Luminous Landscape Forum

Equipment & Techniques => Digital Cameras & Shooting Techniques => Topic started by: ZoranC on June 27, 2012, 01:26:16 am

Title: Michael Reichmann, ETTR and Oly OMD EM5
Post by: ZoranC on June 27, 2012, 01:26:16 am
Now that Michael is using EM5 I am wondering could he share with us his experiences using ETTR on it.

For start, I am wondering what is better to use on it to make sure one didn't overexpose, shadow/highlight blinkies or histogram of all channels? Or both?

Next, when using blinkies, what values would be best for Histogram Settings to indicate over/under exposure and why?

And when it comes to metering when one would want to use Spot Highlight and when one wouldn't want to use it?

Lastly, I am interested in input on Auto Gradation. From what I understand it affects metering (it underexposes by 1/3 of stop) even when one is shooting only RAW. Thus when RAW shooter would want to use it over Normal and vice versa, if ever? And how use of it would interact with ETTR efforts?
Title: Re: Michael Reichmann, ETTR and Oly OMD EM5
Post by: michael on June 27, 2012, 06:45:46 am
No need to over complicate things. Use the blinkers because they are easy to judge at a glance. The highlights are the only important ones unless you are doing lighting on a set and can add fill. I leave the highlight setting at 255.

Just expose so that nothing important blinks, (go over, then back off a touch) and you have optimum raw exposure. Same applies to cany camera with a highlight warning.

Michael
Title: Re: Michael Reichmann, ETTR and Oly OMD EM5
Post by: ZoranC on June 27, 2012, 01:47:07 pm
No need to over complicate things. Use the blinkers because they are easy to judge at a glance. The highlights are the only important ones unless you are doing lighting on a set and can add fill. I leave the highlight setting at 255.

Just expose so that nothing important blinks, (go over, then back off a touch) and you have optimum raw exposure. Same applies to cany camera with a highlight warning.
Thank you! I am a big fan of KISS approach, but for a sake of better understanding please allow me few more questions.

It is my understanding (please correct me if I am wrong) that blinkies, just like histogram, depend not on data in RAW but on data in JPG preview that gets saved with RAW, and in turn JPG preview depends on settings for Picture Mode, Saturation, Contrast and Gradation. Thus am I correct in understanding that one would want them set to something like Muted, toned down Saturation and Contrast, and Normal Gradation while deciding what optimal RAW exposure is? If yes would ETTR at such settings introduce bigger chance of later blowing highlights in the post-processing? I am asking because I think I have seen a shot taken with ETTR and Muted that was not blown yet became blown once one applied iEnhance in post to it's RAW.
Title: Re: Michael Reichmann, ETTR and Oly OMD EM5
Post by: fike on July 11, 2012, 11:40:38 am
Thank you! I am a big fan of KISS approach, but for a sake of better understanding please allow me few more questions.

It is my understanding (please correct me if I am wrong) that blinkies, just like histogram, depend not on data in RAW but on data in JPG preview that gets saved with RAW, and in turn JPG preview depends on settings for Picture Mode, Saturation, Contrast and Gradation. Thus am I correct in understanding that one would want them set to something like Muted, toned down Saturation and Contrast, and Normal Gradation while deciding what optimal RAW exposure is? If yes would ETTR at such settings introduce bigger chance of later blowing highlights in the post-processing? I am asking because I think I have seen a shot taken with ETTR and Muted that was not blown yet became blown once one applied iEnhance in post to it's RAW.

I think Michael is advocating for a simple and common-sense approach.  First deliberately overexpose, and then back off a bit.  Don't make ETTR harder than it needs to be. 

If you really want to get precise about this there are convoluted methods for setting white balance that may help you to get closer to some theoretical maximum exposure, but it is complicated and annoying. 

ETTR isn't extremely helpful in a very high dynamic range environment (where the histogram exceeds the range of the scale on both left and right).  Think of ETTR as more helpful when you have a very narrow hump in the middle of your histogram, with no extra lobes to the right.  In that case (most commonly repeatable in dense fog) in preparation for RAW post-processing, you want to move it over to the right side, but don't try to get it so close that you are likely to accidentally clip because your histogram is using the jpg image for the highlight warning.  In that case you are courting disaster. 
Title: Re: Michael Reichmann, ETTR and Oly OMD EM5
Post by: richarddd on January 02, 2013, 02:43:15 pm
If changing aperture and shutter speed isn't enough to get to the point where there are blinkies (perhaps because of desired DOF or lack of blur), does it make sense to increase ISO?
Title: Re: Michael Reichmann, ETTR and Oly OMD EM5
Post by: ErikKaffehr on January 02, 2013, 02:50:15 pm
Hi,

Increasing ISO is a bad idea. The reason behind ETTR is to maximize exposure. You want to capture as many photons as possible.

Best regards
Erik


If changing aperture and shutter speed isn't enough to get to the point where there are blinkies (perhaps because of desired DOF or lack of blur), does it make sense to increase ISO?
Title: Re: Michael Reichmann, ETTR and Oly OMD EM5
Post by: ErikKaffehr on January 02, 2013, 03:01:04 pm
+1

Erik

No need to over complicate things. Use the blinkers because they are easy to judge at a glance. The highlights are the only important ones unless you are doing lighting on a set and can add fill. I leave the highlight setting at 255.

Just expose so that nothing important blinks, (go over, then back off a touch) and you have optimum raw exposure. Same applies to cany camera with a highlight warning.

Michael
Title: Re: Michael Reichmann, ETTR and Oly OMD EM5
Post by: digitaldog on January 02, 2013, 03:02:08 pm
Increasing ISO is a bad idea.

On some cameras yes, on some, it's the same as ETTR if handled as such.
Title: Re: Michael Reichmann, ETTR and Oly OMD EM5
Post by: richarddd on January 02, 2013, 03:40:10 pm
Increasing ISO is a bad idea. The reason behind ETTR is to maximize exposure. You want to capture as many photons as possible.
Increasing ISO does not change the amount of light coming into the camera. Assume you've already set optimal aperture and shutter speed.

http://www.luminous-landscape.com/forum/index.php?topic=56906.40 suggests the answer to the general question regarding increasing ISO once you've done all you can with exposure is "it depends" on the camera, which is digitaldog's answer.  I'm wondering what the answer is for the E-M5

In my tests so far, increasing ISO at worst doesn't hurt and at best clearly helps, provided that I don't go above 1600.  From what I've read, values between 200 and 400 should be avoided.  Before drawing any conclusions I'd like to hear from others, as I might not be testing properly, ignoring important cases, etc.

Might dynamic range be an issue in addition to noise, or does a change in one make a corresponding change in the other?
Title: Re: Michael Reichmann, ETTR and Oly OMD EM5
Post by: Vladimirovich on January 02, 2013, 04:33:22 pm
Increasing ISO does not change the amount of light coming into the camera. Assume you've already set optimal aperture and shutter speed.

but bigger gain (it it is done not using tags - like GH3 does for ISO above 6400) may result in raw data clipping sooner than lower gain...  however if you claim that you have already the best aperture and shutter speed set and you are not ETTR fan (= you do not want more light, even you have room for that) then based on your particular camera ( Canon-like sensors) you might want to increase the gain to deal w/ lower read noise... is it what you are trying to assert ?
Title: Re: Michael Reichmann, ETTR and Oly OMD EM5
Post by: richarddd on January 02, 2013, 04:52:55 pm
but bigger gain (it it is done not using tags - like GH3 does for ISO above 6400) may result in raw data clipping sooner than lower gain...  however if you claim that you have already the best aperture and shutter speed set and you are not ETTR fan (= you do not want more light, even you have room for that) then based on your particular camera ( Canon-like sensors) you might want to increase the gain to deal w/ lower read noise... is it what you are trying to assert ?
What I'm asking (for the E-M5) is, assuming I've set optimum aperture and shutter speed, is it a good idea in terms of image quality to increase ISO to just below the clipping point (where clipping is determined by blinkies, as Michael stated), so long as you don't go above 1600? 

By optimum aperture and shutter speed, I mean letting in as much light as possible, consistent with desired DOF and blur. I would hope even ETTR fans don't set shutter speeds so low that they get subject or camera blur or aperture so wide they don't get appropriate DOF, but in any event that's not something I want to do.

If this strategy is best to get lower read noise, are there any countervailing issues that would suggest it is not a good idea?
Title: Re: Michael Reichmann, ETTR and Oly OMD EM5
Post by: fike on January 02, 2013, 05:11:10 pm
I don't know the answer, but I am going to speculate a bit. 

As has been mentioned, assuming you keep the aperture and shutter the same you aren't actually increasing the amount of light entering the camera.  What you are doing is boosting the signal (with it the noise) so that it moves the data into the brighter half of the histogram--towards the right.  I think this would be advantageous because you would be moving the data into the region of the histogram where the sampling rate is higher.  The signal to noise ratio would be the same, but the gradations of light sensitivity would be finer. 

I think that you will not see a difference in noise-performance because boosting in camera (increasing ISO) or boosting in post processing (exposure slider in ACR) achieve very similar results.  With that said, I have come full circle because one of the main purposes of ETTR is to increase the signal to noise ratio.  Your strategy doesn't accomplish an increase in SNR. It only allows you to capture more tonality within the dynamic range that is present. 

There is no magic here. If you are working at 1600 ISO and are underexposing by 3 stops, you are not much worse or better off than going to 12,800 ISO and having the "correct" exposure (aside from the fact that you will recover some shadow detail at 12,800 and it will be total crap).  The details of these relationships are not precisely the same for all manufacturers, but generally speaking these relationships hold true. 

So you get half-credit for your approach within the limitations you describe.
Title: Re: Michael Reichmann, ETTR and Oly OMD EM5
Post by: Vladimirovich on January 02, 2013, 05:58:32 pm
By optimum aperture and shutter speed, I mean letting in as much light as possible, consistent with desired DOF and blur.

define possible about exposure time = possible based on clipping or possible based on subject moving/stabilization/etc

if the situation imposes the limit on how long your exposure can be and you are not close to clipping @ lower gain  then you might want (based on your camera) increase the gain to lower read noise - so you test your camera using the tools available to see what is the max gain after which there no advantage in read noise reduction.
Title: Re: Michael Reichmann, ETTR and Oly OMD EM5
Post by: AFairley on January 02, 2013, 07:00:45 pm
This is a case where the proof of the pudding is in the eating and the answer can be empirically shown by shooting both ways (iso boost or no iso boost) and seeing which one PPs best, which, after all, is what you are after.
Title: Re: Michael Reichmann, ETTR and Oly OMD EM5
Post by: richarddd on January 02, 2013, 07:20:08 pm
define possible about exposure time = possible based on clipping or possible based on subject moving/stabilization/etc
Start at high shutter speed.  Decrease until first of (1) clipping and (2) subject moving/stabilization/etc

if the situation imposes the limit on how long your exposure can be and you are not close to clipping @ lower gain  then you might want (based on your camera) increase the gain to lower read noise - so you test your camera using the tools available to see what is the max gain after which there no advantage in read noise reduction.
Which available tools?

Aren't there are issues besides read noise, such as shot noise and dynamic range?
Title: Re: Michael Reichmann, ETTR and Oly OMD EM5
Post by: richarddd on January 02, 2013, 07:25:41 pm
This is a case where the proof of the pudding is in the eating and the answer can be empirically shown by shooting both ways (iso boost or no iso boost) and seeing which one PPs best, which, after all, is what you are after.
As I mentioned above, testing so far shows either no harm or improvement in increasing ISO up to 1600.  I set ISO to 1600, set aperture and shutter speed just below having some areas clip, then retook the same shot at ISO 200. In LR4 I used the exposure slider to increase by 3 stops, which matched brightness levels, then looked at both pictures.

I'm concerned that there are cases I'm not testing (perhaps higher dynamic range, perhaps others) or that I'm otherwise missing something.
Title: Re: Michael Reichmann, ETTR and Oly OMD EM5
Post by: richarddd on January 02, 2013, 07:32:38 pm
I think that you will not see a difference in noise-performance because boosting in camera (increasing ISO) or boosting in post processing (exposure slider in ACR) achieve very similar results.
Is there any reason to believe boosting ISO and boosting in post will achieve similar results?  I've seen examples with some cameras showing quite different results. For example, the second post in this thread http://www.luminous-landscape.com/forum/index.php?topic=56906.0 Alas, that thread deals with cameras other than the E-M5
Title: Re: Michael Reichmann, ETTR and Oly OMD EM5
Post by: Vladimirovich on January 02, 2013, 09:07:12 pm
such as shot noise

you can't do anything about shot noise because you can't increase exposure time (you are bounded by either raw data clipping that you do not want or your subject movement or your lens/sensor stabilization or something else) and you can't increase aperture - your own words... so you can't get more light = your shot noise is what it is... the only chance you have is to pour some liquid nitrogen to cool your sensor  ::) or to may be reduce the read noise (if your camera has such sensor/adc/firmware) by increasing the gain (that is only if your exposure time is limited not by clipping).

and dynamic range?

what about DR ? you were in a situation when you can get more exposure time to get more light... it is either clipping or subject moving or something else... if it is not clipping you can try to lower read noise affecting your deep shadows by applying a bigger gain (but test your own camera and converter).
Title: Re: Michael Reichmann, ETTR and Oly OMD EM5
Post by: Vladimirovich on January 02, 2013, 09:14:03 pm
Is there any reason to believe boosting ISO and boosting in post will achieve similar results?
certainly, at some point it makes no sense to further increase the gain in camera (unless you need in camera JPG for whatever reason)... but it is specific to a camera model (and may be even a specific unit) and then probably not all raw converters were created equal.
Title: Re: Michael Reichmann, ETTR and Oly OMD EM5
Post by: Vladimirovich on January 02, 2013, 09:15:55 pm

Which available tools?


I think you were pointed to such tools more than one time in the past.
Title: Re: Michael Reichmann, ETTR and Oly OMD EM5
Post by: richarddd on January 02, 2013, 09:55:50 pm
@Vladimirovich, regarding shot noise and DR, I was trying to figure out if there were measures other than read noise that might be relevant. Shot noise isn't relevant, as you say, because we are assuming a fixed exposure. DR shouldn't be an issue if we're below highlight clipping.

Regarding available tools, are we communicating somewhere else? I haven't posted here about ETTR and the E-M5. In any event, others on this forum might be interested in the answer.
Title: Re: Michael Reichmann, ETTR and Oly OMD EM5
Post by: Vladimirovich on January 02, 2013, 10:14:52 pm
Shot noise isn't relevant, as you say, because we are assuming a fixed exposure.

it is relevant, you just can't increase SNR (signal vs shot noise, which is the dominant component beyond dark shadows) by increasing exposure... and your question is, if I understand it correctly, what else can you do to make your data better... the only noise that you can reduce are the noise because of the temperature and because of charge readout/ADC circuitry... hence cool your sensor/circuitry and/or try to play w/ gain increase (in you do not clip)...

Title: Re: Michael Reichmann, ETTR and Oly OMD EM5
Post by: Graystar on January 03, 2013, 01:12:22 am
Increasing ISO is a bad idea. The reason behind ETTR is to maximize exposure. You want to capture as many photons as possible.

That's my understanding of ETTR as well.  As far as I'm aware, ETTR refers to increasing exposure to make use of unutilized sensor DR.  If you shift the histogram to the left to protect highlights...that's not ETTR.  Sure, you may be trying to get those highlights as far right as possible without clipping, But ETTR specifically refers to setting an exposure greater than Standard Exposure for the expressed purpose of reducing noise.

With that as a definition, I'm quite certain that increasing ISO provides no benefit.  The reason is that exposure is reduced in post processing...not increased.

There have been images demonstrating that some cameras produce better results at high ISO.  This is done by taking ISO 1600 or ISO 3200 images (and sometimes increasing exposure on those by a few stops) and comparing to images taken at ISO 100 or ISO 200 and then increasing exposure in post by 5 EV or more.  The reason that these examples give the results they do is that the difference in noise (of which there really is less of at the higher ISO settings) is amplified by the boosted exposure.

But with ETTR, exposure isn't increased in post...it's reduced.  And when you reduce exposure, some of the bits representing the improved noise levels will simply be chopped off.  So whereas the noise difference in those "proof of concept" images was boosted into the sunlight, any real noise difference in actual images will be suppressed into obscurity by the application of negative EC.

No one's been able to show me a pair of images demonstrating improved noise via ETTR with elevated ISO levels.  If anyone has some I'd be interested in seeing them.  But until then, I have to agree with those who say ETTR only valid at base ISO.
Title: Re: Michael Reichmann, ETTR and Oly OMD EM5
Post by: Vladimirovich on January 03, 2013, 02:27:28 am
OP has limits on exposure time and aperture and provided that raw data is not clipped in his situation the question is - whether it makes sense to increase the gain in camera... he can't get more light (remember - no increase in exposure time or aperture), can he get somewhat less noise... the answer is - yes, he can, by increasing the gain (in some cameras, up to a certain limit, provided that he is not clipping raw data or otherwise getting close enough to the saturation capacity which may /or may not/ be not good for some cameras) reduce the read noise... what's wrong w/ that ? he can't get bigger SNR by getting more light under his conditions, he might get bigger SNR in shadows by getting less read noise.
Title: Re: Michael Reichmann, ETTR and Oly OMD EM5
Post by: richarddd on January 03, 2013, 06:07:57 am
No one's been able to show me a pair of images demonstrating improved noise via ETTR with elevated ISO levels.  If anyone has some I'd be interested in seeing them.  But until then, I have to agree with those who say ETTR only valid at base ISO.
See reply #1 in http://www.luminous-landscape.com/forum/index.php?topic=56906.0 (the link was posted in reply #16 to this thread)  The photos were taken with an older Canon, but I believe the E-M5 has the same behavior.
Title: Re: Michael Reichmann, ETTR and Oly OMD EM5
Post by: Graystar on January 03, 2013, 08:20:05 am
See reply #1 in http://www.luminous-landscape.com/forum/index.php?topic=56906.0 (the link was posted in reply #16 to this thread)  The photos were taken with an older Canon, but I believe the E-M5 has the same behavior.
Please reread my post.  This is exactly the type of "proof" I was referring to.  The image taken at ISO 100 had its exposure increased in post to match the image taken at ISO 1600.  That is not what happen in the ETTR method.  ETTR does the reverse...an overexposed image has its exposure reduced in post.  In the "proof" the noise difference is amplified.  With ETTR the noise difference is diminished.

The EM-5 has a sensor made by Sony, and typically the latest Sony sensor don't exhibit this behavior.  The EM-5's SNR is curve at DxOMark doesn't seem to indicate that it does, so I would be surprised if the EM-5 behaved this way.
Title: Re: Michael Reichmann, ETTR and Oly OMD EM5
Post by: richarddd on January 03, 2013, 09:05:19 am
Please reread my post.  This is exactly the type of "proof" I was referring to.  The image taken at ISO 100 had its exposure increased in post to match the image taken at ISO 1600.  That is not what happen in the ETTR method.  ETTR does the reverse...an overexposed image has its exposure reduced in post.  In the "proof" the noise difference is amplified.  With ETTR the noise difference is diminished.

The EM-5 has a sensor made by Sony, and typically the latest Sony sensor don't exhibit this behavior.  The EM-5's SNR is curve at DxOMark doesn't seem to indicate that it does, so I would be surprised if the EM-5 behaved this way.
My posts are about setting exposure and wondering if increasing ISO to just below the clipping point will help image quality. Exposure is set at the maximum consistent amount with desired blur (or lack of) and DOF.  

The photos I posted show increasing ISO decreases noise.

Is your post responsive to the question of whether holding exposure constant and increasing ISO (up to say 1600) will improve image quality for the E-M5?

In your post, you agree that "Increasing ISO is a bad idea. The reason behind ETTR is to maximize exposure. You want to capture as many photons as possible."  As increasing ISO does not change the number of photons captured, why is increasing ISO a bad idea?

The usual reason increasing ISO is a bad idea is that it is often associated with decreasing exposure, which is a bad idea. That's not an issue here; we're holding exposure constant.
Title: Re: Michael Reichmann, ETTR and Oly OMD EM5
Post by: Vladimirovich on January 03, 2013, 09:44:05 am
The EM-5 has a sensor made by Sony
it does not.
Title: Re: Michael Reichmann, ETTR and Oly OMD EM5
Post by: Vladimirovich on January 03, 2013, 09:48:14 am
The EM-5's SNR is curve at DxOMark doesn't seem to indicate that it does
and we are not talking about SNR @ 18% which is what DxOMark curve shows... @ 18% it is dominated by shot noise... we are talking about shadows (deep) and read noise reduction... there the relevant curve is DR curve (non Sony looking curve by the way... it is Panasonic curve).
Title: Re: Michael Reichmann, ETTR and Oly OMD EM5
Post by: fike on January 03, 2013, 10:37:25 am
That's my understanding of ETTR as well.  As far as I'm aware, ETTR refers to increasing exposure to make use of unutilized sensor DR.  If you shift the histogram to the left to protect highlights...that's not ETTR.  Sure, you may be trying to get those highlights as far right as possible without clipping, But ETTR specifically refers to setting an exposure greater than Standard Exposure for the expressed purpose of reducing noise.

With that as a definition, I'm quite certain that increasing ISO provides no benefit.  The reason is that exposure is reduced in post processing...not increased.

...

I was under the impression that there were two benefits to ETTR.
1) as you say increase signal to noise ratio.
2) Place the signal you have into the right half of the histogram where you have more exposure gradations, ie higher sampling rate. 

In the case of #1, there is no benefit to increasing ISO as the OP mentions.  In these case of #2, there is a small benefit because after normalization of the tone curve, it may reduce the effect of posterization.

 
Title: Re: Michael Reichmann, ETTR and Oly OMD EM5
Post by: Graystar on January 03, 2013, 10:59:00 am
it does not. [have a sony sensor]

A few months ago there was news that the president of Olympus had stated that the EM-5 sensor is made by Sony.

http://www.imaging-resource.com/news/2012/07/06/olympus-ceo-identifies-sensor-supplier-for-flagship-mirrorless-camera

Unfortunately, the link to the original story at Yahoo Japan no longer works.  However, Sony DID invest some 645 million dollars in Olympus at the end of September, and if Sony wasn't supplying sensors to Olympus previously, it definitely is going to do so, as that fact was already announced in October...
http://www.43rumors.com/olympus-and-sony-conference-in-tokyo
Title: Re: Michael Reichmann, ETTR and Oly OMD EM5
Post by: Graystar on January 03, 2013, 11:18:20 am
My posts are about setting exposure and wondering if increasing ISO to just below the clipping point will help image quality. Exposure is set at the maximum consistent amount with desired blur (or lack of) and DOF. 

The photos I posted show increasing ISO decreases noise.

Is your post responsive to the question of whether holding exposure constant and increasing ISO (up to say 1600) will improve image quality for the E-M5?

In your post, you agree that "Increasing ISO is a bad idea. The reason behind ETTR is to maximize exposure. You want to capture as many photons as possible."  As increasing ISO does not change the number of photons captured, why is increasing ISO a bad idea?

The usual reason increasing ISO is a bad idea is that it is often associated with decreasing exposure, which is a bad idea. That's not an issue here; we're holding exposure constant.

If you decrease ISO, then you can increase exposure with a slower shutter or wider aperture.  If you have a shutter and aperture limit, then you can't really perform ETTR, as the goal of ETTR is to maximizing your signal.  Sure, you can increase your ISO to push the histogram to the right...but your exposure will always be less than optimal.

As to whether you get any benefit from pushing the histogram to the far right via ISO, as I said I believe there is no benefit.  The reason is that the process of applying negative EC to correct the exposure will likely wipe out any gains in reduced noise.  Those gains are in the least significant bits of data, and those are the bits that get dropped when you apply negative EC, rendering any gains practically invisible.
Title: Re: Michael Reichmann, ETTR and Oly OMD EM5
Post by: digitaldog on January 03, 2013, 11:24:22 am
Please reread my post.  This is exactly the type of "proof" I was referring to.  The image taken at ISO 100 had its exposure increased in post to match the image taken at ISO 1600.

In the example I have to offer below, based on the post referred to, it's the opposite. The ISO image was shot "normal" based on an incident meter. The ISO 800 image had to be normalized in post to match as we expect to see with all ETTR images. Otherwise the default rendering appears "over exposed" (but of course, it's not).

http://digitaldog.net/files/100vs800iso.jpg

ISO 800 on this Canon has far less noise than 100. Due to ETTR as Guillermo Luijk first illustrated.
Title: Re: Michael Reichmann, ETTR and Oly OMD EM5
Post by: Graystar on January 03, 2013, 11:35:57 am
I was under the impression that there were two benefits to ETTR.
1) as you say increase signal to noise ratio.
2) Place the signal you have into the right half of the histogram where you have more exposure gradations, ie higher sampling rate. 

In the case of #1, there is no benefit to increasing ISO as the OP mentions.  In these case of #2, there is a small benefit because after normalization of the tone curve, it may reduce the effect of posterization.

Actually that case #2 has also been debunked by our in-house science professor, Dr. Emil Martin and his work on image noise.  Also, the levels reasoning forgets that image values are not taken directly from RAW values.  A RAW value must be demosaiced, during which time a dozen or more other RAW values will influence the final result of the one pixel's RAW conversion.  Bottom line is that the "levels" limitation only really exists for the least significant bits, which render to near black anyways.
Title: Re: Michael Reichmann, ETTR and Oly OMD EM5
Post by: richarddd on January 03, 2013, 11:52:13 am
If you decrease ISO, then you can increase exposure with a slower shutter or wider aperture.  If you have a shutter and aperture limit, then you can't really perform ETTR, as the goal of ETTR is to maximizing your signal.  Sure, you can increase your ISO to push the histogram to the right...but your exposure will always be less than optimal.
Higher ISO amplifies the signal. Whether it does a good job is an empirical question.

As to whether you get any benefit from pushing the histogram to the far right via ISO, as I said I believe there is no benefit.  The reason is that the process of applying negative EC to correct the exposure will likely wipe out any gains in reduced noise.  Those gains are in the least significant bits of data, and those are the bits that get dropped when you apply negative EC, rendering any gains practically invisible.
The Canon shot I posted shows a clear benefit from shooting with a higher ISO, even though exposure (i.e., light on the sensor) was unchanged.

We won't necessarily be reducing brightness in post, although it's a distinct possibility.  Exposure is a given and ISO is set so that important highlights are just below clipping, but not beyond 1600. This does not necessarily mean a shot that is too bright.
Title: Re: Michael Reichmann, ETTR and Oly OMD EM5
Post by: bjanes on January 03, 2013, 11:55:24 am
Thank you! I am a big fan of KISS approach, but for a sake of better understanding please allow me few more questions.

It is my understanding (please correct me if I am wrong) that blinkies, just like histogram, depend not on data in RAW but on data in JPG preview that gets saved with RAW, and in turn JPG preview depends on settings for Picture Mode, Saturation, Contrast and Gradation. Thus am I correct in understanding that one would want them set to something like Muted, toned down Saturation and Contrast, and Normal Gradation while deciding what optimal RAW exposure is? If yes would ETTR at such settings introduce bigger chance of later blowing highlights in the post-processing? I am asking because I think I have seen a shot taken with ETTR and Muted that was not blown yet became blown once one applied iEnhance in post to it's RAW.

You are correct. The histogram and blinkies are from the JPEG preview and not from the raw file itself. Most cameras allow some headroom for highlights by using an ISO sensitivity somewhat higher than the Ssat value measured by DXO, resulting in underexposure. So that the LCD preview and histogram data do not show underexposure, they are adjusted to the right which effectively masks the degree of underexposure. The KISS approach yields satisfactory but not optimal results and one must know one's camera to get better results.

I don't have the Olympus model under discussion, so I will show results for the Nikon D800e. The principles are the same for any camera, but the results will vary. The first step is to expose a uniformly illuminated 18% gray card according to the camera meter (the same results can be obtained by exposing a white target or the computer screen filled with white). I used my computer screen calibrated to a WB of 6500K. The spike in the histogram is somewhat to the right of the expected mid value because of the tone curve applied by the camera (the Nikon standard picture control in this case). The Adobe RGB pixel value was 156 rather than the 117 that would be expected with midgray.

(http://bjanes.smugmug.com/Photography/D800-ETTR/i-hP6t8Rk/0/O/D800eHistogramMidGray_0001_small.png)

The next step is to look at the raw file. Rawdigger is excellent for this purpose. The gray value for the green channels is 3 EV short of clipping, whereas 18% gray should be 2.47 EV from clipping. This allows for 0.5 EV of highlight headroom. The raw pixel value in the 14 bit raw file for the green is 1986 to give a saturation of 12.1%. In the gamma 2.2 Adobe RGB this corresponds to an 8 bit pixel value of 98.

(http://bjanes.smugmug.com/Photography/D800-ETTR/i-9fTnxbd/0/O/MidGray_0001-Sel-3439-2079-359x519.png)

To get an ETTR exposure, one would have to add 3 EV, and this does show the green channels just short of clipping.

(http://bjanes.smugmug.com/Photography/D800-ETTR/i-md5MVL4/0/O/MidGrayExpPlus_3-Sel-3219-1999-799x759.png)

The next step is to look at the camera histograms and blinking highlights for these exposures. As shown in the table below, the plus 2.33 EV exposure gives no clipping in the camera histogram and blinking highlights, and the plus 2.67 EV exposure does show clipping in the histogram and blinking highlights. Backing off of the plus 2.67 EV exposure to the Plus 2.33 one as suggested would give a raw file 0.67 EV short of full ETTR. The camera histogram for the Plus 2.33 EV exposure is shown (one should use the RGB histogram in case there is clipping in the color channels). The ACR results rendered into Adobe RGB are also shown for those who judge ETTR exposure with ACR/LR. For the most accurate results one should use PV2010 with -0.35 EV exposure to compensate for the baseline offset ACR uses for this camera.

(http://bjanes.smugmug.com/Photography/D800-ETTR/i-ftJLVRG/0/O/D800eHistoPlus2_33.png)

(http://bjanes.smugmug.com/Photography/D800-ETTR/i-fw64zwM/0/O/ExcellResults.png)

Bill




Title: Re: Michael Reichmann, ETTR and Oly OMD EM5
Post by: Graystar on January 03, 2013, 12:02:45 pm
In the example I have to offer below, based on the post referred to, it's the opposite. The ISO image was shot "normal" based on an incident meter. The ISO 800 image had to be normalized in post to match as we expect to see with all ETTR images. Otherwise the default rendering appears "over exposed" (but of course, it's not).

http://digitaldog.net/files/100vs800iso.jpg

ISO 800 on this Canon has far less noise than 100. Due to ETTR as Guillermo Luijk first illustrated.

Thanks for the example.  Can you clarify your statement...I think you might have left out an ISO value when you said, "The ISO image was shot 'normal'..."  I think you meant, "the ISO 100 image was shot 'normal'..." but don't want to put words in your mouth.  Presuming this is the case...

I see that the bright area of the ISO 100 image has RGB values of just over 200.  On my Nikon D90 I know that such levels would just begin to clip if overexposed by one EV (the Nikon D800 has similar exposure characteristics.)  I'm surprised you were able to amplify that image by 3 stops without any clipping.  That 5D MII must provide a tremendous amount of headroom.

Do you know of the 5D MIII produces a similar result?
Title: Re: Michael Reichmann, ETTR and Oly OMD EM5
Post by: Vladimirovich on January 03, 2013, 12:10:09 pm
The reason is that the process of applying negative EC to correct the exposure will likely wipe out any gains in reduced noise.  Those gains are in the least significant bits of data, and those are the bits that get dropped when you apply negative EC, rendering any gains practically invisible.

we shall assume that you tested all raw converters  8) ?
Title: Re: Michael Reichmann, ETTR and Oly OMD EM5
Post by: Vladimirovich on January 03, 2013, 12:16:37 pm
The next step is to look at the raw file. Rawdigger is excellent for this purpose. The gray value for the green channels is 3 EV short of clipping, whereas 18% gray should be 2.47 EV from clipping. This allows for 0.5 EV of highlight headroom.
camera metering (spot metering or matrix/centerw off a uniform surface) does not care what is your target actual reflectance... it is assuming some (manufacturer defined) reflectance regardless whether you meter off true 18% grey card or white snow and in most cases it is 12.x% (some camera models do differently)... and head room for different channels will be different for different illumination... so you test for a daylight, but you test for tungsten bulb as well and so on.
Title: Re: Michael Reichmann, ETTR and Oly OMD EM5
Post by: Graystar on January 03, 2013, 12:17:34 pm
The Adobe RGB pixel value was 156 rather than the 117 that would be expected with midgray.

Oy...that drives me crazy.  What I realized is that the picture control has some sort of auto-levels function that increasing the brightness.  I created a custom picture control from the Neutral PC using ViewNX, and that gives me the expected value on the camera.
Title: Re: Michael Reichmann, ETTR and Oly OMD EM5
Post by: Vladimirovich on January 03, 2013, 12:19:06 pm
That 5D MII must provide a tremendous amount of headroom.
may be Canon's just meter in such way, despite the fact that their sensors do not do a good job in deep shadows
Title: Re: Michael Reichmann, ETTR and Oly OMD EM5
Post by: Graystar on January 03, 2013, 12:25:37 pm
we shall assume that you tested all raw converters  8) ?

Well...they say that when you assume you make an ASS of U and ME.  So I'd say better not to assume.  8)

It seems to be the case with Raw Therapee, Lightroom, and ViewNX.  I'd assume others are the same. (wait...what did I just do?) ;)
Title: Re: Michael Reichmann, ETTR and Oly OMD EM5
Post by: digitaldog on January 03, 2013, 01:00:02 pm
I think you might have left out an ISO value when you said, "The ISO image was shot 'normal'..."  I think you meant, "the ISO 100 image was shot 'normal'..." but don't want to put words in your mouth.  Presuming this is the case...

Do you know of the 5D MIII produces a similar result?

I used an incident meter, set at ISO 100 and used that for the ISO 100 image in terms of shutter and aperture. I then left everything alone expect I upped the ISO to 800.

Don't know about the Mark III. I believe most of not all Canon's exhibit this behavior but not necessarily other camera systems. In the original linked forum posts, I think someone suggested Nikon works differently.
Title: Re: Michael Reichmann, ETTR and Oly OMD EM5
Post by: Graystar on January 03, 2013, 01:28:11 pm
I used an incident meter, set at ISO 100 and used that for the ISO 100 image in terms of shutter and aperture. I then left everything alone expect I upped the ISO to 800.

Got it.  Thanks...I'll keep that example in mind when discussing these points.
Title: ETTR and Oly OMD EM5 -- and bracketing ISO if there is time
Post by: BJL on January 03, 2013, 01:32:10 pm
See reply #1 in http://www.luminous-landscape.com/forum/index.php?topic=56906.0 (the link was posted in reply #16 to this thread)  The photos were taken with an older Canon, but I believe the E-M5 has the same behavior.
On the contrary, the sensor of the E-M5 behaves far more like the sensors of the Pentax K-5 and Nikon D7000 mentioned in that reply:
Quote
If you use the same shutter speed and f-stop, noise will be more visible the lower the ISO. ...
On some cameras (e.g. Pentax K5, Nikon D7000) the difference will be negligible, in some others (e.g. Canons) will be monstruous.
Putting aside the debate over whether the E-M5 sensor shares a Sony heritage with the K-5 and D7000 sensors, the measurements of dynamic range and SNR at 18% mid-tone level at DxOMark show that for image quality in raw files at equal exposure level (same f-stop and shutter speed), there is no gain beyond about ISO setting of 800: underexposing as low as 800 then pushing in raw conversion will probably give very similar noise characteristics, with the advantage of extra highlight headroom. Then again, E-M5 raw files have rather abundant raw headroom, over one stop, so that might not be a big deal.

By the way, as far as noise levels near midtone level, the graphs of SNR at 18% suggest that there is very little benefit to going above the minimum ISO setting of 200, and no measurable gain in going beyond 400.

On the other hand, underexposing will degrade in-camera JPEG's.  I use JPEG+raw, partly to have the JPEG's as "proofs" with all of the Olympus fixes for distortion and such applied, so when I have time, I prefer to bracket in this situation, starting with a frame at a "correct" combination of exposure level and ISO sensitivity setting.
Title: Re: ETTR and Oly OMD EM5 -- and bracketing ISO if there is time
Post by: Vladimirovich on January 03, 2013, 01:50:14 pm
Putting aside the debate over whether the E-M5 sensor shares a Sony heritage with the K-5 and D7000 sensors
"sony heritage" is multiple ADC on die... Panasonic and Nikon/Aptina (?) designs also have such things... however implementations of Sony sensors in Pentax/Nikon cameras show linear DR graph.... and non Sony implementations show a bit of Canon like non linearity near lowest ISO (see Panasonic GH2, non Sony Nikons) and EM5 DR graph is exactly non Sony like...
Title: Re: Michael Reichmann, ETTR and Oly OMD EM5
Post by: bjanes on January 03, 2013, 01:58:45 pm
camera metering (spot metering or matrix/centerw off a uniform surface) does not care what is your target actual reflectance... it is assuming some (manufacturer defined) reflectance regardless whether you meter off true 18% grey card or white snow and in most cases it is 12.x% (some camera models do differently)... and head room for different channels will be different for different illumination... so you test for a daylight, but you test for tungsten bulb as well and so on.

That is true regarding the results obtained with targets of different reflectance. That is why I said one could meter off the white of a computer screen rather than from a gray card. This will give the same results as reading off a gray card. Nikon and most other meters are calibrated for a luminance value that is roughly equivalent to the reflectance of 12% gray. This was pointed out by Thom Hogan (http://www.bythom.com/graycards.htm) way back in 2003.

With tungsten illumination, the head room for the green channel does not change much as shown below. Since green is heavily weighted for the luminance histogram, there won't be much of a change with tungsten as compared to daylight illumination. However, there will be less headroom for red and more headroom for the blue.

Bill
Title: highlight blinkies are over-cautious for raw, especially with the E-M5
Post by: BJL on January 03, 2013, 02:17:22 pm
You are correct. The histogram and blinkies are from the JPEG preview and not from the raw file itself. Most cameras allow some headroom for highlights by using an ISO sensitivity somewhat higher than the Ssat value measured by DXO, resulting in underexposure.
...
As shown in the table below, the plus 2.33 EV exposure gives no clipping in the camera histogram and blinking highlights, and the plus 2.67 EV exposure does show clipping in the histogram and blinking highlights. Backing off of the plus 2.67 EV exposure to the Plus 2.33 one as suggested would give a raw file 0.67 EV short of full ETTR.
Thanks Bill. The short version is that:
1. From your test, it looks safe to push the D800E exposure level up in, say, half stop increments and then use the the first level that _does_ give highlight blinkies, rather than backing off to avoid them.
2. This should be even safer with the E-M5, which gives raw files with more highlight headroom than most (a bigger gap between the ISO Standard Output Sensitivity calibration of its JPEG output and and the ISO Saturation based measurement of its raw output).


P. S. A camera having an ISO Standard Output Sensitivity [SSOS] higher than its ISO Saturation-based Sensitivity [Ssat] is not "underexposing". The two numbers are simply different measurements of different output formats, with related but different purposes. Imatest calculates both SSOS and Ssat, and has a useful discussion at http://www.imatest.com/docs/sensitivity_ei/
The size of the gap is primarily a measure of raw highlight headroom and the Imatest page above gives this highlight headroom in its equation (8):
HdrSOS(f-stops) = log2( SSOS ⁄ Ssat ) + 0.5      (f-stops or EV or zones)

As I have mentioned in numerous times, Japanese camera makers are more or less required by their industry association CIPA to use ISO Standard Output Sensitivity as the primary method of calibration of the ISO sensitivity settings on their cameras. Insisting that SSOS = SSat would forbid a camera from offering more or less than 1/2 stop of highlight headroom in its raw files, which would be an absurdly rigid restriction on camera design.
Title: Re: Michael Reichmann, ETTR and Oly OMD EM5
Post by: richarddd on January 03, 2013, 04:11:07 pm
If it would be helpful, I'd be happy post a rawdigger analysis of an E-M5 RAW image of a gray card, or any other test I can easily perform.  If so, please be very specific as to what you'd like to see.
Title: Re: highlight blinkies are over-cautious for raw, especially with the E-M5
Post by: bjanes on January 03, 2013, 10:28:15 pm
P. S. A camera having an ISO Standard Output Sensitivity [SSOS] higher than its ISO Saturation-based Sensitivity [Ssat] is not "underexposing". The two numbers are simply different measurements of different output formats, with related but different purposes. Imatest calculates both SSOS and Ssat, and has a useful discussion at http://www.imatest.com/docs/sensitivity_ei/
The size of the gap is primarily a measure of raw highlight headroom and the Imatest page above gives this highlight headroom in its equation (8):
HdrSOS(f-stops) = log2( SSOS ⁄ Ssat ) + 0.5      (f-stops or EV or zones)

As I have mentioned in numerous times, Japanese camera makers are more or less required by their industry association CIPA to use ISO Standard Output Sensitivity as the primary method of calibration of the ISO sensitivity settings on their cameras. Insisting that SSOS = SSat would forbid a camera from offering more or less than 1/2 stop of highlight headroom in its raw files, which would be an absurdly rigid restriction on camera design.

The D800e uses REI (recommended exposure index) for its speed rating and I understand that this is also in compliance with the CIPA standards. The REI is arbitrary and the maker could assign whatever value they wish, so that standard is not lilmiting. In my own tests with the D800e exposing the target at ISO 100 according to the light meter reading gives a sensor saturation of 12.2% which is in accordance with the Ssat standard. If I interpret this correctly, Nikon is essentially using Ssat. What I don't understand is that DXO states that the Ssat speed for the D800e is 73 for the manufacturer's ISO of 100.

Regards,

Bill
Title: Re: Michael Reichmann, ETTR and Oly OMD EM5
Post by: Graystar on January 04, 2013, 10:23:22 am
If it would be helpful, I'd be happy post a rawdigger analysis of an E-M5 RAW image of a gray card, or any other test I can easily perform.  If so, please be very specific as to what you'd like to see.

Do you also have a Nikon or a Canon?  I'd be interested to see if there is a difference in spot meter exposure settings.  So if both cameras are set to ISO 200 and, say, f/8, and mounted with similar lenses, and you spot meter a neutral reference...is there a difference in the shutter speed?

I ask because I suspect that the EM-5 is overexposing by one stop, and that's why it's performance appears to be so good.  When I compare the camera settings of sample images taken at Imaging Resource or DPReview, it always seems like the EM-5 is taking in one stop more light at the same ISO under the same lighting.

In nearly all cases, when DxOMark indicates a lower ISO than the rating, it's because the signal is boosted to the correct level.  This isn't cheating because amplifying the signal adds noise.  It's nothing more than how the manufacturer has decided to process the signal.  However, in the case of Olympus's latest cameras...I think they're cheating.  They're ISO levels measure so low because they're actually changing the exposure from what it's supposed to be.  Basically, I think they're exposing ISO 200 shots as if they are ISO 100.  As far as I'm aware, DxOMark doesn't test for the correct aperture/shutter settings.

Well, that's my suspicion but I don't have an EM-5 to prove it.  And like I said...this is based on sample images (and the fact that DxOMark measure the ISO to be an entire stop off.)  I could just be imagining it all.

Olympus used to indicate how it rates ISO (they used SOS like Pentax) but they don't list the method on the new cameras.  Hmmm....
Title: Re: Michael Reichmann, ETTR and Oly OMD EM5
Post by: fike on January 04, 2013, 10:51:54 am
Hi Folks,
This has all been beaten to death at another photo (review) site's forums.  

YES, the OM-D is about 2/3rds of a stop less sensitive than a Canon.  
YES, the same shutter and aperture will yield a different ISO on an OM-D and many other cameras.

SO, comparing noise between an OM-D and, say for example, a 7D requires the 7D to be set to 1600 ISO while the OM-D is set to 2500 ISO.  

FINALLY, despite this peculiarity the OM-D still has less noise and more detail at these equivalent ISOs than any other MFT and many APS-C cameras.  Of course, it can't touch full frame sensors and some of the latest APS-C sensors are cleaner.  
Title: Re: Michael Reichmann, ETTR and Oly OMD EM5
Post by: Graystar on January 04, 2013, 11:41:23 am
This has all been beaten to death at another photo (review) site's forums.
Sorry, I did not realize that this was already discussed in other forums.  I do visit other forums, but I never visit the 4/3 areas, so I didn't know (it hasn't really bothered me enough to ask about it until richarddd offered tests.)

Quote
YES, the OM-D is about 2/3rds of a stop less sensitive than a Canon.
That's what DxOMark shows as well, which is in line with the difference in area.

Quote
FINALLY, despite this peculiarity the OM-D still has less noise and more detail at these equivalent ISOs than any other MFT and many APS-C cameras.
There, I beg to differ.  That's only at ISO 200 and because Olympus is exposing ISO 200 as if it's ISO 100.  Most APS-C cameras have a real ISO 100, and that's where they leave the OM-D and new Pens behind.

The comparison at equivalent aperture/shutter values is important because when you start shooting in low light, those are your limits.  People will buy an OM-D thinking they're going to get great performance in low light...but they're not because their ISO will be higher than a Canon or Nikon (for any given aperture/shutter combo.)  In the end, they're going to get 2/3rds stop more noise...not the APS-C noise level that they were led to believe the camera can deliver.  The deception of the buyers is what bothers me most about this...they're not getting what they think they're getting.
Title: Re: Michael Reichmann, ETTR and Oly OMD EM5
Post by: fike on January 04, 2013, 12:03:58 pm
...
...not the APS-C noise level that they were led to believe the camera can deliver.  The deception of the buyers is what bothers me most about this...they're not getting what they think they're getting.


I agree that their ISO ratings look deceptive. I will not comment about whether this was deliberate deception by Olympus or not.  BUT, despite that 2/3rds stop difference, the OM-D is better than my 7D in resolution and noise performance.  This is born out by DXO and many testers.  I believe that I said "MANY" other APS-C cameras.  The diversity of options in that space is pretty wide and I think it is enough to say that the OM-D MFT sensor is in the same quality range with the APS-C sensors even when you factor in an equivalent exposure that requires "different" ISO settings. Some APS-C are better some are worse. I don't really care about debating the last 3% of quality and I would estimate they all fall in that area.
Title: Re: Michael Reichmann, ETTR and Oly OMD EM5
Post by: Graystar on January 04, 2013, 12:21:02 pm
BUT, despite that 2/3rds stop difference, the OM-D is better than my 7D in resolution and noise performance.  This is born out by DXO and many testers.

That's true, but the 7D is over three years old now.  When comparing the OM-D to APS-C cameras with Sony sensors released around the same time, it rates over 2 dB less (a smidge over 2/3 EV more noise.)  This makes sense to me, as all the sensors are made by Sony and of similar tech, so you'd expect the difference to be based on size.  The bottom line is that there's nothing special about the OM-D sensor.  What's different is the exposure.  Canon can issue a firmware update for all their cameras changing the ISO 100 label to ISO 800...and now the OM-D performance will be abysmal in comparison!

That's definitely not what the CIPA had in mind when they issued the latest digital ISO standard.
Title: Re: Michael Reichmann, ETTR and Oly OMD EM5
Post by: richarddd on January 04, 2013, 05:38:47 pm
Do you also have a Nikon or a Canon?  I'd be interested to see if there is a difference in spot meter exposure settings.  So if both cameras are set to ISO 200 and, say, f/8, and mounted with similar lenses, and you spot meter a neutral reference...is there a difference in the shutter speed?

I ask because I suspect that the EM-5 is overexposing by one stop, and that's why it's performance appears to be so good.  When I compare the camera settings of sample images taken at Imaging Resource or DPReview, it always seems like the EM-5 is taking in one stop more light at the same ISO under the same lighting.

In nearly all cases, when DxOMark indicates a lower ISO than the rating, it's because the signal is boosted to the correct level.  This isn't cheating because amplifying the signal adds noise.  It's nothing more than how the manufacturer has decided to process the signal.  However, in the case of Olympus's latest cameras...I think they're cheating.  They're ISO levels measure so low because they're actually changing the exposure from what it's supposed to be.  Basically, I think they're exposing ISO 200 shots as if they are ISO 100.  As far as I'm aware, DxOMark doesn't test for the correct aperture/shutter settings.

Well, that's my suspicion but I don't have an EM-5 to prove it.  And like I said...this is based on sample images (and the fact that DxOMark measure the ISO to be an entire stop off.)  I could just be imagining it all.

Olympus used to indicate how it rates ISO (they used SOS like Pentax) but they don't list the method on the new cameras.  Hmmm....
I don't currently have a Nikon or Canon to compare.

The issue of the accuracy of Olympus's ISO rating has been debated extensively on another site.  For that matter, the meaning of ISO for RAW has been debated extensively.
Title: Re: Michael Reichmann, ETTR and Oly OMD EM5
Post by: allegretto on January 04, 2013, 11:04:11 pm
Just a subjective observation here, but a careful one if not quantitative.

Have used several systems this year (Nikon, Fuji, Sony and Leica) and just used an OM-D. Found, relatively that it tends to overexpose many scenes. Immediately de-comped  a third and probably needs one more click.

Which in a way I guess is saying that you can't cheat physics.
Title: Re: Michael Reichmann, ETTR and Oly OMD EM5
Post by: BJL on January 05, 2013, 07:59:02 pm
Just a subjective observation here, but a careful one if not quantitative.

Have used several systems this year (Nikon, Fuji, Sony and Leica) and just used an OM-D. Found, relatively that it tends to overexpose many scenes. Immediately de-comped  a third and probably needs one more click.
Do you mean in standard JPEG output? Because in raw files, the observation from DxO etc. is that it goes in the opposite direction, placing raw output at lower levels than other cameras.

Are there any measurements on the EM5 about:
(a) how the shutter speeds chosen by the in-camera metering with aperture priorty mode compare to other cameras? (In particular, does it choose slower shutter speeds than "advertised" by the ISO setting, which would require choosing a higher "ISO" setting to get comparable shutter speeds to other cameras in the same lighting conditions?)
(b) how the tonal level placement in default JPEG output looks when one chooses exposure setting based on external metering? (That is, does it give overexposed or underexposed JPEG output when used with an external meter?)

I ask because the DxO measurements of Ssat on raw files say nothing about the answers to these questions, and this discussion seems to confuse no less than three different issues by the persistent use of "ISO" as if it has a unique well defined meaning.
Title: Re: Michael Reichmann, ETTR and Oly OMD EM5
Post by: BJL on January 05, 2013, 08:20:16 pm
When comparing the OM-D to APS-C cameras with Sony sensors released around the same time, it rates over 2 dB less (a smidge over 2/3 EV more noise.)  This makes sense to me, as all the sensors are made by Sony and of similar tech, so you'd expect the difference to be based on size. ...
 
Canon can issue a firmware update for all their cameras changing the ISO 100 label to ISO 800...and now the OM-D performance will be abysmal in comparison!

That's definitely not what the CIPA had in mind when they issued the latest digital ISO standard.
I agree; that is why I am curious about comparisons at equal exposure level, such as same light source, same aperture ratio, shutter speeds of 100, 200, 400, etc. DxO does not do this, since its Ssat measure has nothing to do with how much exposure the sensor gets, but just about how much the photosite signals are subsequently amplified to produce raw output.

In this context, it is interesting to compare the DxO results for the three 2012 Olympus models, all using the same sensor AFAIK. The two newer "consumer" models, EPL5 and EPM2, score a bit better than the EM5 in the "sports" score, apparently because they have slightly "higher" curves for DR and SNR at 18%. But the dots on those curves for equal ISO settings are mostly the same or slightly lower for the new models, but also slightly to the right which moves the curves up. So the curves are only higher because the consumer models have slightly higher DxO Ssat scores, due to those two consumer models placing their raw output at slightly higher levels. This is purely a matter of how much analog (and/or digital) amplification is applied, and the slightly greater amplification used with the consumer models does not improve noise levels at all.
Title: Re: Michael Reichmann, ETTR and Oly OMD EM5
Post by: allegretto on January 07, 2013, 03:15:19 pm
thanks, no I mean RAW

will post examples when I figure out how you guys do the posting of such large files to this site...
Title: Oly OMD EM5 _overexposing_ raw?! Under what usage conditions?
Post by: BJL on January 07, 2013, 03:52:11 pm
thanks, no I mean RAW

will post examples when I figure out how you guys do the posting of such large files to this site...
As an EM5 user who has followed the debate over its "sensitivity", this is very intriguing!

Apart from trying to post examples, can you say how you are choosing exposure levels when this overexposure occurs? With in-camera metering in aperture priority mode? If so, with which metering mode? (I am trying to work out some weird abberations with its multi-pattern metering mode, especially with adaptor mounted SLR lenses, so center weighted has been far safer with those old lenses.)

Or are you using an external light meter set to the same ISO exposure index as the EM5’s ISO sensitivity setting?

Also, how are you converting from raw to the RGB files in which you see overexposure? (Or are you looking directly at level placement in the raw files?)
Title: Re: Michael Reichmann, ETTR and Oly OMD EM5
Post by: allegretto on January 08, 2013, 10:37:22 pm
all fair questions

just using center weighted and manual ISO settings

very busy week professionally (not a photographer), dont mean to be a dyk, will get the images
Title: Re: Michael Reichmann, ETTR and Oly OMD EM5
Post by: BJL on January 09, 2013, 09:52:32 am
Thanks.
    I look forward to your sample images, but no hurry ... after all, I am testing my newish EM5 myself, but appreciate additional info. from other users in this forum.

all fair questions

just using center weighted and manual ISO settings

very busy week professionally (not a photographer), dont mean to be a dyk, will get the images
Title: Re: Michael Reichmann, ETTR and Oly OMD EM5
Post by: Jack Hogan on January 10, 2013, 12:30:46 pm
What I'm asking (for the E-M5) is, assuming I've set optimum aperture and shutter speed, is it a good idea in terms of image quality to increase ISO to just below the clipping point (where clipping is determined by blinkies, as Michael stated), so long as you don't go above 1600?

This excellent chart by Bill Claff (http://home.comcast.net/~NikonD70/Charts/PDR_Shadow.htm#OM-D E-M5,EOS 5D Mark III) suggests that after ISO 400 you get very little IQ improvement raising the ISO on an EM5 - I would stop thereabouts because every time you raise the ISO by one stop you reduce highlight headroom by the same amount.  For instance if you are 'properly' ETTR'd at ISO1600, you only gain about 0.1 of a stop in noise performance compared to ISO 400 (I challenge you to see the difference :-), but you give up two stops of highlight headroom.  Of course in such a situation ISO 400 most likely entails correcting brightness in post.

Different story for Canons, as you can see...
Title: Re: Michael Reichmann, ETTR and Oly OMD EM5
Post by: allegretto on January 12, 2013, 04:07:25 pm
How does one post a true RAW file here? I have RX1 and OM-D shoot off here but don't know how to get the full sized files uploaded. Will only take up to 5Meg.

Thanks...
Title: Re: Michael Reichmann, ETTR and Oly OMD EM5
Post by: stamper on January 13, 2013, 06:28:40 am
I think you will have to us a site like this.

http://www.mediafire.com/about/

and then post a link that anyone can download. It has a free subscription.
Title: Re: Michael Reichmann, ETTR and Oly OMD EM5
Post by: gerafotografija on January 14, 2013, 01:24:42 am
This excellent chart by Bill Claff (http://home.comcast.net/~NikonD70/Charts/PDR_Shadow.htm#OM-D E-M5,EOS 5D Mark III) suggests that after ISO 400 you get very little IQ improvement raising the ISO on an EM5 - ...

Although I am generally scientifically inclined, I would have to brush up on light detector physics and engineering to add to the quantitative debate. After reviewing various sensor performance curves with the tool linked to above, it looks to me like the sensor generations seem to be progressing to a flatter curve, but it is not clear to me whether that represents an improvement or not.

However, having put a few thousand frames of captures through my OMD, maybe the following highly subjective and aesthetically directed comments may provide useful input? Someone else may be available to explain to me why I am doing these things, and whether there are better choices.

My first inclination was to use the OMD at base iso whenever possible, because the colors come out absolutely gorgeous and if you ettr and botch it up there is headroom to recover quite a bit in pp (Aperture 3 recovery in my case). You could argue that specific good color examples may have also been due to good lighting, or good lens drawing, or post processing for optimal curves, but I have personally never been able to "nail" the color tones and ranges so easily with any camera previously. Admittedly I gave it a shot with Kodachrome and Velvia back in the day.

That said, when I get to iso 1600 and beyond, I start to think about converting to monochrome because the colors start to go wrong. I can still get reasonable results at 1600 and with NR up to ~4000, but it no longer looks natural, and if there is skin involved, a careful B&W treatment with virtual red or yellow filtering added keeps the image reasonably flattering if not ideal.

Now that I have digested this thread a bit, my recent and completely unrelated switching to auto iso with a limit of 400 or 800 in less than perfect lighting conditions makes sense a bit. Keep in mind that I am mainly an enthusiastic amateur working with ambient lighting, and only rarely supplementing it with fill flash or other sources. The reason I switched to auto iso was to still get good colors across the DR (shadow through highlights), but in less than ideal conditions. I have never felt the need for NR up to 400 or so with this camera after getting a reasonable exposure, and usually a light touch of denoising at 800 with moderate at 1600, and then nothing but B&W along with heavier NR above that. This upper end is probably the range of conditions that moved me to switch over to and then push process Tri-X 10 to 20 years ago.

Assuming the sensor at different gains keeps the Rs Gs and Bs looking the way my eyes expect them to be as the overall luminance drops, I am with the group that proposes to push iso whenever possible to get the histogram to push all the way to the right. If my rods and cones start to diverge with respect to what they expect to be seeing on the screen or print, and I start to see bias or artifacts rather than natural (or at least pleasing or useful) renditions of colors and textures, that triggers a different approach to the image in pp and a slightly different plan during capture – I tend to focus my attention on mid-tones when exposing for monochrome.

I have no idea whether the base iso vs higher iso comparisons proposed take any of these aspects of IQ into account, but you can be sure I will be following along closely to see if I can learn a way to wring out another eV or so of useful range out of this sensor.
Title: Re: Michael Reichmann, ETTR and Oly OMD EM5
Post by: richarddd on January 14, 2013, 09:22:40 am
This excellent chart by Bill Claff (http://home.comcast.net/~NikonD70/Charts/PDR_Shadow.htm#OM-D E-M5,EOS 5D Mark III) suggests that after ISO 400 you get very little IQ improvement raising the ISO on an EM5 - I would stop thereabouts because every time you raise the ISO by one stop you reduce highlight headroom by the same amount.  For instance if you are 'properly' ETTR'd at ISO1600, you only gain about 0.1 of a stop in noise performance compared to ISO 400 (I challenge you to see the difference :-), but you give up two stops of highlight headroom.  Of course in such a situation ISO 400 most likely entails correcting brightness in post.
If you are below clipping, why do you need more highlight headroom?
Title: Re: Michael Reichmann, ETTR and Oly OMD EM5
Post by: richarddd on January 14, 2013, 09:29:29 am
I have no idea whether the base iso vs higher iso comparisons proposed take any of these aspects of IQ into account, but you can be sure I will be following along closely to see if I can learn a way to wring out another eV or so of useful range out of this sensor.
I worry that we focus too much on quantitative measures that don't capture all technical aspects of image quality.  To take a random example, it makes a difference whether or not noise has patterns.
Title: Re: Michael Reichmann, ETTR and Oly OMD EM5
Post by: Jack Hogan on January 14, 2013, 11:43:51 am
If you are below clipping, why do you need more highlight headroom?
Good question.  I would turn it around: how many times have we blown highlights that we wished to keep?  It's an almost free lunch: the extra DR costs virtually nothing - except perhaps a little EC in post :-)
Title: Re: Michael Reichmann, ETTR and Oly OMD EM5
Post by: Jack Hogan on January 14, 2013, 12:12:57 pm
Assuming the sensor at different gains keeps the Rs Gs and Bs looking the way my eyes expect them to be as the overall luminance drops...

I have no idea whether the base iso vs higher iso comparisons proposed take any of these aspects of IQ into account, but you can be sure I will be following along closely to see if I can learn a way to wring out another eV or so of useful range out of this sensor.

The answer depends on two separate components: the sensor, and the raw converter.  As far as the sensor goes, since it is based on Exmor technology which is quite linear with virtually no pattern noise, it should make no difference whatsoever to your rods and cones as long as the raw data is properly rendered - which brings us to the raw converter.   Most don't have a problem brightening up tones two or three stops.  If pushed hard, LR supposedly may exhibit issues because it 'Does the Twist' (http://dcptool.sourceforge.net/Hue%20Twists.html).  I don't think this should be an issue for the EM5, but perhaps someone will let us know?

Cheers Jack
Title: Re: Michael Reichmann, ETTR and Oly OMD EM5
Post by: richarddd on January 14, 2013, 01:00:12 pm
Good question.  I would turn it around: how many times have we blown highlights that we wished to keep?  It's an almost free lunch: the extra DR costs virtually nothing - except perhaps a little EC in post :-)
Sure, but my line in this thread is whether this is a productive technique:

1) Increase exposure until the first of (a) edge of clipping in highlights we wish to keep and (b) too slow shutter speed or too deep depth of field

2) If not yet at clipping, increase ISO until the first of (a) edge of clipping in highlights we wish to keep and (b) 800 or 1600

Practical application depends on being able to detect clipping, but the E-M5's orange highlight blinkies seem adequate (as MR says in the beginning of this thread).

In this technique, we will not blow highlights we wish to keep, by definition, so loss of highlight headroom would not be an objection.

If one is not prepared for a little adjustment in post, one shouldn't be shooting RAW :-)
Title: Re: Michael Reichmann, ETTR and Oly OMD EM5
Post by: Jack Hogan on January 14, 2013, 05:16:09 pm
2) If not yet at clipping, increase ISO until the first of (a) edge of clipping in highlights we wish to keep and (b) 800 or 1600

Right, great technique.  And my suggestion, after perusing Bill Claff's chart for the EM5, is however b) 400.

Cheers,
Jack
Title: Re: Michael Reichmann, ETTR and Oly OMD EM5
Post by: richarddd on January 15, 2013, 08:56:46 am
Right, great technique.  And my suggestion, after perusing Bill Claff's chart for the EM5, is however b) 400.
As I read the charts, by going from 400 to 800 or 1600, you improve noise at the cost of decreased dynamic range.  Decreased noise is a good thing, while decreased DR should not hurt if you are not in highlight clipping range.

Do you disagree with this analysis or is there another factor I'm ignoring?

BTW, increasing exposure (light on sensor) is always a good thing, so long as you're not clipping (or using too slow a shutter speed or too shallow DOF). Increasing ISO is more about trade-offs, primarily between noise and DR as I understand it.
Title: Re: Michael Reichmann, ETTR and Oly OMD EM5
Post by: Jack Hogan on January 15, 2013, 10:44:18 am
As I read the charts, by going from 400 to 800 or 1600, you improve [minimally] noise at the cost of  [1 or 2 stops of] ... dynamic range.  Decreased noise is a good thing, while decreased DR should not hurt if you are not in highlight clipping range.

Do you disagree with this analysis or is there another factor I'm ignoring?

That's basically it, and I agree with your statement as modified above.

Quote
BTW, increasing exposure (light on sensor) is always a good thing, so long as you're not clipping (or using too slow a shutter speed or too shallow DOF). Increasing ISO is more about trade-offs, primarily between noise and DR as I understand it.

Yes, but in the fine technique that you have outlined, exposure is always maximized according to artistic constraints first - and it doesn't change one bit (it stays maximized) when you raise the in-camera ISO setting.   The benefit in SNR obtained by raising ISO has nothing to do with exposure and all to do with the sensor electronics: increasing ISO in some cameras results in reduced sensor read noise.  If you look at the chart by Bill Claff you'll realize that for Canon sensors your statement just above is relatively accurate but for a so-called ISOless sensor such as the D7000's it is not because b) in your procedure above would be base ISO, so by raising ISO all you'd be doing is lowering the clipping point of the recorded data (while maintaining the exact same exposure) - no benefits whatsoever.

Back to your EM5, which is ISOless above 1200.  I can see a SNR benefit going from 200 to 400.  From 400 to 1200 you gain very little (1/6 of a stop of SNR - I challenge you to see it :-) while reducing the clipping point 1.5 stops: is it worth the risk?  Your camera, your scene, your call.  Past that, though, you only have something to lose by raising ISO.

Just out of curiosity, if you had a D7000, would you feel comfortable setting blur and dof (that is shutter speed and f/number, that is exposure) according to your procedure and artistic intent and leaving ISO at base forever, forgetting it even existed?  Your exposure would be the same than otherwise, your IQ would not suffer one bit :) and you'd have the best DR in town.  Nevertheless some people just do not feel comfortable getting properly exposed, dark images OOC.

Jack
Title: Re: Michael Reichmann, ETTR and Oly OMD EM5
Post by: richarddd on January 15, 2013, 03:50:15 pm
I agree that the benefits of pushing ISO decrease rapidly above 400 on the E-M5.

I don't have a D7000 or other DSLR.  From what I've read, leaving ISO at base and adjusting in post works well with modern Nikon DSLRs, other than that it makes it hard to review pictures in camera, as the image on the LCD is often too dark.

Title: Re: Michael Reichmann, ETTR and Oly OMD EM5
Post by: digitalphotographer on September 11, 2013, 11:37:12 pm
This excellent chart by Bill Claff (http://home.comcast.net/~NikonD70/Charts/PDR_Shadow.htm#OM-D E-M5,EOS 5D Mark III) suggests that after ISO 400 you get very little IQ improvement raising the ISO on an EM5 - I would stop thereabouts because every time you raise the ISO by one stop you reduce highlight headroom by the same amount. 

Thanks for the suggestion for the EM5! In this case, it seems limiting Auto-ISO to ISO400 and doing manual "push" or EV adjustments in LR/ACR will be a better option to have more highlights headroom and controls, am I right?

Just want to make sure I understand this, when the ISO is top-limited at 400, I will still get benefits from ETTR, is this correct?
Title: Re: Michael Reichmann, ETTR and Oly OMD EM5
Post by: bjanes on September 12, 2013, 07:26:03 am
I agree that the benefits of pushing ISO decrease rapidly above 400 on the E-M5.

I don't have a D7000 or other DSLR.  From what I've read, leaving ISO at base and adjusting in post works well with modern Nikon DSLRs, other than that it makes it hard to review pictures in camera, as the image on the LCD is often too dark

Contrary to popular belief, the Nikon D800 is not really ISOless and some improvement in shadow performance can be obtained by increasing the ISO, but the effect is relatively small, less than 1/2 stop as shown by Bill Claff's previously mentioned graph.

A basic understanding of the sources of noise in digital captures is helpful and Emil's post (http://theory.uchicago.edu/~ejm/pix/20d/tests/noise/index.html) is an excellent source. The two main sources of noise are shot noise (which is predominant in the highlights) and read noise (which predominates in the deep shadows). We are not interested in noise per se, but in the signal to noise ratio. Shot noise is maximal in the highlights, but the SNR increases as the square root of exposure.

Read noise predominates in the deep shadows. In cameras where the electronics (primarily the analog to digital converter) is not well matched to the full well capacity of the sensor (such as with most Canons), read noise can be reduced by increasing the ISO gain. If f/stop and shutter speed considerations prevent attaining full well, one can increase the ISO gain with the Canon to get a better DR. However, head room is decreased and one can clip the highlights if the ETTR is overdone. With the D800e, the DR improvement is less than 1/2 stop with ETTR under these conditions, but the risk of highlight clipping remains.

Thermal noise is important in astrophotography, but is minimal  with exposures of less than a second, so it is not important to cool the sensor with ordinary terrestrial photography.

The presumed increase in tonality towards the right end of the histogram where more levels are present is largely a red herring, since these potential levels are dithered by noise and effectively obliterated, as explained by Emil in some detail.

Bill



Title: Re: Michael Reichmann, ETTR and Oly OMD EM5
Post by: Vladimirovich on September 12, 2013, 10:44:09 am
Thanks for the suggestion for the EM5! In this case, it seems limiting Auto-ISO to ISO400 and doing manual "push" or EV adjustments in LR/ACR will be a better option to have more highlights headroom and controls, am I right?
you are not getting more "highlights headroom" with a higher gain... higher gain makes sense if you want to get better SNR (Signal/Noise) in (deep) shadows if the scene's total DR and your intent in terms of what you can tolerate to be clipped in raw will allow you to increase the gain...
Title: Re: Michael Reichmann, ETTR and Oly OMD EM5
Post by: digitalphotographer on September 12, 2013, 03:12:21 pm
you are not getting more "highlights headroom" with a higher gain... higher gain makes sense if you want to get better SNR (Signal/Noise) in (deep) shadows if the scene's total DR and your intent in terms of what you can tolerate to be clipped in raw will allow you to increase the gain...
Thanks for the clarification. What I meant to say was, when shooting EM5 at max ISO 400 and using ETTR, I will have more highlight headroom than shooting at ISO800 or ISO1600, with the benefit of reducing shadow noise on the RAW files. Is this understanding correct?
Title: Re: Michael Reichmann, ETTR and Oly OMD EM5
Post by: Vladimirovich on September 12, 2013, 05:24:34 pm
Thanks for the clarification. What I meant to say was, when shooting EM5 at max ISO 400 and using ETTR, I will have more highlight headroom than shooting at ISO800 or ISO1600, with the benefit of reducing shadow noise on the RAW files. Is this understanding correct?

yes... but practically w/ Sony sensor in EM5 (beyond the tests, which are correct indeed) I doubt that you need to bother w/ ISO400 - just stay at lowest ISO and use maximum exposure (time and aperture) possible in your situation and that 's it... real life scene w/ real life converters - you will not notice that gain in shadows.
Title: Re: Michael Reichmann, ETTR and Oly OMD EM5
Post by: digitalphotographer on September 12, 2013, 05:49:02 pm
yes... but practically w/ Sony sensor in EM5 (beyond the tests, which are correct indeed) I doubt that you need to bother w/ ISO400 - just stay at lowest ISO and use maximum exposure (time and aperture) possible in your situation and that 's it... real life scene w/ real life converters - you will not notice that gain in shadows.

I know this is a thread about EM5, but recently I also acquired an RX1R as a second camera, would you also suggest shooting at base ISO for the 24MB FF too? Or is RX1R resembling the D800 more, ie going "ISOless" at higher range?
Title: Re: Michael Reichmann, ETTR and Oly OMD EM5
Post by: Vladimirovich on September 12, 2013, 06:08:52 pm
I know this is a thread about EM5, but recently I also acquired an RX1R as a second camera, would you also suggest shooting at base ISO for the 24MB FF too? Or is RX1R resembling the D800 more, ie going "ISOless" at higher range?
see the graphs of DxOMark tests... like w/ EM5 the tests will suggest to go one stop higher (if you do not risk to clip anything important to you)... will you see the end result (which is indeed there, test numbers wise) ? try yourself
Title: Re: highlight blinkies are over-cautious for raw, especially with the E-M5
Post by: BJL on March 19, 2018, 02:36:42 pm
Sorry to dig up this old discussion, but the topic is active again in other threads, so maybe this one is worth bringing to peope's attention.

As I have mentioned in numerous times, Japanese camera makers are more or less required by their industry association CIPA to use ISO Standard Output Sensitivity as the primary method of calibration of the ISO sensitivity settings on their cameras.
To which Bill Janes replied
The D800e uses REI (recommended exposure index) for its speed rating and I understand that this is also in compliance with the CIPA standards. The REI is arbitrary and the maker could assign whatever value they wish, so that standard is not lilmiting. In my own tests with the D800e exposing the target at ISO 100 according to the light meter reading gives a sensor saturation of 12.2% which is in accordance with the Ssat standard.
My original statement needs some refinement. As far as I can tell, the flexibility of "REI" is there simply to allow flexibility in multi-zone light metering algorithms; insisting on SSOS would restrict cameras to some standard averaged brightness metering. In practice it seems that basic metering modes do more or less follow SSOS. In fact, if a camera were to deviate significantly form this, its handling of simple low contrast scenes would produce out-of-camera JPEGs significantly lighter or darker than from other cameras, and customers would notice, and complain. In other words, for basic metering, or metering on a subject of fairly uniform brightness (as with spot metering), market forces keep in-camera metering close to SSOS, with placement of metered mid-tones at a JPEG level near 112, which is just under 18%.

Also, allowing for the highlight roll-off typically used in default JPEG conversion tone curves, it makes sense that mid-tones are placed a bit below 18% of FWC (or maximum raw level): a 1/2 stop of roll-off requires placing them at about 12.5%, close to what you observed with the D800e. This is also what the ISO 12232(2006) standard hints at when it defines SSat: it mentions adding 1/2 stop of headroom.