Expose Left, or Expose Right (or neither?)

[Glenn NK]

Obviously I can't count (or can't read); two things to watch carefully and I listed three? Yikes.

[Carl in Louisiana]

Carl:

I'm late - on the west coast.

I haven't changed my contrast setting on my 30D for four or five years and I shoot everything with it. Same for my three year old 5DII.

Three things I do watch carefully:

1) because there can be so much variation in dynamic range in one image, the type of metering set in camera can be important. For extreme closeup/macro I often use spot metering for example.

2) particularly the camera histogram,

3) and of course I check the histogram in Lightroom when processing to be sure I have the right settings.

Occasionally I get an image that "leaks out" the right side of the histogram, but most of the time I can recover it in PP.

Glenn

Hey Glenn,

Can you expound a little more on number one please, perhaps with some examples.

Oh and who is counting, so long as I get the desired effects I'm looking for.

[Carl in Louisiana]

Okay, I went on the Google search and came up with tons of info on this subject. Lots of pros and cons. This site does have examples of what I really wanted to see and here it is for those that are interested. Right now I am open to comments either for or against. Also most that I have read suggest using spot metering with ETTR, focusing on the brightest subject in the shot (not the sky).
http://photography-on-the.net/forum/...1#post12699441

[GrahamH]

Hi,
Whether I expose R of L, depends on what is important in the image. If the darks are more important, then I prefer to slightly over expose to make sure I get the necessary detail. If that causes clipping - is the clipped area important? Can I use HDR or similar methods to get the image that I have in mind. Is this a one shot deal of can I bracket to my hearts content and assess at a later time.

In general I tend to prefer to ETTR as long as it is within stop of two of the average DR, but too many other factors come into play making it a general positioning rather than a hard and fast rule. Also bear in mind ETTR for a RAW file is NOT the same as ETTR for a JPEG file due to the greater DR range in RAW. I can clip up to a couple of stops (compared to a JPG) if necessary and still recover the information.
Basically the more control I have over the image, the better it is likely to be.
Graham

[Brev00]

It all depends. I think the greatest value of pushing the exposure to the right is reduction of noise. This concerns me when I am shooting landscapes with a lot of blue sky. I find that accurately or underexposing a sky typically leads to a pleasant color undermined by noise that becomes quite obvious in a black and white conversion. So, I spot meter the sky and add some exposure to keep the skies clean(er). When shooting my red roses, I must, on the other hand underexpose by at least a stop to prevent the red channel from clipping. Accurate color trumps noise in such situations. Then, there are times I underexpose to keep my shutter speed up in limited light. Getting the action frozen is more important than noise in this instance. There are other times when I don't need to fuss so much and I keep the meter in the middle. Rob Sheppard, the outdoor photographer, says that color reproduction is most true in this type of even exposure. So, I keep that in the back of my mind along with the rest as I confront the scene before me. All this being said, I find shooting raw is very forgiving and usually so am I.

[Carl in Louisiana]

Thank You Graham and Larry,

I do shoot in RAW and after the last post I have done some more research via Google and come up with some more very interesting in camera settings for high ISO settings that I picked up from the Canon forum and the poster added his images before and after PP and I must say they were very impressive over the noise that I have in my images. If I can come close to what I saw it will be a big improvement. I plan on trying them out and will post my results as I have them.

[Glenn NK]

Carl:

In regard to the metering method: for a uniformly bright image (plain white piece of paper), it doesn't matter what method is used. If there are "hot spots", then the spot metering may be the better method so as to avoid clipping them.

In any event, I actually rely on the histogram much more than I rely on the meter. Quite often during focus stacking, I run the camera in Manual, look at the metered results in Av or Tv, then manually set both f/stop and shutter. But I always take a test shot and check the histogram. My cameras' histograms are marked with vertical lines - each space corresponds roughly to one stop, making the next adjustment easier.

Returning to the ETTR/ETTL discussion - Larry's comment about signal to noise ratio is part of what ETTR is all about.

Does Rob Sheppard come from a film background ("colour reproduction is most true in this even type of exposure".)?

In digital I suspect that this is probably meaningless. The whole effort in digital is to capture the most amount of information. A brightly exposed digital image has more info than a dark one, and also has a higher signal to noise ratio - more signal, less noise - the shadows are full of noise.

The LCD is meaningless and dangerous in judging colour - even with JPEG images.

Glenn

[Colin Southern]

It's a topic that - unfortunately - stirs up much mis-information.

In short, what one NEEDS to do, depends on a number of factors, with the biggest one usually being the dynamic range of the scene that's being captured. If it's purely reflective (ie only around 4 stops) then it really doesn't matter how you expose it - but as the dynamic range of the scene increases, then correct exposure becomes more and more critical (as it does with increasing ISO).

Some always advocate "exposing to the right" so that they capture the most information, but - unfortunately - if it's a relatively low dynamic range scene then doing this can bite you in the bum because what usually happens is that one channel will be closer to clipping that the other two, and if this happens to push the levels into the sensors non-linear response region then you end up with a weird colour shift in the highlights that's nigh on impossible to correct in post-production.

Conversely, if one doesn't push the exposure of high ISO shots then the shadow data can get mixed in equal proportions with the noise floor resulting in significant amounts of visible noise.

So one really just has to know their camera - know their subject - and know the appropriate compromises that are best for any given situation. If in doubt, experiment.

As a general rule though, ETTR produces cleaner images only in THEORY - in practice, most modern cameras are capturing at least 12 stops of DR (at base ISO), and we're typically only using about 6 to 8 of them anyway - so we don't need the extra data in practice.

[Carl in Louisiana]

Carl:

In regard to the metering method: for a uniformly bright image (plain white piece of paper), it doesn't matter what method is used. If there are "hot spots", then the spot metering may be the better method so as to avoid clipping them.

In any event, I actually rely on the histogram much more than I rely on the meter. Quite often during focus stacking, I run the camera in Manual, look at the metered results in Av or Tv, then manually set both f/stop and shutter. But I always take a test shot and check the histogram. My cameras' histograms are marked with vertical lines - each space corresponds roughly to one stop, making the next adjustment easier.

Returning to the ETTR/ETTL discussion - Larry's comment about signal to noise ratio is part of what ETTR is all about.

Does Rob Sheppard come from a film background ("colour reproduction is most true in this even type of exposure".)?

In digital I suspect that this is probably meaningless. The whole effort in digital is to capture the most amount of information. A brightly exposed digital image has more info than a dark one, and also has a higher signal to noise ratio - more signal, less noise - the shadows are full of noise.

The LCD is meaningless and dangerous in judging colour - even with JPEG images.

Glenn

Thank You Glenn,
Your info is helping in the learning process. I came across this thread and in the back of my mind it tweaked something and when that happens I find myself digging probably to deep looking for an answer.
I have found one that is suitable for me at my level of knowledge about photography, which is not very much at this stage, to give it a go.
It is not to what I will call for me the extreme end of ETTR. So like Colin stated I will experiment with it at that level which is suitable for my knowledge and if I grasp it then maybe I'll look a little deeper into the subject.
Thank You all for your input as I have learned much from it and it has helped me to do some research on my own. I hope others have learned as much or more.
If what I use helps me in photography and improves my images I will be very glad to pass it on.

[Glenn NK]

OK, more reading - and two more that suggest ETTR, and why.

http://schewephoto.com/ETTR/index.html

http://www.lightstalking.com/why-sho...#ixzz1mQAzPQJr

After reading many articles about this, it appears that the link I provided previously (Luminous Landscape), had a major error. In the article by Michael Reichmann it was stated that one half of the information collected by the sensor is in the brightest stop (at the right of the histogram).

Apparently this is false.

HOWEVER, this does not invalidate the reasoning for using ETTR. The above two links clearly indicate that ETTR is the better approach. The proviso (of course) is that one doesn't blow out the highlights that are important (significant?).

To this end, the RGB histogram is the only useful tool - the luminance (single white colour) histogram is not of much use. Reason? Suppose you are shooting a red flower that fills all or most of the frame, and use only the luminance histogram - it may look fine. But when you look at the flower in Photoshop or Lightroom, it is obvious that the red channel is blown, and that there is lost detail. This is not surprising at all - the flower is red!! This has happened to me.

While there are reasons to ETTR, I can't find any good reasons to ETTL.

When I shot colour slide film, underexposure provided better colours when the slide was viewed on a screen - overexposure washed them out. A digital sensor is very different from film.

Glenn

[John Morton]

While there are reasons to ETTR, I can't find any good reasons to ETTL.

When I shot colour slide film, underexposure provided better colours when the slide was viewed on a screen - overexposure washed them out. A digital sensor is very different from film.

Glenn

As a matter of course, I tend to shoot a lot of HDR sequences since I am often out with my PC (shift/tilt) lens and a tripod.

When I look for a single capture out of an HDR sequence to work with, it is inevitably one which would fit into the category of "ETTL"; because these are the images which are least likely to have any kind of motion blur, since hte shutter speed used in those captures will be faster than the ETTR images.

[Carl in Louisiana]

OK, more reading - and two more that suggest ETTR, and why.

http://schewephoto.com/ETTR/index.html

http://www.lightstalking.com/why-sho...#ixzz1mQAzPQJr

After reading many articles about this, it appears that the link I provided previously (Luminous Landscape), had a major error. In the article by Michael Reichmann it was stated that one half of the information collected by the sensor is in the brightest stop (at the right of the histogram).

Apparently this is false.

HOWEVER, this does not invalidate the reasoning for using ETTR. The above two links clearly indicate that ETTR is the better approach. The proviso (of course) is that one doesn't blow out the highlights that are important (significant?).

To this end, the RGB histogram is the only useful tool - the luminance (single white colour) histogram is not of much use. Reason? Suppose you are shooting a red flower that fills all or most of the frame, and use only the luminance histogram - it may look fine. But when you look at the flower in Photoshop or Lightroom, it is obvious that the red channel is blown, and that there is lost detail. This is not surprising at all - the flower is red!! This has happened to me.

While there are reasons to ETTR, I can't find any good reasons to ETTL.

When I shot colour slide film, underexposure provided better colours when the slide was viewed on a screen - overexposure washed them out. A digital sensor is very different from film.

Glenn

Hey Glenn,

Thanks, I found the first link in my google search but not the second one. I did not mean to sound as if I was not going to use ETTR but that I was going to take small steps until I understand just what it is I am doing.
I did pick up some in camera settings from the Canon forum and just about the same settings from the fred miranda site. I have set my camera up to those setting and played around with them just a bit today with +1/3 ETTR. I was very impressed with the outcome of those shots. However I have some other areas that I have shot in before that I want to test these settings in. Those I will post on the forum when I get them shot and PP.
Thanks for the tip about the RGB histogram as I had read about it in one of the articles on one of the sites I visited. My camera is still set with the RGB histogram.
The in camera setting I am using are for the Canon 7D, 60D, T2i, T3i, ect. and are for those of us who shoot high ISO so far I am liking them very much but like I said I want to test them some more in different areas before making a final decision on them.
I spent most of the day going through my manuel and making all the adjustments to the camera. One thing is for sure I know my camera a lot better than I did before this all started and thats a good thing. I may be slow but I am steady!

[OldFrank]

Does anyone bracket these days. I notice that it hasn't been mentioned. If for no other reason it provides an excellent learning experience especially with raw. I often use a modified approach. I use the cameras spot metering, pick several points in the scene, and shoot with them as references. It's interesting to process each in LR4 and compare the differences.

frank

[John Morton]

Does anyone bracket these days. I notice that it hasn't been mentioned. If for no other reason it provides an excellent learning experience especially with raw. I often use a modified approach. I use the cameras spot metering, pick several points in the scene, and shoot with them as references. It's interesting to process each in LR4 and compare the differences.

frank

Frank, that reminds me so much of what I loved most about my last film camera, the Olympus OM-4 T, which had a metering system that allowed one to take up to eight individual spot meter reading for each exposure and lock these readings into the camera's exposure memory. The camera would then calculate the optimum exposure based upon these individual spot meter readings. It was always a lot of fun taking photographs with that camera.

[ajohnw]

Rather than ettr and ettl perhaps it's better to understand the effects and wonder just what the software in the camera is doing but that is rather difficult to determine. It can sometimes be estimated from technical reviews that show a camera's dynamic range. They vary from make to make. There are loads on dpreview. All similar cost cameras tend to be the same but vary in their position in the brightness range. This is mostly down to software and. it's easy to be miss led. There are even indications that cameras can say It's a Test Card.

It's a bit ridiculous to say sensor capture bright light more efficiently without mentioning in what way. To a first approximation sensor capture light linearly. That's one of their problems. We see logarithmically over an enormous range. A pixel on the other hand simply counts photons. More photon's and more output until it overflows. A pixel has inherent noise that builds up as an exposure is taken. As far as the output is concerned it's just the same as light and the rate it builds up at will vary from pixel to pixel. There are steps taken to minimise its' effects. Un exposed pixels for one that can be used to estimate noise levels. Dark frame subtraction is also possible and even variations on that. When a pixel has received little light the noise contributes more than if it had received a lot.. Sensors also have a gain control of one sort or another which shows up as ISO in the camera. A simple view of this is that it amplifies noise as well as the light level that has been captured. This is why a high iso shot with relatively low variations in light levels will show no noise but others will show noise in darker areas. Any effects other than these are down to software in the camera. Higher quality sensors tend to have less inherent noise. Pixel sensitivity to light is even likely to vary across the frame just as noise levels do..

What this thread is really talking about is what the camera manufacturer chooses to do with the output of the sensor. On that point even in film slr days all were designed to take decent shots of people with a certain amount of sky or grass in the background, I found Nikon excelled on that score. Best metering of the lot. Much argument at the camera club but some eventually agreed. It was also very very accurate which helps and more importantly predictable. Looked at this way it's hardly surprising the dslr's do similar things. Over expose and the highlights are gone - nothing for jpg engines to work on. The best thing to do is to spend time getting to know a specific camera. My Pen for instance has a tendency of leaving 2 stops at the dark end in raw - maybe it must give decent sky and subject exposures if it can in its jpg's. Adobe tends to bring out more highlight definition than the camera manufacturers do - camera curves, software etc. Never really gone into my 300D it's the most amazing more or less point and shoot camera I have ever used. 5D pass as well but that in my case is many for use on a microscope where things are decidedly different. Used normally it's a more difficult camera to use.

Histograms - think about the fact that the sensors record white light, even uv and ir fall into their scope if allowed to get there. They don't see red, the filter on a pixel looks after that aspect, They just record light. The bayer filter provides the colour information and uv and ir filters remove that problem. The histograms refer to jpg exposures which is why more can be obtained from a raw. Perhaps some camera could have a raw / jpg histogram option but the latter would mean that the exposure would have to be right making metering more difficult and in some scenes the light levels would exceed the capabilities of the sensor anyway. There are other factors as well. The raw development process uses interpolation - software again often involving camera response curves / colour profile files. RGB and Adobe - more software the bayer filters remain the same. A PC generally has more processing power than a camera allowing more complex software to do more to enhance a shot.

Really all of this comes down to knowing your camera, the type scene and exposing accordingly. One method isn't going to always produce the best results. Often the camera will select the best option but is probably optimised for what ordinary camera users shoot most often.

It's usually easy to see noise in a shot against exposure time and iso settings. Just take several exposure with the lens cap on taking care to not let any light in. Vary exposure and iso. There is no need to check raw on the cameras I use but software can do all sorts of strange things. This aspect seems to get worse and worse as pixel counts increase and as a result of that iso ratings too.

Atrophotographer make a lot of use of the sensors response to low light. Long carefully guided exposures used to be needed. Now they take a series of much shorter exposures and add them up. Used this way digital cameras can see much better then we can through telescopes. The only limitation is residual light in the "dark" sky as it isn't dark really and to a certain extent noise in the cameras sensor. They alternate dark frame shots to help with that and subtract them. There are some incredible shots about on the web that previously would have needed a huge telescope.

John

[Glenn NK]

Does anyone bracket these days. I notice that it hasn't been mentioned. If for no other reason it provides an excellent learning experience especially with raw. I often use a modified approach. I use the cameras spot metering, pick several points in the scene, and shoot with them as references. It's interesting to process each in LR4 and compare the differences.

frank

Yes, one setting on my camera takes three successive shots - underexposed, as metered, and overexposed. Although the original reason was to combine them, in most cases the images were at the ocean at sunset, so wave movement made this difficult. Now I pick the exposure that retains most of the highlights (excluding specular highlights) and recover as much of the shadows as possible in LR.

Glenn

[Carl in Louisiana]

Does anyone bracket these days. I notice that it hasn't been mentioned. If for no other reason it provides an excellent learning experience especially with raw. I often use a modified approach. I use the cameras spot metering, pick several points in the scene, and shoot with them as references. It's interesting to process each in LR4 and compare the differences.

frank

Yes I do! I like to play around with HDR images some and therefore the need for bracketing. I hope to being doing some in the near future.
Next month will be 'Cruisin" The Coast" about a 45 minute drive from here and I really like going over for that. Love the old cars and the memories they bring back. Not sure I'll be doing any bracketing there but will be looking for that opportunity.

[Colin Southern]

While there are reasons to ETTR, I can't find any good reasons to ETTL.

It all depends on the dynamic range of the scene Glenn -- if it's relatively low, with ETTR you can end up with highlights (usually biased towards one channel) getting into the sensors non-linear response region as it gets close to saturation. with a normal exposure of a reflective scene (ie around 4 stops of DR) most cameras will have in the region of 1.3 to 2 stops of "safety margin"; assuming 2 stops - and thus a 2 stop ETTR - you then have blacks being recorded as midtones - and they then have to be shifted back 2 stops in post production - which is fine in theory, but in practice, I've found that (a) they never quite look right and (b) if some of the highlights have crept into the non-linear response region then you'll get some weird colour shifts in the highlights that you'll never be able to correct properly.

So all that begs the question of "why bother to ETTR when you have a camera that's capable of capturing at least 12, and you're only using 6 (4 for the scene, and 2 stops of "wastage" from a "normal exposure" giving you a 2 stop safety margin)". To me, it just doesn't make any sense - especially when coupled with the fact that when one looks at the image as a whole either on screen or in a normal size print (assuming the monitor or print isn't - LITERALLY - 6 feet wide!) then you're not going to be able to resolve the noise anyway.

In my opinion - YES - ETTR has it's uses - infact it's practically mandatory when using high ISO modes or shooting scenes with a high DR - but for normal DR range photography ETTR is far more about advantages gained in THEORY than any that are gained in practice.

[arith]

I see some people advocating that you deliberately either over-expose or under-expose shots (in RAW), so that you can go back and either recover more shadow detail or cheat the sensor out of an extra stop of Noise Reduction, or to get more dynamic range, or some reason. Does anyone favor shooting either over or under-exposed on purpose? In what situations if any?

Also, I shoot Canon, but do I understand correctly that on Nikon, the light/exposure meter is reversed? (Left is bright and right is dark?)

Actually haven't had a chance to read the rest of the thread yet but: mostly I expose to the right. Now if I'm doing a portrait I only go over a little bit and recently I found night shots seemed to look better by exposing bang in the middle, but I don't know for sure.

It is easy to get it completely wronge though, as I found at a canal with brightly painted bits; exposing two stops over on the sky left me with problems.

[Carl in Louisiana]

Actually haven't had a chance to read the rest of the thread yet but: mostly I expose to the right. Now if I'm doing a portrait I only go over a little bit and recently I found night shots seemed to look better by exposing bang in the middle, but I don't know for sure.

It is easy to get it completely wronge though, as I found at a canal with brightly painted bits; exposing two stops over on the sky left me with problems.

Hey Steve,

In my research on this subject I found a forum that use's ETTR (sparingly ie. +1/3 - +2/3 ETTR) for shooting higher ISO and I must say with outstanding results, as the Gent had posted images he had taken as proof, and he has plenty of them. The settings I obtained are for Canon cameras (I listed above).

I shoot mostly indoors and without flash and that is why I am so interested in this thread and the material I found on other forums to study and learn the technics applied. I would like to bring my photography to the next level and think this route is it. I have made the necessary in camera adjustments and will be testing these along with the suggested PP steps. I hope to be posting some of these shots soon and hopefully with near results as I have seen others do.