Expose to the right (ETTR) or underexpose?

[Jaybob]

Advertisement (gone after registration)

...exactly what ETTR has always meant to the discerning, namely "give as much exposure as you can without blowing the brightest highlights that matter".

 

My understanding exactly.

[pgk]

In fact, ETTR makes sense only for high-contrast scenes when a regular exposure, i. e. a centered histogram, means blowing some highlights and blocking some shadows. Since blocked shadows usually are less objectionable to the eye than blown highlights, it makes sense to reduce exposure in order to save the highlights from blowing, at the expense of more blocked shadows. Usually. Exceptions to the rule may happen; it depends on the subject and the photographer's intentions.

Thanks 01af. You have described my empirical findings quite succinctly. As with many aspects of photography, 'rules' are only valid if used appropriately and should be broken when it is effective to do so.....

[Advertisement]

The Leica Q3 is available for pre-order at B&H Photo Video and Adorama

[tobey bilek]

Put the histo in the center. If blacks are clipped and there is space on the right, give more exposure. I see no point in pushing a too bright image and then pulling it back.

 

If you are already to white clip point, then you need to boost blacks in computer or do HDR.

[mjh]

I see no point in pushing a too bright image and then pulling it back.

The point is less noise and more finely resolved tonal values. Assuming the M9 has a dynamic range of 11 EV at its base sensitivity, a scene contrast of 9 EV is resolved into 8176 distinct values if placed in the middle of the histogram, but 16,352 distinct values if shifted by 1 EV to the right. (It’s 4064 vs. 16,256 values for a scene contrast of 7 EV and so on.)

[sandymc]

The point is less noise and more finely resolved tonal values. Assuming the M9 has a dynamic range of 11 EV at its base sensitivity, a scene contrast of 9 EV is resolved into 8176 distinct values if placed in the middle of the histogram, but 16,352 distinct values if shifted by 1 EV to the right. (It’s 4064 vs. 16,256 values for a scene contrast of 7 EV and so on.)

 

And when shooting compressed, squashes all that into a total of 256 distinct values in raw - so all those extra values from ETTR will really help.

 

Sorry, I should stay out of this thread, but sometimes I just can't resist......

 

Sandy

[mjh]

And when shooting compressed, squashes all that into a total of 256 distinct values in raw - so all those extra values from ETTR will really help.

They do. There are still more values per f-stop on the right than there are to the left, even when the difference isn’t quite as pronounced in the compressed DNG format. For example, in the 7 EV case it is 116 vs. 233 values (of 256).

Edited October 18, 2012 by mjh

[sandymc]

Advertisement (gone after registration)

They do. There are still more values per f-stop on the right than there are to the left, even when the difference isn’t quite as pronounced in the compressed DNG format. For example, in the 7 EV case it is 116 vs. 233 values (of 256).

 

Actually, no, the top stop goes from code 181 to 255; a 1 EV delta gets you 74 extra values max. But the point is, Leica's compression deliberately throws away thousands and thousands of values, and you don't see many complaints about image quality. Which tells you how much all the additional ETTR values gain for you in terms of practical image quality.

 

Sandy

[pgk]

in terms of practical image quality.

 

Sandy

And surely, in debating a subject where the final image quality is defined not in numerical terms, but in its actual acceptability within a final viewed image, this is the point. There is little point chasing a theoretical concept if its not tested out to see whether it actually has a practical bearing in reality. And in all honesty I'd rather not spend yet more time post processing theoretically better raw files if I don't actually have to.

[k-hawinkler]

Is my understanding correct that in the new Leica M a lossless compression scheme is implemented?

That would have some bearing on the discussion here, will it not?

[mjh]

Actually, no, the top stop goes from code 181 to 255; a 1 EV delta gets you 74 extra values max.

The difference quoted is for 7 EV placed in the middle of a 11 EV dynamic range (which leaves the rightmost 2 EV unused) vs placed flush with the right edge. +2 EV gains you 128 additional values at the right hand side whereas you lose just 11 values on the left, resulting in a net gain of 117 values, i.e. 233 rather than just 116 values in total.

 

But the point is, Leica's compression deliberately throws away thousands and thousands of values, and you don't see many complaints about image quality. Which tells you how much all the additional ETTR values gain for you in terms of practical image quality.

When you have just 256 values, max, you don’t want to throw away any more of those if you don’t have to.

[douglasf13]

This statement, from beginning to end, is inaccurate, except for "don't blow highlights".

 

Regardless of camera model, sensor type, film stock, OR ISO, "proper" exposure will always yield a better technical photograph. The BEST way to accomplish proper exposure in digital photography is to utilize ETTR. The only real "correctable" mistake you can make shooting RAW is in white balance, exposure errors will almost always compromise the image quality, unless of course, that's what you're looking for.

 

Jay

 

It seems like you misunderstand how digital photography fundamentally works. ISO is not really part of exposure in digital photography. Shutter and aperture speed control the amount of light that hits the sensor, and the ISO just boosts the amplification after the fact. Whether one should boost that signal with the camera's internal analog or digital gain boost (ISO,) or use the raw converter to boost the gain after the fact, depends on the camera and raw converter.

 

For example, if you shoot Canon, it is generally better to boost ISO in-camera up to higher ISOs. With several sensors, like in the M9, you can boost ISO in the raw converter, rather than raising in-camera ISO, and potentially even improve your DR by doing it this way. Of course, results are highly dependent on the raw converter being used.

 

So, I agree that exposure is important, but, in digital, exposure really only refers to your shutter speed and aperture. ISO is just camera gain after the fact, which can be done in different ways. Several cameras are essentially ISO-less, meaning that you can focus on the amount of light hitting the sensor, and boost the gain later in the raw converter.

 

p.s. because of the above, ETTR really only applies to shooting at base ISO, btw. ETTR can be good from a SNR standpoint, although, as others have mentioned, it can be a problem in regards to color. It can also be a problem if one underexposes the subject too much in an attempt to save highlights that may not be as important to the shot.

Edited October 18, 2012 by douglasf13

[AbbeyFoto]

LuLa has a pretty good explanation of the idea behind ETTR at Optimizing Exposure.

 

Thanks for the link. I found the article very illuminating but left me needing a fuller and clearer exposition. I read "Film has a characteristic tone curve. The human eye sees logarithmically. But digital sensors are inherently linear." And yet "Most current DSLRs and digital backs have a dynamic range somewhere between 8 and 12 stops." Furthermore, "it [ETTR] definately doesn't mean blowing the highlights." Does this mean digital sensors has a highly nonlinear response? Outside the DR a digital sensor "clips"? Rather than a smooth curve we have cutoff corners? And with colour filters in front of a sensor the clipping is dependent on the colour of the light hitting the sensor? Hence the "clipping" depends on the combination of colour filtration used in almost all digital cameras and the spectral nature of the light hitting the sensor. White and black cats does not help here. The article hints at this issue but is not as clear as I would wish.

[AbbeyFoto]

.... ISO is not really part of exposure in digital photography. Shutter and aperture speed control the amount of light that hits the sensor, and the ISO just boosts the amplification after the fact. Whether one should boost that signal with the camera's internal analog or digital gain boost (ISO,) or use the raw converter to boost the gain after the fact, depends on the camera and raw converter.

 

For example, if you shoot Canon, it is generally better to boost ISO in-camera up to higher ISOs. With several sensors, like in the M9, you can boost ISO in the raw converter, rather than raising in-camera ISO, and potentially even improve your DR by doing it this way. Of course, results are highly dependent on the raw converter being used.

 

So, I agree that exposure is important, but, in digital, exposure really only refers to your shutter speed and aperture. ISO is just camera gain after the fact, which can be done in different ways. Several cameras are essentially ISO-less, meaning that you can focus on the amount of light hitting the sensor, and boost the gain later in the raw converter.

 

p.s. because of the above, ETTR really only applies to shooting at base ISO, btw. ETTR can be good from a SNR standpoint, although, as others have mentioned, it can be a problem in regards to color. It can also be a problem if one underexposes the subject too much in an attempt to save highlights that may not be as important to the shot.

 

I quote this message as it is symptomatic of my need for a clear and systematic exposition of the issues raised in this thread. I think I understand much of what has been said but I am left with a feeling I really do not really understand the digital process. I would ask for guidance on a careful, accurate, detailed and yet approachable discussion of these issues.

[FlashGordonPhotography]

While I understand the theory and on some of my cameras I see the difference, I don't do it on my M9. The M9's sensor acts like no other I've ever owned. The amount of detail rendered in the shadows is amazing and the highlight shoulder is very steep. I did my own personal test with some exposure bracketing using the cameras histogram as a reference. In all cases , *for me*, I preferred the colours and tonal range when I pushed the exposure in post compared to pulling. Using ETTR seemed to leave me with a harsh and nasty looking fall off in the highlights.

 

So for me, (and you're free to disagree) my M9 is not an ETTR camera.

 

Gordon

[Jaybob]

It seems like you misunderstand how digital photography fundamentally works. ISO is not really part of exposure in digital photography. Shutter and aperture speed control the amount of light that hits the sensor, and the ISO just boosts the amplification after the fact. Whether one should boost that signal with the camera's internal analog or digital gain boost (ISO,) or use the raw converter to boost the gain after the fact, depends on the camera and raw converter.

 

For example, if you shoot Canon, it is generally better to boost ISO in-camera up to higher ISOs. With several sensors, like in the M9, you can boost ISO in the raw converter, rather than raising in-camera ISO, and potentially even improve your DR by doing it this way. Of course, results are highly dependent on the raw converter being used.

 

So, I agree that exposure is important, but, in digital, exposure really only refers to your shutter speed and aperture. ISO is just camera gain after the fact, which can be done in different ways. Several cameras are essentially ISO-less, meaning that you can focus on the amount of light hitting the sensor, and boost the gain later in the raw converter.

 

p.s. because of the above, ETTR really only applies to shooting at base ISO, btw. ETTR can be good from a SNR standpoint, although, as others have mentioned, it can be a problem in regards to color. It can also be a problem if one underexposes the subject too much in an attempt to save highlights that may not be as important to the shot.

 

Yes, I have no idea how digital photography works. Thank you for pointing that out to me as the crowning moment to your SIX posts here. I'm not sure how I managed to make it this far in life.

 

If you dont have an ISO to work with be it 160, 500, 1000, or whatever, you dont have an exposure to figure out. Shooting all your exposures at 160 and ramping it up in your raw converter doesn't make you much of a photographer, in fact it may be the most fundamentally unsound approach to the art and science of the craft of photography I have ever encountered. Boosting the ISO later? Why? Where's the SKILL in that? Nailing it, both moment AND exposure, is more important to me, Nikon, Sony, or meterless Leica. You may as well shoot with your phone.

 

Show me your M9 histograms, files, and your final examples of this impossible to follow technique. It seems as though the whole concept of it is flawed and following it would yield nothing but unusable exposures. What do I know? I don't understand digital photography.

Edited October 18, 2012 by Jaybob

[douglasf13]

I quote this message as it is symptomatic of my need for a clear and systematic exposition of the issues raised in this thread. I think I understand much of what has been said but I am left with a feeling I really do not really understand the digital process. I would ask for guidance on a careful, accurate, detailed and yet approachable discussion of these issues.

 

I think the confusing thing about digital is that people think of ISO as if it is having a film camera with infinite film speeds built in, but that is not really the case. It is more like pushing film after development. Our digital cameras have a single, base ISO, and raising the ISO requires either pushing the exposure in a raw converter, or pushing the exposure in the camera with the camera's analog or digital gain. Some cameras do better with the former, others with the latter. It entirely depends on the sensor design.

 

Granted, if you don't push the ISO in camera, your preview images will be dark, so the trade off may still be worth it.

 

p.s. this is all assuming that you shoot raw. if you shoot jpeg, using in-camera ISO is more mandatory.

[douglasf13]

Yes, I have no idea how digital photography works. Thank you for pointing that out to me as the crowning moment to your SIX posts here. I'm not sure how I managed to make it this far in life.

 

If you dont have an ISO to work with be it 160, 500, 1000, or whatever, you dont have an exposure to figure out. Shooting all your exposures at 160 and ramping it up in your raw converter doesn't make you much of a photographer, in fact it may be the most fundamentally unsound approach to the art and science of the craft of photography I have ever encountered. Boosting the ISO later? Why? Where's the SKILL in that? Nailing it, both moment AND exposure, is more important to me, Nikon, Sony, or meterless Leica. You may as well shoot with your phone.

 

Show me your M9 histograms, files, and your final examples of this impossible to follow technique. It seems as though the whole concept of it is flawed and following it would yield nothing but unusable exposures. What do I know? I don't understand digital photography.

 

You managed this far because camera makers decided to call camera gain "ISO," in order for film shooters making the transition to digital easily understandable, especially since our cameras are still geared towards jpeg output. If you're shooting jpeg, it certainly does matter, and I should have made the distinction that we're talking about raw. Either way, you know the base ISO of the M9, 160, and that is all the guide that you really need. Exposure is simply the amount of light hitting the sensor, which is not dictated by ISO in a digital camera.

 

Quite frankly, what I'm talking about is nothing new, and is a common topic in other forums. Feel free to criticize my 6 posts, but I've only recently rejoined this forum, and, although I'm embarrassed to admit it, I have something like 15,000 posts in other enthusiast forums, so please don't go there.

 

I'm not going to take the time to do tests which you can do on your own, but here's one of several links illustrating what I'm talking about from raw expert Guillermo Luijk. The thread is in regard to the Pentax K-5, another essentially ISO-less camera. Noise and DR comparision: Canon 5D vs 5D2 vs 7D vs Pentax K5

 

p.s. I do still use ISO 800 or 1250 on the M9, because I like to have a usable jpeg review on the back of the camera. Also, with my NEX cameras, since the LCD and EVF rely on boosting the ISO in order to see the frame properly, I boost ISO in those cameras, too. So, even though I may be loosing a little DR by boosting ISO, it is often still worth it to me.

Edited October 19, 2012 by douglasf13

[Jaybob]

If I walked into a dark room and shot everthing at 160, on RAW, (of course), and tried to boost it later, you know what I'd have? People who understand enough about photography (my clients who PAY me) mad enough to murder me. It doesn't take more than that to realize that that underexposing and boosting the blacks, shadows, exposure, highlights, whites, and contrast or the "ISO" later, IS the worst photographic technique ever devised. There NO benefit to doing it that way, ZERO. I dont care if you're shooting M9, D4 or whatever it is that Canon is selling to those dopes. PENTAX?? really?.

 

I understand ISO and gain. I also understand that the lower I keep my ISO the better my images look. I never turn my D3 above 2500, because the images start to look suspect above that point and I haven't shot a JPG since 2003. I prefer to get the exposure right so I'm not dicking around in LR for days at a time fixing things. That's how digital photography works in the real world.

 

SO....Show me examples of how it works. Histograms, files exposed using this method, and final images...

Edited October 19, 2012 by Jaybob

[douglasf13]

If I walked into a dark room and shot everthing at 160, on RAW, (of course), and tried to boost it later, you know what I'd have? People who understand enough about photography (my clients who PAY me) mad enough to murder me. It doesn't take more than that to realize that that underexposing and boosting the ISO later IS the worst photographic technique ever devised, M9, D4 or whatever it is that Canon is selling to those dopes.

 

That is fine by me. No one says you have to use these techniques. This thread is about the nuance of exposure, and as I mentioned above, it may not always be practical in use. If your clients handle your raws, I can understand their confusion. If the photographer doesn't understand the concept, I doubt a client would.

 

You reference the D4, the D3 and Canon cameras, all of which this method doesn't apply to. You should, try it with your NEX-5, though (although maybe one stop above base ISO.) Try using the same shutter speed and aperture at both ISO 200 and ISO 1600, and boost the ISO 200 image in your raw converter to match the ISO 1600 image. If you're using LR, the original ISO 1600 image will likely be slightly better. If you're using RPP, the opposite should be true.

 

p.s. the link I posted above has plenty of discussion and examples of the ISO-less Pentax K-5.

Edited October 19, 2012 by douglasf13

[Advertisement]

Hallo,

Look here: Leica M

[Jaybob]

That is fine by me. No one says you have to use these techniques. This thread is about the nuance of exposure, and as I mentioned above, it may not always be practical in use. If your clients handle your raws, I can understand their confusion. If the photographer doesn't understand the concept, I doubt a client would.

 

You reference the D4, the D3 and Canon cameras, all of which this method doesn't apply to. You should, try it with your NEX-5, though (although maybe one stop above base ISO.) Try using the same shutter speed and aperture at both ISO 200 and ISO 1600, and boost the ISO 200 image in your raw converter to match the ISO 1600 image. If you're using LR, the original ISO 1600 image will likely be slightly better. If you're using RPP, the opposite should be true.

 

p.s. the link I posted above has plenty of discussion and examples of the ISO-less Pentax K-5.

 

Exposing Toward the Right, without blowing highlights while using a reasonable ISO has been working just fine for me since I first picked up a D1 at the end of 1999, thanks anyway. Shooting everything at 160? no thanks. I did that with the Digilux 2.

 

If your argument is hinging on me following that link, I would never employ those techniques..."photon noise?" are you serious? Why complicate something so simple to get the same, and inherently more controllable, result? I'd sooner go back to film.

Edited October 19, 2012 by Jaybob