[dosdan]

100 -1600, i.e 5 stops

No, it's 4 stops.

[Class A]

No, it's 4 stops.

Agreed.

Not much I can agree with in Smeggypant's post, but let's leave it at "agreeing to disagree".

[Smeggypants]

No, it's 4 stops.

yes my bad - sorry. Yes it's 4 stops

[Smeggypants]

Agreed.

Not much I can agree with in Smeggypant's post, but let's leave it at "agreeing to disagree".

Well both theory and practise shows a reduction in dynamic range by using base ISO and underexposing, rather than setting an ISO that gives a proper exposure.

You disagree? Ok fair enough. can you explain how you've got a different practical result from your K-5?

---

Last edited by Smeggypants; 03-23-2011 at 07:39 PM.

[falconeye]

Interesting this thread is still alive. Let me avoid to comment on any particular argument though

The thing which makes me reply is a notion of theory vs. practice.

So, this is what I do when shooting with my K-5 and how the K-5 changed the way I use the dials:

Say, I normally would shoot with Auto-ISO 80-1600. Then, in many instances, I find myself setting the K-5 to Auto-ISO 80-400 and EV-2 exp.compensation at the same time. I do this whenever there is a chance for blown highlights like sky, sunsets, windows etc. I don't do it always because I sometimes want to use the full DR of ISO 80. YMMV

This isn't ISOless already. But it's a better approximation than with previous cameras.

[Smeggypants]

Well both theory and practise shows a reduction in dynamic range by using base ISO and underexposing, rather than setting an ISO that gives a proper exposure.

You disagree? Ok fair enough. can you explain how you've got a different practical result from your K-5?

Still waiting for your findings here Class A

[Smeggypants]

Still waiting for your findings here Class A

......

[Smeggypants]

Bumping for unaddressed answers ....

[Class A]

Well both theory and practise shows a reduction in dynamic range by using base ISO and underexposing, rather than setting an ISO that gives a proper exposure.

What "theory " predicts a reduction in dynamic range?
If you don't "crush" shadows (is "crushing" more acceptable to you than "clipping" in this context? Note that you can "lose" shadow details not only due to the reasons you mentioned but also because you are not giving the few photons enough of a chance to be registered) then both shooting alternatives can capture the same DR. Shooting at higher ISO settings even bears the danger of losing DR by highlight clipping.

In "practise " all sorts of things can happen, depending on what one does.

The thing is: After ISO 1600, your K-5 does behave like an ISOless camera. It just underexposes and pushes in post. Between ISO 80-1600 there is some analogue amplification going on but because the K-5's excellent SONY sensor has such low noise, that analogue amplification is almost not needed anymore. By all means, use ISO settings higher than ISO 80/100 (I would too) but don't think that it makes a gigantic difference to underexposing and pushing in post.

Of course, here "underexposing" means to "reduce the shutter speed by the same factor a corresponding ISO setting would have". If you go beyond that, of course it will typically show in increased shadow noise in comparison.

Ok fair enough. can you explain how you've got a different practical result from your K-5?

What makes you think that I've got a K-5? Or are you saying that I'm not qualified to engage in discussions about "ISOless sensors" because I don't own a K-5?

[philbaum]

.....
So, this is what I do when shooting with my K-5 and how the K-5 changed the way I use the dials:

Say, I normally would shoot with Auto-ISO 80-1600. Then, in many instances, I find myself setting the K-5 to Auto-ISO 80-400 and EV-2 exp.compensation at the same time. I do this whenever there is a chance for blown highlights like sky, sunsets, windows etc. I don't do it always because I sometimes want to use the full DR of ISO 80. YMMV

This isn't ISOless already. But it's a better approximation than with previous cameras.

Guys,
I don't want to beat a dead horse, but i want to understand the issue as understanding is key to getting the most of this or any camera. I've been reading this thread over and over.

Finally, i think i can understand the issue based on the histogram. Lets say for example that we have an image that encompasses a total of 10 ev. This total DR of 10 ev pixels are nicely distributed over the 256 tonal values of the histogram (because of my amazing skills no doubt ) as i took this first picture, call it SCENE A.

In the second picture I decide to use a new strategy, and badly underexpose the picture such that only 10% of the photosites have enough photons to register a tonal value. The good news is that none of the highlights have been clipped. Call this SCENE B. I try to correct the image in pp by using the exposure and contrast sliders. Yet despite my best skills, the developed image doesn't look right.

I think the problem with Scene B is that it was exposed in a manner that only allowed what would have been the brightest tones to even register. Although one can apply more gain using pp software, there will never be the richness (multiplicity of 256 tones) or the range of sampling of the real image. Dpreview has discussion of the histogram which describes a tone-poor image as having "combing".

Histogram: Digital Imaging: Glossary: Learn: Digital Photography Review

Obviously, scene B was an extreme example, and the tactic of falconeye to use -2EV to preserve the highlights, will likely have much more minor repercussions in the loss of some tonal values. Instead there will be advantages in 2 ways, preserving highlights and taking advantage of a larger DR because of a lower ISO.

Having some understanding of what's happening in the camera allows one to tailor the strategy to the image. If the image is a moody dark indoor image, then the clipping of highlights is of less concern than preserving a rich number of tones in the shadows. So one might want to put a +2ev onto a 400 iso setting. Is that conclusion right?

[dosdan]

While I've been concentrating on the Pentax K-5, let's look at the Nikon D7000 too. Compare the read noise characteristics of the K-5 & the D7000 between base ISO & ISO1600.

Sensorgen.info data for Pentax K-5
Sensorgen.info data for Nikon D7000

K-5

ISO	Actual ISO	Equiv. read noise in electrons
80	70	3.3
100	91	3.5
200	183	3.1
400	363	2.6
800	717	2.4
1600	1417	1.9

D7000

ISO	Actual ISO	Equiv. read noise in electrons
100	83	3.1
200	167	3
400	328	3
800	658	3
1600	1269	2.9

You can see a difference in noise performance, with the Pentax apparently having a little higher read noise at base ISO, but less (-0.6 stops) noise at ISO1600. This difference suggests that the D7000 is an actual "ISOless" implementation, i.e. no change in analogue gain used when altering the ISO sensitivity, just digital gain/multiplication, so the read noise is constant, while the K-5 does makes use of analogue gain change to achieve a slight improvement in high ISO noise performance.

Dan.

[glanglois]

Well, that would put an end to one question, at least it was still open in my mind. There is a significant physical difference between the Sony sensors in the two cameras. It's not just a matter of processing somewhere after the A/D conversion.

I suppose that everyone else figured that out while I was napping ??

[dosdan]

There is a significant physical difference between the Sony sensors in the two cameras.

I think it's the same sensor; just a matter of programming. The programmable gain amp on the sensor chip, that Falconeye mentioned here, is either not switched on in the D7000, or if it is, its gain is fixed, rather than being varied to alter ISO sensitivity.

Dan.

[Smeggypants]

While I've been concentrating on the Pentax K-5, let's look at the Nikon D7000 too. Compare the read noise characteristics of the K-5 & the D7000 between base ISO & ISO1600.

Sensorgen.info data for Pentax K-5
Sensorgen.info data for Nikon D7000

K-5

ISO	Actual ISO	Equiv. read noise in electrons
80	70	3.3
100	91	3.5
200	183	3.1
400	363	2.6
800	717	2.4
1600	1417	1.9

D7000

ISO	Actual ISO	Equiv. read noise in electrons
100	83	3.1
200	167	3
400	328	3
800	658	3
1600	1269	2.9

You can see a difference in noise performance, with the Pentax apparently having a little higher read noise at base ISO, but less (-0.6 stops) noise at ISO1600. This difference suggests that the D7000 is an actual "ISOless" implementation, i.e. no change in analogue gain used when altering the ISO sensitivity, just digital gain/multiplication, so the read noise is constant, while the K-5 does makes use of analogue gain change to achieve a slight improvement in high ISO noise performance.

Dan.

So you're equating analogue gain with 'ISO' and making an assumptice leap on the 7000 becuase of it's slightly different read noise figures?

And there's me thinking ISOLess, was a camera without an ISO parameter?

Those data sheets also show the K5 and D7000 sensor as havign different pixel sizes. I thought it was generally accepted that the sensor was the same but with additional proprietary addons?

[Smeggypants]

I think it's the same sensor; just a matter of programming. The programmable gain amp on the sensor chip, that Falconeye mentioned here, is either not switched on in the D7000, or if it is, its gain is fixed, rather than being varied to alter ISO sensitivity.

Dan.

How do you know the analogue gain on the D7000s sensor is fixed?