good histogram = dark images in surf pics...

[jjj]

I think people are over complicating things here. Learn how to manually expose correctly for your own purposes and equipment and you do not have to use exposure compensation.
As for your audio analogy how would you you know what levels to set if you are recording in situations where the volume varies as much as light levels do? It's not strictly analogous anyway as you can capture a shot with two very different exposures and still get a good photo. They may differ certainly, but both can still be good. There only tends to be a good recording and a bad recording, there's not really an equivalent of a high key and a low key image in recording. In music yes, but not the recording itself.

When using slide which is way less forgiving of exposure variation that raw files, I'd simply take an exposure off my hand and add 2/3rds of a stop and got slides just how I liked them. Very simply, very effective. And if you care enough about exposure for outdoor work you should learn the Sunny 16 rule and learn to judge by eye. Quite useful.
Just tested it in fact, I took some shots out of my office window, firstly by guessing exposure [iso 100, 1/100 at f9] and was spot on, then I used auto and moved shot several different framings of scene. Auto [TV] gave me 1/100 at f7.1, f9, f11, f13 despite the light not changing in the slightest. Scene was mostly trees, hills, a few houses and some sky. This is why I usually can't be bothered with auto, too much faff, not less faff.
Auto exposure is hardly likely to have improved since the 80s as the same problems will always be there, the camera cannot know what you require from a scene in exposure terms, just like a camera does not know where you want point of focus to be. It can make a good guess at both and for mundane shots it's probably fine, most of the time. The only AF I've ever really been impressed with was the one on the Canon EOS 3, because it was eye controlled and worked really well. A shame it's not on more modern cameras, same goes for the clever spot metering on the OM4/OM3.

[hjulenissen]

I think people are over complicating things here. Learn how to manually expose correctly for your own purposes and equipment and you do not have to use exposure compensation.

With photography software and cameras alike working actively to not tell you what is going on (and to mask issues like excessive noise or clipping), it can be hard to get feedback on what you are doing.

Of course, if the end-result is "good enough" and that is what you are after, then everything is fine.

As for your audio analogy how would you you know what levels to set if you are recording in situations where the volume varies as much as light levels do?

With digital, you try to maximize levels while allowing for some headroom to avoid clipping. I guess it comes down to experience and gut-feeling based on some level of technical understanding (similar to photography?).

It's not strictly analogous anyway as you can capture a shot with two very different exposures and still get a good photo. They may differ certainly, but both can still be good. There only tends to be a good recording and a bad recording, there's not really an equivalent of a high key and a low key image in recording. In music yes, but not the recording itself.

I think that one exposure is "ideal" in the sense that it maximize information about the scene. There are modifications of this, depending on how much motion blur/DOF/diffraction limits your choice in exposure. For a perfectly still, softly lit scene and using a tripod, there is a wide range of exposure times that could be chosen, but only one that makes the highlights barely touch the sensor saturation level.

Just like in the case of music, (for the limited scenario drawn), this is what you want. And just like music, allowing for some headroom makes it a lot easier to avoid blowing the whole thing.

One critique of ETTR (as I have understood it) is that exposure tends to be about optimizing many (partially) conflicting goals, and that putting too much emphasis on one is counter productive.

Auto exposure is hardly likely to have improved since the 80s as the same problems will always be there, the camera cannot know what you require from a scene in exposure terms, just like a camera does not know where you want point of focus to be. It can make a good guess at both and for mundane shots it's probably fine, most of the time. The only AF I've ever really been impressed with was the one on the Canon EOS 3, because it was eye controlled and worked really well. A shame it's not on more modern cameras, same goes for the clever spot metering on the OM4/OM3.

A camera cannot know how much motion blur or how much DOF you want. Those are artistic choices, and they somehow have to be input (or guessed). Within those confines, a camera can pretty safely assume that the user does not want to have visible sensor clipping or excessive noise.

-h

[Petrus]

Auto exposure is hardly likely to have improved since the 80s


same goes for the clever spot metering on the OM4/OM3.

There has been massive improvement in autoexposure systems since eighties. At that time there basically were only full screen averaging, semi spot metering or fixed zones (middle, 4 segments around it, usually the top one slightly damped). Now cameras can have many variable segments with (semi) intelligent guessing about the subject, and focus point can be given extra weight in exposure calculation. I shoot 95% automatic now, if the exposure is off, I usually rather correct with exposure compensation than turn to manual. I might dial in some compensation even before starting to shoot when I know AE will not nail it.

About Olympus OM3/OM4: YES, it had the best (spot) metering of any film era camera. Place the spot on the light-side cheek, adjust the exposure to +2/3, Bingo!

[nma]

The answers to the seemingly simple questions regarding " good histogram = dark images in surf pics..." have become very confusing to me. DigitalDog seems to be arguing that there is no absolute in evaluating exposure because it is intertwined with the raw converter. Others argue that Rawdigger can tell you something important. Still others rely on rules of thumb left over from the days of slide film.  And still others rely on the "blinkies" derived from the in-camera jpeg. Then there is the claim that the histogram is not very useful, perhaps even misleading.

Here are some PRACTICAL arguments for further discussion:

1. If you have a good histogram using a linear tone curve (in your raw converter), meaning the dynamic range of the scene is encompassed so the elements fall between 1 and 255 (or whatever the number of elements on the abscissa) then it should be possible to adjust the levels in the raw converter to achieve an optimal rendering. I want to add that the highlight end of the histogram should be close to the far right, otherwise it is not a good histogram. No clipping.

2. If you have a good histogram and the image is dark when you import into your raw converter, this does not indicate a fault in your exposure.  This means the raw converter defaults are not optimal for rendering your image.  Merely adjust the levels of the converter to taste.

3. The "blinkies" are an indicator but since they rely on the jpeg and the in-camera jpeg settings, they are not a reliable indicator of clipping in the raw image.

4. Someone should explain how you can have a "good histogram" using a linear tone curve and not have the optimal exposure possible given the lighting conditions. I do understand that you could purposely deviate from that exposure level to emphasize the shadow values, for example, but that is not what I mean by an optimal exposure.

[digitaldog]

I think people are over complicating things here.

No, you just find it complicated. It's really exposure 101, you're probably not there yet. Re-read what hjulenissen wrote, it's spot on:

I think that one exposure is "ideal" in the sense that it maximize information about the scene.

Exactly! ETTR is the beginning of the process of producing that. What's the actual sensitivity of the sensor and the role of ISO settings? What do I have to do from what the meter recommendation (again for a JPEG or film) to produce maximum information in the raw? Simple exposure and development testing as we did when shooting film. Nothing complicated about it, it's part of photography from the day it was invented!

There is one optimal, correct, ideal exposure for any scene with of course input by the image creator (espeically if the scene's DR is wider than what the camera can capture). ETTR is simply about providing that by ignoring the incorrect meter/expsoure/development (for JPEG) seen on the camera histogram. The 'to the right' means compensate because you're under exposing the raw data by correctly exposing for the JPEG. If you have a false understanding that the two should be treated the same, you fail to understand the differences in the data, encoding, oh, I'm probably going over your head, sorry. Bottom line is, just as you would never treat a neg 'rated' at ISO 400 the same as a chrome 'rated' at ISO 100 until you tested exposure+development to produce the best results, you don't do the same with digital capture.

Now if you want to point a meter, reflective or incident and treat film, JPEG, raw, any ISO, and the development the same and expect the same results, you clearly don't understand exposure and development, again photography 101.

ETTR should just be called 'proper exposure for raw data' or whatever you want to name it. The only so called conflicting goals is being forced to under expose (and over develop, at the expense of data qualty) because you don't have enough light to fully embrace ETTR, correct exposure. In the film days we DID have to do this and we did see the result; more grain. Stopping the action or producing better data, any photographer will also understand they have to push the process, grain/noise as the net result. Lesser of two evils and again, photography 101.

There has been massive improvement in autoexposure systems since eighties.

The tools may be better, but the photographer who doesn't understand what they are being told and follows the advise incorrectly are not aided here. The better metering you speak of isn't going to help the raw shooter if it's based on the JPEG and they do not implement ETTR.

[digitaldog]

The answers to the seemingly simple questions regarding " good histogram = dark images in surf pics..." have become very confusing to me. DigitalDog seems to be arguing that there is no absolute in evaluating exposure because it is intertwined with the raw converter.

Not quite. I think RawDigger, which I downloaded, looks very interesting but I don't think you have to use it to produce your goals of optimal exposure and development. I'm suggsting too you take baby steps and try nailing exposure + development in the raw converter of you choice. That may make a huge difference in quality.

1. If you have a good histogram using a linear tone curve (in your raw converter), meaning the dynamic range of the scene is encompassed so the elements fall between 1 and 255 (or whatever the number of elements on the abscissa) then it should be possible to adjust the levels in the raw converter to achieve an optimal rendering. No clipping.

First I'd shy away from the Histogram in LR and instead using the clipping overlay technique outlined in my video and elsewhere. Here you can SEE all the pixels that clip both tone and saturation. You may very well decide some areas should be clipped and you have every right to. Again, I'd bracket some scenes, ideally under controlled lighting and see how much compensation or ETTR off the recommend exposure you need. Outlined here: http://www.digitalphotopro.com/technique/camera-technique/exposing-for-raw.html

2. If you have a good histogram and the image is dark when you import into your raw converter, this does not indicate a fault in your exposure. This means the raw converter defaults are not optimal for rendering your image.  Merely adjust the levels of the converter to taste.

It could indicate a falut in your exposure, it depends on the current rendering settings. With ETTR, you'll mostly see images that appear at least initially to be too light. You normalize (Michale's term and a good one) the development in the converter to get what you desire. IF you clipped higlights you wanted to retain, you did over expose!

3. The "blinkies" are an indicator but since they rely on the jpeg and the in-camera jpeg settings, they are not a reliable indicator of clipping in the raw image.

Correct! The data is vastly different hence so is how you expose that data.

4. Someone should explain how you can have a "good histogram" using a linear tone curve and not have the optimal exposure possible given the lighting conditions. I do understand that you could purposely deviate from that exposure level to emphasize the shadow values, for example, but that is not what I mean by an optimal exposure.

I think you are putting a bit too much emphasis on histogram. I don't know what a "good histogram' should look like but I know what I want my images to look like. In my histogram video (did I leave a link?), I show how histograms do not indicate an image is good or that a histogram some would examine alone tells them the image isn't good is indicative of the actual image. The histogram in your converter gives you an idea of the data at whatever stage of the processing you're at so it's difficult to suggest you keep an eye on at all times. You do want to keep your eye on the image itself, on a well calibrated and profiled display. And again, the clipping overlays we have in Photoshop, ACR and LR, and I'm sure other products is far more indicative of the clippings affect on the actual image than the Histogram.

[leuallen]

same goes for the clever spot metering on the OM4/OM3.

Olympus still has this, at least on the EM1. You have spot (regular), spotHI (for highlights) and spotSH (for shadows). I think it is on the EM5 also.

Larry

[bjanes]

In my histogram video (did I leave a link?), I show how histograms do not indicate an image is good or that a histogram some would examine alone tells them the image isn't good is indicative of the actual image. The histogram in your converter gives you an idea of the data at whatever stage of the processing you're at so it's difficult to suggest you keep an eye on at all times. You do want to keep your eye on the image itself, on a well calibrated and profiled display. And again, the clipping overlays we have in Photoshop, ACR and LR, and I'm sure other products is far more indicative of the clippings affect on the actual image than the Histogram.

Andrew,

Your histogram movie is excellent (as are your other recent tutorials), but on thing that you did not point out is the limitation of luminance histograms for determining clipping in the red and blue channels. This limitation is explained by Sean McHugh in his histogram tutorial. The composite RGB histogram does not keep track of the status of individual pixels, whereas the luminance histogram does as explained in the Cambridge tutorial.  Since the red and blue pixel values account for only 30% and 11% of the luminance in any given pixel, they can be 255 and the average including the green (59% of the luminance) can be well below the 255 clipping level.

Regards,

Bill

[Vladimirovich]

3. The "blinkies" are an indicator but since they rely on the jpeg and the in-camera jpeg settings, they are not a reliable indicator of clipping in the raw image.

indeed they are not off the shelf - but in reality I can tune such indicators for both E-M1 and A7 to show raw clipping within 1/3 EV precision... granted OOC JPG becomes unusable

[digitaldog]

This limitation is explained by Sean McHugh in his histogram tutorial.

Thanks Bill. That URL is actually referenced in the video.

[jjj]

There has been massive improvement in autoexposure systems since eighties. At that time there basically were only full screen averaging, semi spot metering or fixed zones (middle, 4 segments around it, usually the top one slightly damped). Now cameras can have many variable segments with (semi) intelligent guessing about the subject, and focus point can be given extra weight in exposure calculation. I shoot 95% automatic now, if the exposure is off, I usually rather correct with exposure compensation than turn to manual. I might dial in some compensation even before starting to shoot when I know AE will not nail it.

You missed the end part of my quote off which changes the meaning enormously.
"Auto exposure is hardly likely to have improved since the 80s as the same problems will always be there, the camera cannot know what you require from a scene in exposure terms"
Not to mention that I tested my modern camera's auto exposure and it was rubbish. As the results I listed above demonstrate. But then I've never been impressed by Canon's metering full stop.
Why faff with correcting, just set the right exposure and shoot away, unless light is dodging behind clouds exposure stays pretty constant outdoors and indoors [nightclubs excluded] it's constant so unless subject in moving in an out of light sources, just set the correct exposure and shoot away.
To illustrate what I mean, the shots I took of the view from office gave me 4 different exposures with auto despite the scene's lighting not changing one iota. So to get a better result, I'd have to first decide what the correct exposure was and then adjust each reframe of composition by a different amount of exposure compensation. If shooting manually, I simply set correct exposure and shoot away, much easier as all I need to consider is the composition.
Alternatively I auto meter from where I know will give the correct exposure and recompose, rinse and repeat for every shot. Still faffier than manual.

[jjj]

Olympus still has this, at least on the EM1. You have spot (regular), spotHI (for highlights) and spotSH (for shadows). I think it is on the EM5 also.

Not quite as multiple/averaged spot metering was also part of it, not to mention the really ergonomic buttons, whereas I think the spotHi and SpotSh are more alternate version of the spot meter mode. Not played with one long enough to even discover it had that feature, so not sure how it works, but from what I can find out, they seem to be modal which is a retrograde step.

[jjj]

No, you just find it complicated. It's really exposure 101, you're probably not there yet. Re-read what hjulenissen wrote, it's spot on: Exactly! ETTR is the beginning of the process of producing that. What's the actual sensitivity of the sensor and the role of ISO settings? What do I have to do from what the meter recommendation (again for a JPEG or film) to produce maximum information in the raw? Simple exposure and development testing as we did when shooting film. Nothing complicated about it, it's part of photography from the day it was invented!

There is one optimal, correct, ideal exposure for any scene with of course input by the image creator (espeically if the scene's DR is wider than what the camera can capture). ETTR is simply about providing that by ignoring the incorrect meter/expsoure/development (for JPEG) seen on the camera histogram. The 'to the right' means compensate because you're under exposing the raw data by correctly exposing for the JPEG. If you have a false understanding that the two should be treated the same, you fail to understand the differences in the data, encoding, oh, I'm probably going over your head, sorry. Bottom line is, just as you would never treat a neg 'rated' at ISO 400 the same as a chrome 'rated' at ISO 100 until you tested exposure+development to produce the best results, you don't do the same with digital capture.

Now if you want to point a meter, reflective or incident and treat film, JPEG, raw, any ISO, and the development the same and expect the same results, you clearly don't understand exposure and development, again photography 101.

ETTR should just be called 'proper exposure for raw data' or whatever you want to name it. The only so called conflicting goals is being forced to under expose (and over develop, at the expense of data qualty) because you don't have enough light to fully embrace ETTR, correct exposure. In the film days we DID have to do this and we did see the result; more grain. Stopping the action or producing better data, any photographer will also understand they have to push the process, grain/noise as the net result. Lesser of two evils and again, photography 101.
The tools may be better, but the photographer who doesn't understand what they are being told and follows the advise incorrectly are not aided here. The better metering you speak of isn't going to help the raw shooter if it's based on the JPEG and they do not implement ETTR.

Wow, that's remarkably patronising. Even for you. Though your reply has very little to do with what I was actually writing about. Probably as you have quoted from three different people and attributed them all to me, all whilst being childishly insulting about my abilities to take a photograph. Yet I'm the one who does photography for a living and you do not. Funny that.

[digitaldog]

Probably as you have quoted from three different people and attributed them all to me...

Actually two of three but missing such items is your specialty. Do you feel that all quotes in the body of a forum post must have individual's attributed when the entire post quoted is just below? If so, forgive me, I'll be sure to include this in the future.

...all whilst being childishly insulting about my abilities to take a photograph.

I don't see anything about your abilities to take a picture. I do see some inabilities to understand a simple process that's been done since photogrpahy was invented. Wasn't you who wrote: I think people are over complicating things here.

Yet I'm the one who does photography for a living and you do not. Funny that.

Like attempting to piss on a few images on my web page that isn't at all a photography site, this is then an attempt to dismiss the statements about exposure I wrote? Are you aware of my days of shooting for a living when perhaps you were in grade school? Any posts not made by a photographer working to make a living today is therefore easily dismissed in your mind? Such a state of mind you find funny, that's kind of a shame.

[jjj]

Actually two of three but missing such items is your specialty.

No one was from me, two were from other posters. This is probably why you have such problems with forums, you have no idea who posted what or in what context.

Do you feel that all quotes in the body of a forum post must have individual's attributed when the entire post quoted is just below? If so, forgive me, I'll be sure to include this in the future.

Attributing quotes to the wrong person is rather obviously misleading. You shouldn't have to have that explained.
And the initial quoted post was four posts above, not just below. But hey why bother with pesky facts?

I don't see anything about your abilities to take a picture. I do see some inabilities to understand a simple process that's been done since photogrpahy was invented. Wasn't you who wrote: I think people are over complicating things here.

I did, but that doesn't mean I do not understand ETTR or other aspects of exposure. Being smart is making complex things simple and people are making a meal out of 'correct' exposure. Besides as a colleague sagely rather sagely said - the right exposure is the one the photographer deems correct. On that point I've tried ETTR and the resultant look doesn't suit my style, so do not bother with it.

Like attempting to piss on a few images on my web page that isn't at all a photography site, this is then an attempt to dismiss the statements about exposure I wrote? Are you aware of my days of shooting for a living when perhaps you were in grade school? Any posts not made by a photographer working to make a living today is therefore easily dismissed in your mind? Such a state of mind you find funny, that's kind of a shame.

Actually I've been told that you moved into colour consulting, because you weren't very good at photography or photoshop.
Oh not the ridiculous 'I'm older than you, so must know more' argument. Two problems with that a)you do not even know how old I am and b)it's bollocks anyway. Never went to grade school either.

[digitaldog]

No one was from me, two were from other posters.

That's exactly correct and what I wrote! One was attributed to you. One was attributed to hjulenissen. The third, the last, my bad wasn't. Two of the three had quoted authors, you missed that too?

This is probably why you have such problems with forums, you have no idea who posted what or in what context.

Pot again calling the kettle black.

Attributing quotes to the wrong person is rather obviously misleading.

Ah, so that's how you see it. My question: Do you feel that all quotes in the body of a forum post must have individual's attributed when the entire post quoted is just below?. is at least answered without your usual ambiguity. Again I'll state: If so, forgive me, I'll be sure to include this in the future!
Meanwhile in two different forums, just this week you state you're so good at reading other's posts, an admission that doesn't hold up under scrutiny. The TIFF/PSD post is a prefect example. I wanted to move on, pointing out you missed a key part of my original writings about being done editing an image then flattening. You're still under the impression that Photoshop is a parametric image editor and that John and Julieann say as such when indeed the do not.

But hey why bother with pesky facts?

I haven't seen any facts provided by you yet. Posting quotes that don't backup what you're trying to suggest shows agreement with your opinions isn't going to fly. You are welcome to your opinions but not your own facts.

I did, but that doesn't mean I do not understand ETTR or other aspects of exposure.

This statement would suggest otherwise:

Don't like ETTR myself for example. Yes it may collect more information, but I'm a photographer not a forensic analyst, so I don't get upset by my shadows being black as that's maybe how I want them to look.  Smiley

Care to provide us a DNG with whatever settings you like, produced using correct ETTR that doesn't allow shadows as black as zero? I've never seen such a case.

Being smart is making complex things simple and people are making a meal out of 'correct' exposure. Besides as a colleague sagely rather sagely said - the right exposure is the one the photographer deems correct. On that point I've tried ETTR and the resultant look doesn't suit my style, so do not bother with it.

Look, this is a technical discussion, the subject is exposure. I know you'd rather talk about how much a working photographer you are, how much better your images are than my 'snap shots', how it's all about anything but the technical aspects of the discussion. As such, this set of discussion's isn't for you. There are proper ways to expose film, digital be it JPEG or raw or capture for HDR etc.

Actually I've been told that you moved into colour consulting, because you weren't very good at photography or photoshop.

Right, just believe whatever you want, it's immaterial to this discussion.

Oh not the ridiculous 'I'm older than you, so must know more' argument. Two problems with that a)you do not even know how old I am and b)it's bollocks anyway. Never went to grade school either.

I don't know anything about you, your lack of transparency is glaring. It doesn't matter. That you are a working photographer today and I was when you were in grade school or a nursing home dosn't matter a lick. You brought this up, the facts that you did is so telling. How does this have anything to do with helping the OP with proper exposure? It doesn't. But it gives you more troll speak to inflict on this audience.

[bjanes]

and I using LR Rawdigger can tune real time (pre shot) blinkies (E-M1) or zebra (A7) to show me when clipping starts in raw with less than 1/3 EV precision and understand what is the headroom in raw when I do spot metering.. that is I don't need clueless histogram in LR/ACR at all... that is exposure 101 nowadays.

I agree that the use of Rawdigger (or a similar program that shows the raw histogram and clipping) is the key to obtaining good exposure. It is useful to photograph a Stouffer wedge for testing, since the steps are precisely 0.3 EV apart. I photographed the wedge with the Nikon D800e using the default picture control. A near ideal luminance histogram is shown here. The highlights are just short of clipping.



Rawdigger shows that the highlights are about 2/3 EV below clipping. Increasing the exposure by 0.3 EV causes some clipping in the camera histogram, but the highlights as shown by Rawdigger are still slightly short of clipping.





Increasing exposure by an additional 0.3 EV results in clipping as shown by Rawdigger. The camera histogram shows additional clipping. Using ACR with PV2010 and a linear tone curve and the exposure offset of -0.3 EV (the exposure offset used by ACR and LR) gives a fairly accurate evaluation of the raw file. PV2012 with default settings appears too bright, but step one is not clipped because of automatic highlight recovery.







The camera histogram is highly useful if one knows how to use it. Since most histograms are somewhat conservative, slight clipping can be allowed. However, with current high performance sensors, one can still get excellent quality with highlights 2/3 EV below clipping, at least at base ISO. In difficult situations such as low light or high dynamic range subjects, one might give an additional +2/3 or 1 EV exposure in a second image. HDR bracketing would give additional insurance.

Regards,

Bill

[digitaldog]

Using ACR with PV2010 and a linear tone curve and the exposure offset of -0.3 EV (the exposure offset used by ACR and LR) gives a fairly accurate evaluation of the raw file. PV2012 with default settings appears too bright, but step one is not clipped because of automatic highlight recovery.

So what's the consensus of this automatic highlight recovery? Isn't this ACR taking two good channels and building data in the 3rd? That's been the case for far longer than PV2012 I believe, but prior to PV2012, the recovery was rather ugly.

[bjanes]

So what's the consensus of this automatic highlight recovery? Isn't this ACR taking two good channels and building data in the 3rd? That's been the case for far longer than PV2012 I believe, but prior to PV2012, the recovery was rather ugly.

Yes, that is my understanding. However, the automatic recovery with PV2012 is done without the users intervention and knowledge. I agree with Bart: it is important to know what is really going on with one's images.

Bill

[digitaldog]

Yes, that is my understanding. However, the automatic recovery with PV2012 is done without the users intervention and knowledge.

So again, in this case, the development plays a critical role in the total picture.
So if you get a data point about clipping with Raw Digger but you want to use PV2012 (which has lots of improvements over 2003), what do you now do? Adjust exposure or leave it based on your development preferences?