Ad
Ad
Ad
Pages: [1] 2 3 4 »   Bottom of Page
Print
Author Topic: Is expose to the right ever wrong?  (Read 12342 times)
hdomke
Full Member
***
Offline Offline

Posts: 149


WWW
« on: December 06, 2011, 09:51:41 AM »
ReplyReply

Luminous Landscape has taught me to err on the side of over-exposure and not under-exposure.
Provided that one does not clip any highlight information that is desired, is it ever wrong to expose to the right?
Are there circumstances where one might get better results not doing ETTR (expose to the right)?
Logged

Henry

Henry Domke Fine Art
www.henrydomke.com
Mark D Segal
Contributor
Sr. Member
*
Offline Offline

Posts: 6894


WWW
« Reply #1 on: December 06, 2011, 09:55:03 AM »
ReplyReply

Luminous Landscape has taught me to err on the side of over-exposure and not under-exposure.
Provided that one does not clip any highlight information that is desired, is it ever wrong to expose to the right?
Are there circumstances where one might get better results not doing ETTR (expose to the right)?

No - you are thinking analog, not digital. LULA would not teach you to "over-expose". In digital, over-expose means you clip highlights having detail you want to preserve. As long as you don't do that, you are not over-exposing. The image may LOOK brighter than you want it to be in the final analysis, but that's just a matter of adjusting it back down in post-capture processing; at least you will be starting with the maximum luminance information obtainable. That's what ETTR does for you.
Logged

Mark D Segal (formerly MarkDS)
Author: "Scanning Workflows with SilverFast 8....." http://www.luminous-landscape.com/reviews/film/scanning_workflows_with_silverfast_8.shtml
Guillermo Luijk
Sr. Member
****
Offline Offline

Posts: 1283



WWW
« Reply #2 on: December 06, 2011, 10:52:05 AM »
ReplyReply

Are there circumstances where one might get better results not doing ETTR (expose to the right)?

There are several cases where ETTR is not the best option, for example:
- If your subject of interest is in the deep shadows, and the scene has a high dynamic range. For example you take a picture of your cat at home in a dark room with a window facing a sunny day outside. ETTR would mean to expose in such a way that the window is preserved, but your cat would be captured so dark that will display a lot of noise when lifting the shadows in post processing.
- If to achieve ETTR you are forced to use a shutter speed not quick enough to avoid motion blurring, or you are forced to use such an aperture that your DOF is insufficient. In these cases, some extra noise because of not-ETTR is desirable vs the lack of sharpness
- If the scene is so dynamic that you cannot do the calculations or trial&error to achieve proper ETTR. In that situation it is better not to ETTR and underexpose slightly in order to prevent highlight clipping.
 
Logged

jonathanlung
Jr. Member
**
Offline Offline

Posts: 70


« Reply #3 on: December 06, 2011, 11:03:55 AM »
ReplyReply

To add to Mark's comments that you should, if anything, err on the side of underexposing with digital since one can recover more from the shadows than with film. One has about two stops latitude on overexposure on negative film (and one under), depending on the scene.

Of course, ETTR's purpose (roughly speaking) is to maximize capture quality (w.r.t. noise and number of levels of luminosity distinguished) after applying exposure compensation in post while staying within the scene's dynamic range. But if what you want/need is not provided by ETTR, then it can be "wrong", just like trying to use any tool for something it wasn't intended for.

In addition to the examples from Guillermo, here are some more:

For instance, if you need to use unadjusted output from the camera (for example, to hand an out-of-camera JPEG to a client on the spot), then you don't have the option of adjusting exposure compensation in post; ETTR is not suitable here. Or, if you know you can afford to blow highlights, then you're not staying within the scene's dynamic range. For example, maybe you're going to crop your image to eliminate the part with blown out highlights. Here, you can increase exposure even more to maximize capture quality. But you really do need to make sure you can afford/want to have clipped highlights and/or will be removing those sections of the image later.

If you're shooting RAW and can adjust later, the exposure you would get from using ETTR should be the minimum and only to be exceeded if you know what you're doing!
« Last Edit: December 06, 2011, 11:06:08 AM by jonathanlung » Logged
Mark D Segal
Contributor
Sr. Member
*
Offline Offline

Posts: 6894


WWW
« Reply #4 on: December 06, 2011, 12:33:09 PM »
ReplyReply

There are several cases where ETTR is not the best option, for example:
- If your subject of interest is in the deep shadows, and the scene has a high dynamic range. For example you take a picture of your cat at home in a dark room with a window facing a sunny day outside. ETTR would mean to expose in such a way that the window is preserved, but your cat would be captured so dark that will display a lot of noise when lifting the shadows in post processing.
- If to achieve ETTR you are forced to use a shutter speed not quick enough to avoid motion blurring, or you are forced to use such an aperture that your DOF is insufficient. In these cases, some extra noise because of not-ETTR is desirable vs the lack of sharpness
- If the scene is so dynamic that you cannot do the calculations or trial&error to achieve proper ETTR. In that situation it is better not to ETTR and underexpose slightly in order to prevent highlight clipping.
 

Guillermo, all of which adds-up to saying that we have capture situations where compromises are necessary, which of course is true; but one should perhaps emphasize this doesn't impair the general principle that ETTR make the most sense for those who want to retain the maximum possible amount of usable, high quality information in a raw capture.

On your first point, if the scene DR substantially exceeds that of the sensor and you want it all, it may be necessary to make several exposures and blend them - if the situation allows. However, if you intend to ignore or crop-out the areas where the extreme DR manifests itself worst, then of course concentrate the exposure on the material one will keep - and make it an ETTR capture. On your third point, I agree with the idea of being extra-cautious not to clip highlights in conditions of real uncertainty.
Logged

Mark D Segal (formerly MarkDS)
Author: "Scanning Workflows with SilverFast 8....." http://www.luminous-landscape.com/reviews/film/scanning_workflows_with_silverfast_8.shtml
deanwork
Sr. Member
****
Offline Offline

Posts: 702


« Reply #5 on: December 06, 2011, 08:10:26 PM »
ReplyReply

In two situations: when shooting very high speed iso, and when shooting jpegs fomat that can't handle the data build up in the  highlights.

Most of the pixels are in the high-end and in shooting in moderate to flat lighting it is a good method IF you are processing parametrically . Shooting in bright sunlight where the high values are extreme you have to be a little more careful, like in Tucson in the summer....


Logged
mouse
Full Member
***
Offline Offline

Posts: 165


« Reply #6 on: December 07, 2011, 11:25:32 PM »
ReplyReply

Being a frequent visitor to this forum, I have gained great respect for Guillermo's knowledge and expertise in this (and other) areas; thus I should know better than to quibble with his response.  However on this occasion I venture to do so because, even in this august assembly,  I believe there are readers not sufficiently able to read between the lines and may misinterpret the answers provided here.

There are several cases where ETTR is not the best option, for example:
- If your subject of interest is in the deep shadows, and the scene has a high dynamic range. For example you take a picture of your cat at home in a dark room with a window facing a sunny day outside. ETTR would mean to expose in such a way that the window is preserved, but your cat would be captured so dark that will display a lot of noise when lifting the shadows in post processing.

I thought that we had reached a consensus on the meaning of ETTR.  Expose in such a way that the highlights are preserved, only if those highlights contain detail which is important to your image.  In this particular example I think the correct advise would be to forget the window and then expose as far to the right as the cat and the dark room will allow.  In other words, ETTR and then ETTR some more.


Quote
- If to achieve ETTR you are forced to use a shutter speed not quick enough to avoid motion blurring, or you are forced to use such an aperture that your DOF is insufficient. In these cases, some extra noise because of not-ETTR is desirable vs the lack of sharpness


Here again the less knowledgable reader needs to know that in these circumstances (fixed shutter speed and/or aperture), employing ETTR through increasing the ISO (with many cameras) will produce a cleaner image, without sacrificing sharpness or desired DOF.

With much respect.
m
Logged
Tim Gray
Sr. Member
****
Offline Offline

Posts: 2002



WWW
« Reply #7 on: December 08, 2011, 09:44:57 AM »
ReplyReply

There is a post by Ctein on TOP http://theonlinephotographer.typepad.com/the_online_photographer/2011/10/expose-to-the-right-is-a-bunch-of-bull.html which is typical of the misunderstanding of what ETTR is all about.  The problem is that the ETTR is the third of three factors which must be considered:

1.  Don't blow highlights that have detail you want to preserve.
2.  Use the appropriate shutter speed, f stop and ISO
3.  If, subject to 1 and 2 there is latitude to increase the exposure, do so.

The goal of ETTR is to provide as much flexibility in post as is practical.  If that's not important to you (and for some it really isn't) then don't worry about ETTR.
Logged
Guillermo Luijk
Sr. Member
****
Offline Offline

Posts: 1283



WWW
« Reply #8 on: December 08, 2011, 09:49:07 AM »
ReplyReply

The goal of ETTR is to provide as much flexibility in post as is practical.  If that's not important to you (and for some it really isn't) then don't worry about ETTR.

I would be more precise than talking about 'flexibility'. I would say:

If noise is not a problem in your application, you don't need to worry about ETTR.
Logged

Mark D Segal
Contributor
Sr. Member
*
Offline Offline

Posts: 6894


WWW
« Reply #9 on: December 08, 2011, 10:02:56 AM »
ReplyReply

There is a post by Ctein on TOP http://theonlinephotographer.typepad.com/the_online_photographer/2011/10/expose-to-the-right-is-a-bunch-of-bull.html which is typical of the misunderstanding of what ETTR is all about.  The problem is that the ETTR is the third of three factors which must be considered:

1.  Don't blow highlights that have detail you want to preserve.
2.  Use the appropriate shutter speed, f stop and ISO
3.  If, subject to 1 and 2 there is latitude to increase the exposure, do so.

The goal of ETTR is to provide as much flexibility in post as is practical.  If that's not important to you (and for some it really isn't) then don't worry about ETTR.

Hi Tim,

I think this is an excellent summary of good advice. The problem with Ctein's argument is that he focuses mainly on noise, dismisses it too easily in my opinion - regardless of the technical progress since 2003, and doesn't take into account that ETTR allows for capturing more information, resulting in more assurance of smooth tonal gradation, given the greater amount of tonal information the sensor captures further up the tone scale, (ref. Jeff Schewe in his ACR books). Some people think this is a theoretical consideration not seen in practice, but others have seen it in practice. Ctein makes a separate point that in-camera histograms are not necessarily a reliable indicator of clipping. I think this can be true and also well-known, but all it means is to be careful - understand what your camera is showing and perhaps allow a slight margin if in doubt. It doesn't vitiate the general principle.
Logged

Mark D Segal (formerly MarkDS)
Author: "Scanning Workflows with SilverFast 8....." http://www.luminous-landscape.com/reviews/film/scanning_workflows_with_silverfast_8.shtml
Guillermo Luijk
Sr. Member
****
Offline Offline

Posts: 1283



WWW
« Reply #10 on: December 08, 2011, 10:40:59 AM »
ReplyReply

Some people think this is a theoretical consideration not seen in practice, but others have seen it in practice.

Unfortunately for us all, none of them wanted to show others what they saw.
Logged

Mark D Segal
Contributor
Sr. Member
*
Offline Offline

Posts: 6894


WWW
« Reply #11 on: December 08, 2011, 11:09:06 AM »
ReplyReply

I should have said others claim to have seen it in practice. The smoothness of tonal gradations can be an elusive quality to observe unambiguously. Nonetheless it makes sense to me in principle that if you can safely expose in a manner that preserves more information rather than less, it makes sense to do so. What's the downside, noting that I used the word "safely" ?
Logged

Mark D Segal (formerly MarkDS)
Author: "Scanning Workflows with SilverFast 8....." http://www.luminous-landscape.com/reviews/film/scanning_workflows_with_silverfast_8.shtml
Guillermo Luijk
Sr. Member
****
Offline Offline

Posts: 1283



WWW
« Reply #12 on: December 08, 2011, 11:19:20 AM »
ReplyReply

Nonetheless it makes sense to me in principle that if you can safely expose in a manner that preserves more information rather than less, it makes sense to do so. What's the downside, noting that I used the word "safely" ?

I don't think anyone says there is any downside in ETTR, I simply claim there is no advantage in it beyond SNR improvement. If that is true, talking about ETTR is talking about noise, being anything else a fallacy.
Logged

Mark D Segal
Contributor
Sr. Member
*
Offline Offline

Posts: 6894


WWW
« Reply #13 on: December 08, 2011, 12:01:43 PM »
ReplyReply

Guillermo, The SNR improvement alone is a very good reason to retain this practice regardless of how good the sensors and noise reduction programs have become. I have noticed this even using a Canon 1Ds3 at 400ISO. Images with histograms further to the right display better noise properties than those with histograms further to left, all else roughly equal, and noise mitigation programs always involve at least some sacrifice of IQ, even if a little. So why not avoid this if possible? As for the benefit of conserving more luminance information, you may be "exposing yourself to the right a bit" :-)  saying this a fallacy, unless you can demonstrate that there are no imaging conditions in which it may be useful.
Logged

Mark D Segal (formerly MarkDS)
Author: "Scanning Workflows with SilverFast 8....." http://www.luminous-landscape.com/reviews/film/scanning_workflows_with_silverfast_8.shtml
hjulenissen
Sr. Member
****
Offline Offline

Posts: 1677


« Reply #14 on: December 08, 2011, 12:24:58 PM »
ReplyReply

saying this a fallacy, unless you can demonstrate that there are no imaging conditions in which it may be useful.
It is impossible to prove a negative. There have been attempts at showing improvements in bright tones due to ETTR, but in my view they did not show benefit. I know of none that shows benefit.

That is no proof, but to me it is a strong indication.

-h
Logged
Mark D Segal
Contributor
Sr. Member
*
Offline Offline

Posts: 6894


WWW
« Reply #15 on: December 08, 2011, 12:27:45 PM »
ReplyReply

I would expect most of the improvement to show in better gradation of dark tones, because that is where the purported information paucity would be greatest, absent deliberate exposure technique to optimize it.

Yes, indeed, hard to prove a negative and that's why I like the expression "never say never".
Logged

Mark D Segal (formerly MarkDS)
Author: "Scanning Workflows with SilverFast 8....." http://www.luminous-landscape.com/reviews/film/scanning_workflows_with_silverfast_8.shtml
Guillermo Luijk
Sr. Member
****
Offline Offline

Posts: 1283



WWW
« Reply #16 on: December 08, 2011, 12:33:50 PM »
ReplyReply

I would expect most of the improvement to show in better gradation of dark tones

But dark tones are dithered by noise, so extra levels mean no practical advantage. Camera makers choose the number of bits of their ADC not to be the limiting factor for any posterization issues even in the dark shadows (where less differentiated levels are encoded per stop) at base ISO (where less dithering noise is present).

I am a fan of ETTR and I use it whenever possible, but I think the only advantage it provides is cleaner images thanks to SNR improvement. The fallacy begins with the having more tonal levels/more information bla bla bla claim, which has zero effect in practice, and nobody has shown any evidence of the opposite so far. The only thing we have are articles (like Michael's) claiming that ETTR is good because it adds extra tonal levels, and people repeating what they read in those articles.

Regards
Logged

Mark D Segal
Contributor
Sr. Member
*
Offline Offline

Posts: 6894


WWW
« Reply #17 on: December 08, 2011, 12:56:03 PM »
ReplyReply

But dark tones are dithered by noise, so extra levels mean no practical advantage. Camera makers choose the number of bits of their ADC not to be the limiting factor for any posterization issues even in the dark shadows (where less differentiated levels are encoded per stop) at base ISO (where less dithering noise is present).


Unless I'm missing something, which is entirely possible, this seems to set aside the notion that through increased exposure of the darker scene tones, they become lighter captured tones as one is moving them rightward up the scale where there is a higher number of encoded levels, and the S/N is greater.
Logged

Mark D Segal (formerly MarkDS)
Author: "Scanning Workflows with SilverFast 8....." http://www.luminous-landscape.com/reviews/film/scanning_workflows_with_silverfast_8.shtml
hjulenissen
Sr. Member
****
Offline Offline

Posts: 1677


« Reply #18 on: December 08, 2011, 01:03:09 PM »
ReplyReply

Unless I'm missing something, which is entirely possible, this seems to set aside the notion that through increased exposure of the darker scene tones, they become lighter captured tones as one is moving them rightward up the scale where there is a higher number of encoded levels, and the S/N is greater.
The point of Guillermo is (I believe) that there is sufficient noise prior to the ADC that it is this noise that limits the performance, and not the ADC levels. Adding noise to a signal tends to "dither" it, meaning that the quantization noise is effectively decorrelated with the signal, and can be considered a noise source in itself. Going to 16 bits or 24 bits of true ADC resolution might only measure the sensor/electronics noise with higher precision, instead of having any more precise data about the desired (noiseless) "signal". Using a 24-bit audio ADC to digitize old worn-out tapes might not give any better audio compared to a 16-bit one, as the desired signal is mainly degraded by the noise and flaws of the tape.

If there are two noise sources, and one is a lot bigger than the other, then you often might as well forget about the smaller one (spectral characteristics might also play a part).

http://en.wikipedia.org/wiki/Dithering

http://www.sengpielaudio.com/calculator-spl.htm

-h
« Last Edit: December 08, 2011, 01:07:52 PM by hjulenissen » Logged
Guillermo Luijk
Sr. Member
****
Offline Offline

Posts: 1283



WWW
« Reply #19 on: December 08, 2011, 01:12:48 PM »
ReplyReply

this seems to set aside the notion that through increased exposure of the darker scene tones, they become lighter captured tones as one is moving them rightward up the scale where there is a higher number of encoded levels, and the S/N is greater.

There are extra levels thanks to ETTR, but as I said in the previous post: "extra levels mean no practical advantage" in this case, because noise dithers any possible posterization.

Left image is 8-bit, right is just 5-bit (8 times less encoded levels):



Both look similar and are equally robust thanks to noise. The same thing happens in the lower stops of a RAW file.
« Last Edit: December 08, 2011, 01:15:32 PM by Guillermo Luijk » Logged

Pages: [1] 2 3 4 »   Top of Page
Print
Jump to:  

Ad
Ad
Ad