gets darker, 50% less information describes it.

This graph shows the six stop exposure range of the typical raw file with a total of 4096 distinct tones, ranging from black to white. Stop #1, the brightest stop, contains 2048 tones, or 50% of the total tonal bandwidth contained in the file. Stop #2 contains half as much as stop #1, which is 1024. This halving of tonal ranges continues with each successively darker stop.



As you can see by the histogram in the graph, this file was intentionally underexposed using the JPEG exposure philosophy. This results in half of the information available to describe this file not being used because it wasn’t used to describe highlights that weren’t captured. To make matters worse, only 1.6% of the file’s bandwidth is used to describe the darkest shadows.

This is a problem because when less information is used to describe shadow detail, noise becomes more of a problem. This noise is accentuated when the histogram is moved to the right in post-processing to lighten, compensating for the underexposure. If the file had been properly exposed, (moving the histogram one stop to the right), not only would all of the bandwidth have been used to capture the highlights, but the shadows would have moved one stop to the right where twice as much information would have been used to describe them.-

That's an interesting read right there.

[venchka]

...an example of blown highlights. 'Nuff said?

http://www.texasphotoforum.com/forum...d.php?t=104999

[Dobick]

This discussion got way more technical on the ones and zeros than I can successful contribute to, but I can say that one situation which often times requires "underexposure" would be night time sports shooting. The reason for under in this situation is that you are pegged out on ISO, and you have your aperture as wide open as it will go, but you need 1/400 (or faster) to really stop the action crisply, then a consideration might be made to underexpose by as much as a stop, and push it up later in post processing.

That has nothing to do with highlights vs shadows, and everything to do with the physics of fast moving objects...