Expose to the right (ETTR) or underexpose?

[douglasf13]

Advertisement (gone after registration)

The first is correct. The M9 applies (some) noise reduction at higher ISO settings prior to writng the raw file specifically to correct noise that cannot be addressed in further postprocessing (Stefan Daniel at the introduction of the M9), so shooting at base ISO and pushing in postprocessing will render inferior results.

The second is not the case. The M9 sensor does not support pixel binning.

 

My take - ETTR is useful to sacrifice unimportant highlights in a controlled manner in order the preserve shadow detail and reduce shadow noise.

 

I am film-conditioned over the years to regard " get it right in the camera" as a basic mantra.

 

There have been several examples in this thread alone showing the M9 results pushed in the LR4 converter being superior. It could simply be that Adobe's noise reduction is currently better than Leica's in-camera attempt.

 

If you're a "get it right in the camera" shooter, then ETTR doesn't make sense, either, because you likely have to pull exposure back in the converter.

Edited October 26, 2012 by douglasf13

[Advertisement]

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

[k-hawinkler]

There is a difference in method between pushing in-camera (ISO) and in the raw converter, but both are done after the exposure. Which method provides superior results depends on the camera and raw converter being used. With the M9 and LR4, in particular, it seems that pushing in the converter is at least as good, and possibly a little better.

 

 

Could that be explained by the fact that LR4 is 3 years newer (more development) than the guts of the M9?

Did anyone compare with older versions of LR?

Edited October 26, 2012 by k-hawinkler

[k-hawinkler]

A logical conclusion from the statement referred to.

 

 

That answer reminds me of Aristotle and horse's teeth:

 

Google Answers: Searching for source of "horse's teeth" parable

 

http://www.flowofhistory.com/units/birth/3/FC22

Edited October 26, 2012 by k-hawinkler

[douglasf13]

Could that be explained by the fact that LR4 is 3 years newer (more development) than the guts of the M9?

Did anyone compare with older versions of LR?

 

I'd imagine that has at least something to do with it, as LR4 NR has gotten much better over the years, but I haven't tested it thoroughly, so I'm not sure.

[k-hawinkler]

I'd imagine that has at least something to do with it, as LR4 NR has gotten much better over the years, but I haven't tested it thoroughly, so I'm not sure.

 

 

If true - and LR NR has definitely improved - that seems to be another argument to rely more on post-processing rather than in-camera processing as the latter gets finally locked in.

[algrove]

So I guess you guys are talking about Jpegs?

[k-hawinkler]

Advertisement (gone after registration)

So I guess you guys are talking about Jpegs?

 

No, I am not.

[Lindolfi]

With the need of more experimental evidence in this thread, I have made another set of exposures of a MacBeth color card with the settings of 160 and 1250 ISO and at equal exposure, but now in daylight (6500 Kelvin) instead of the former tests in Tungsten light (2800 Kelvin) and at shorter exposure times (1/60 and f5/6). This time there is no statistical difference anymore between the 3 stops pushed 160 ISO and 1250 ISO exposures for all color patches of the MacBeth card in standard deviation of the luminance.

 

Most likely the different result of the tests comes from the longer exposure times and/or the low blue light contribution in the 2800 Kelvin exposures.

 

Also I have tested 0 sharpening and 0 noise reduction settings when loaded in LightRoom and that still does not show a difference between pushed 160 and 1250 ISO.

 

So there is no support for the statement that pushing is better, nor inferior compared to conventional use of higher ISO in these conditions. Also image quality is very much the same.

 

So what it boils down to is that it is your choice of in camera signal amplification from your ISO setting or post processing by pushing in LightRoom development: it makes no difference, with the exception that you can blow out highlights with the former method and not with the latter. Advantage of the former method is that you have an easier review of exposure on location from the histogram on the LCD.

 

If required I can post a link to the DNG files of the MacBeth exposures.

[k-hawinkler]

Bert,

 

Thanks.

Do you stand by your former tests in tungsten light?

[Lindolfi]

Bert,

 

Thanks.

Do you stand by your former tests in tungsten light?

 

Yes. But I'm still not sure whether it is the long exposure time or the high red contribution in tungsten that makes the difference, so I'll do a test with a 1000x filter in daylight to see. More later.

[Lindolfi]

No, it is not the long exposure time. With the 1000x filter I exposed for 8 seconds in 6500 Kelvin and there is still no difference between the 3 stops pushed 160 and the 1250 ISO in noise. So it must be the high red contribution.

Edited October 27, 2012 by Lindolfi

[k-hawinkler]

Bert,

 

Thanks. Very interesting!

How can you contrast that with green and blue?

[Lindolfi]

Well it must have something to do with the white balance correction, which needs to amplify the blue channel mainly when shooting in tungsten light, while in 6500 Kelvin there is sufficient blue light.

[dwbell]

Just a hypothetical usage situation then.

 

You set say f/4 1/4000 for Sunny 16 (EV15). You then, as light reduces adjust a combination of shutter and aperture to suit until you reach your practical lower limit for both depending on, for example, hand shake, subjective movement and depth of field.

 

At this point you could just keep shooting at those (subjective) values. The screen would not help you in terms of exposure composition of course, because it would all be too dark. However the image quality would not suffer at all.

 

This presumably is no the case on the 5DII for example as it applies in camera noise reduction at high ISO.

[Lindolfi]

This presumably is no the case on the 5DII for example as it applies in camera noise reduction at high ISO.

 

You are right. Just did the test with the Canon 5DII and the luminance noise in the 3 stops pushed 160 ISO is about 55% higher than the 1250 ISO exposure of the MacBeth card. You also see some scottish noise patterns (banding along rows and lines of pixels) in the 160 ISO pushed version that is not visible in the 1250 ISO version.

[k-hawinkler]

Would a Nikon camera behave similarly to the 5DII?

[jaapv]

Could you pick up the alleged NR that Leica applies @ high ISO? ( maybe only at 2500?)

[Jaybob]

You are right. Just did the test with the Canon 5DII and the luminance noise in the 3 stops pushed 160 ISO is about 55% higher than the 1250 ISO exposure of the MacBeth card. You also see some scottish noise patterns (banding along rows and lines of pixels) in the 160 ISO pushed version that is not visible in the 1250 ISO version.

 

similar to these NEX files?

Welcome, dear visitor! As registered member you'd see an image here…

Simply register for free here – We are always happy to welcome new members!

[Advertisement]

Hallo,

Look here: Leica M

[Jaybob]

There have been several examples in this thread alone showing the M9 results pushed in the LR4 converter being superior. It could simply be that Adobe's noise reduction is currently better than Leica's in-camera attempt.

 

If you're a "get it right in the camera" shooter, then ETTR doesn't make sense, either, because you likely have to pull exposure back in the converter.

 

....another set of exposures of a MacBeth color card with the settings of 160 and 1250 ISO and at equal exposure, but now in daylight (6500 Kelvin) instead of the former tests in Tungsten light (2800 Kelvin) and at shorter exposure times (1/60 and f5/6). This time there is no statistical difference anymore between the 3 stops pushed 160 ISO and 1250 ISO exposures for all color patches of the MacBeth card in standard deviation of the luminance.

 

So there is no support for the statement that pushing is better, nor inferior compared to conventional use of higher ISO in these conditions. Also image quality is very much the same.

 

So what it boils down to is that it is your choice of in camera signal amplification from your ISO setting or post processing by pushing in LightRoom development: it makes no difference, with the exception that you can blow out highlights with the former method and not with the latter. Advantage of the former method is that you have an easier review of exposure on location from the histogram on the LCD.

 

 

The one M9 example , where the "numbers dont lie", has shown one underexposure to be marginally better on a set of uncompressed DNGs. The other examples have hardly shown superiority, in fact they have shown almost no noticable difference.

 

A million times more than more than ISO, Noise reduction doesn't have anything at all to do with exposure, at least in this discussion. Proper (I mean dead on) Color balance has more to do with exposure than anything. Lightroom Noise reduction is applied by the user moving the sliders, not "behind the scenes", unless your import settings apply Contrast or Color Noise reduction as the default setting, (they do!). They also can apply brightnesss and contrast, or a lot of other things... My understanding of behind the scenes In camera noise reduction was it was only on JPGs, regardless, it has always in the OFF position on my cameras.

 

I consider mysyself a "get it how I want it, AND give me options later..." RAW shooter in camera, using the histogram in a way that is close to what I need, AND is still going to help me the most later. Having those higher registers of the histogram more populated with information, makes a LOT more sense to me.

Edited October 27, 2012 by Jaybob

[Jaybob]

Would a Nikon camera behave similarly to the 5DII?

 

Just my opinion, meaning I haven't done any testing, but speaking as a Nikon snob, the D3 and D4 generation of sensor is a completely different breed of animal a breed to which the normal rules do not apply. (but one I still try to ETTR...)

 

Again toward, not THROUGH.