Dynamic Range measurement

Dynamic range is very important to me.

I have just read a test report on the new Canon 70D which claims that the DR tests out at 11.53 which according to Amateur Photographer is less than Nikon currently offer.

How is this tested? Anyone know? Is it possible to achieve in real life or is it another of these fictional figures (like the EU mpg figure quoted in car adverts here in the UK which everyone accepts is impossible to achieve but nonetheless is a legal requirement)?

Different sites test it in different ways. Shooting a Stouffer step card is one common method. As to the DR of the 70D, it's pretty on par with all other Canon cameras and all are less than what Nikon currently offers.

As to whether or not you can get this is real life photos, a lot depends on the conditions.

Some protocols here: DXO dynamic range protocols and browse a bit for more explanations

Sony sensors, Pentax have better DR, but at higher ISO there is less difference, or different results I have heard?

Lowner wrote in post #16318916
Is it possible to achieve in real life or is it another of these fictional figures?

Well, it really depends on when, where and what you shoot, and afterwards it depends on how much post processing you do, as well as the extent to which you perform highlight and/or shadow recovery.

Yea, if you really need may DR, right now you need to get a D800. Anything else probably isn't going to make that much difference.

Depending on your application, Magic Lantern (available for 5DII and III, 7D, some rebels and **D) has a feature called DualISO. I've used this and while not 'perfect' it's gotten to the stage where it really is pretty darn good. I would even use it for landscapes printing up to at least 19".

Bust as I said, if you need max DR for detailed landscapes to print large, you best DSLR bet is a D800.

good information in this thread so far.

another thing to keep in mind is that any DR advantage Nikon has will disappear at or just before ISO 800.

hes gone wrote in post #16321681
=he's gone;16321681]good information in this thread so far.

another thing to keep in mind is that any DR advantage Nikon has will disappear at or just before ISO 800.

Actually, it's closer to ISO 3200 and even when noise is present, it's much more random and cleans up easier than with any of the Canon sensors.

hes gone wrote in post #16321681
=he's gone;16321681]good information in this thread so far.

another thing to keep in mind is that any DR advantage Nikon has will disappear at or just before ISO 800.

To be specific, you only get roughly 14 stops of DR at ISO100, and then it goes down a stop for each next ISO value:
ISO100 - 14ev
ISO200 - 13ev
ISO400 - 12ev
ISO800 - 11ev
ISO1600 - 10ev
ISO3200 - 9ev

With the new Canon cameras, everything from ISO100-800 looks the same, and then begins to slope down in DR, which gives it an edge at high ISOs:
ISO100 - 11.5ev
ISO200 - 11.5ev
ISO400 - 11.5ev
ISO800 - 11.5ev
ISO1600 - 11ev
ISO3200 - 10.7ev

gjl711 wrote in post #16321751
Actually, it's closer to ISO 3200 and even when noise is present, it's much more random and cleans up easier than with any of the Canon sensors.

In using a 1Dx on a few occasions, there just isn't any visible noise up to ISO3200, I didn't even feel the need to use NR. In practical terms, you can take all of these figures and apply a huge fudge factor to them regarding how it will effect your work.

Edit: You know what's better than DR or noise performance? Ergonomics. I hate how Nikons handle, and to me the D800 is simply a great sensor-in-a-box. Eventually, all such discussion is fueled by trying to figure out which camera is better, or which would one you'd rather buy, and you should do so based on the collective properties of a camera and not just one point of comparison. You can have a mind blowing sensor in the worst camera.

Many thanks to you all. I need to reread and digest what's been written.

The top end of the dynamic range is fairly easy to pin down - it's when the sensor signal is maximum, the sensel signal does not increase any more as the light intensity is increased (the well is full).

However the bottom of the range requires an arbitrary criterion - how low a gray level can be recorded that is definitely a "signal' or distinguishable image feature and is definitely not "noise". This why a particular "signal to noise ratio" is selected as the bottom of the range - in different fields different values are selected 2:1, 3:1 4:1 etc. When measuring the S/N of one sensel, that selection is straightforward. Unfortunately for the whole sensor area, the selection is not so straightforward and may not take into account any "pattern" that might be present in the noise (banding etc) that would be missed by a simple "average noise per pixel across the frame or even just part of it" approach. In addition, the "noise" can arise from various sources and their proportions will vary with ISO. Sometimes the "banding pattern" is clearly visible on top of the image, while at higher ISOs the electronic noise can overwhelm it and its no longer visible. Complex question.....

Kolor-Pikker wrote in post #16321764
To be specific, you only get roughly 14 stops of DR at ISO100, and then it goes down a stop for each next ISO value:
ISO100 - 14ev
ISO200 - 13ev
ISO400 - 12ev
ISO800 - 11ev
ISO1600 - 10ev
ISO3200 - 9ev

With the new Canon cameras, everything from ISO100-800 looks the same, and then begins to slope down in DR, which gives it an edge at high ISOs:
ISO100 - 11.5ev
ISO200 - 11.5ev
ISO400 - 11.5ev
ISO800 - 11.5ev
ISO1600 - 11ev
ISO3200 - 10.7ev

In using a 1Dx on a few occasions, there just isn't any visible noise up to ISO3200, I didn't even feel the need to use NR. In practical terms, you can take all of these figures and apply a huge fudge factor to them regarding how it will effect your work.

Edit: You know what's better than DR or noise performance? Ergonomics. I hate how Nikons handle, and to me the D800 is simply a great sensor-in-a-box. Eventually, all such discussion is fueled by trying to figure out which camera is better, or which would one you'd rather buy, and you should do so based on the collective properties of a camera and not just one point of comparison. You can have a mind blowing sensor in the worst camera.

Good to know that 6D can be used upto iso 800 with no reduction in DR. Implication is that I may bump up iso more often. Is my understanding ok.

ssmanak wrote in post #16330743
Good to know that 6D can be used upto iso 800 with no reduction in DR. Implication is that I may bump up iso more often. Is my understanding ok.

From a DR perspective, it seems so. But noise is a different kettle-of-fish.

ssmanak wrote in post #16330743
Good to know that 6D can be used upto iso 800 with no reduction in DR. Implication is that I may bump up iso more often. Is my understanding ok.

There is a drop, it's just not significant. The drop looks to be about 2/3 of a stop. By 1600 it's a full stop and by 3200 it's two stops.

Lowner wrote in post #16330851
From a DR perspective, it seems so. But noise is a different kettle-of-fish.

Dynamic range and noise are inextricably linked. I have yet to find how DxO deal with patterned noise (low amplitude but visually noticeable patterns, banding etc), as distinct from random noise (same amplitude but not visually noticeable patterns of "typical" graininess) in their evaluations.

For example DxO uses (the arbitrarily selected criterion of) a signal to noise ratio of only 1 to determine the dynamic range values they report

ssmanak wrote in post #16330743
Good to know that 6D can be used upto iso 800 with no reduction in DR. Implication is that I may bump up iso more often. Is my understanding ok.

I have to admit, I still don't understand what Canon does in the electronics that does this. I guess it means that to have low read noise at ISO800 they have to have much higher read noise at ISO100.

From a sensor perspective it seems crazy that if I only fill the photosites 1/8th of what they can handle, that I get about the same DR as if I fill them up with 8x the light.