DR - how measured and what is it on 1Dx, 5DM3, ...

[RGF]

How is dynamic range measured. I keep hear that such a camera has 12/15 stops of DR.

How exactly is DR measured. From my limited understanding of electronics, if you have A/D converter that is 12 steps the output will be between 0 and 2^12-1 (4095). If you allow the darkest step to have values from 0 - 15 (4 bits of data) then you are left with 8 stops of DR.

Does this make sense? If not, what is the correct way to think about DR.

Thanks

 

[3kramd5]

DR is between the highest and lowest values,

A 12-bit system would provide 12 stops at base amplification if free from noise: 2^0 to 2^1, 2^1 to 2^2,... 2^11 to 2^12.

So in the real camera world, DR in is between saturation (highest possible value) and the noise floor (lowest meaningful value).

Since saturation decreases with an increase in ISO, so does DR.

For those two cameras:

http://www.sensorgen.info/CanonEOS-1DX.html
http://www.sensorgen.info/CanonEOS-5D-Mark-III.html

 

[Marsu42]

A 12-bit system would provide 12 stops at base amplification if free from noise

I admit I never quite understood this - what exactly is the definition of "free from noise"? And if this includes banding, the 5d2 (11.0ev@iso100) or 60d (10.8ev@iso100) would have way less dr than the newer 6d (11.5ev@iso100), at least from what I've seen.

 

[Loswr]

...what is the correct way to think about DR.

The correct way to think about DR is that Canon doesn't have enough, and Sony and Nikon have plenty. If anyone tries to suggest that the additional DR of Sony and Nikon isn't of all that much practical benefit in most situations, then the correct way to think about DR is that more DR means more stops of shadow lifting, and if you cannot lift the shadows at least 4 stops then your camera is unusable.

So speaketh the disciples of the Holy Exmor.

 

[RGF]

...what is the correct way to think about DR.

The correct way to think about DR is that Canon doesn't have enough, and Sony and Nikon have plenty. If anyone tries to suggest that the additional DR of Sony and Nikon isn't of all that much practical benefit in most situations, then the correct way to think about DR is that more DR means more stops of shadow lifting, and if you cannot lift the shadows at least 4 stops then your camera is unusable.

So speaketh the disciples of the Holy Exmor.

Oh Great and Wise Holy Exmor

Thanks. I do understand the practical meaning of DR, just wondering how low a signal you can amplify before you are adding noise to noise.

Is there a standard way to determine DR that is consistent across testers, cameras, ...?

can test DR myself? If so, how?

Thanks

 

[RGF]

DR is between the highest and lowest values,

A 12-bit system would provide 12 stops at base amplification if free from noise: 2^0 to 2^1, 2^1 to 2^2,... 2^11 to 2^12.

So in the real camera world, DR in is between saturation (highest possible value) and the noise floor (lowest meaningful value).

Since saturation decreases with an increase in ISO, so does DR.

For those two cameras:

http://www.sensorgen.info/CanonEOS-1DX.html
http://www.sensorgen.info/CanonEOS-5D-Mark-III.html

If these results translate into real results in the field, then the 5D M3 has a much maligned reputation.

At ISO 100, DR is 11.0, for the Nikon D800 at ISO 100 the DR is 13.4

At ISO 800 5DM3 is 10.5 vs 11.2 for D800

At ISO 3200 5DM3 is 9.5 and D800 is 9.2

The D810 increases the DR at ISO 100 by 0.4.

So if the 5DM4 increases it DR by 2-3 stops, it will be on par with the D800/810 (though probably at a lower resolution).

 

[Loswr]

Simply put, read noise can be measured, that's basically the noise floor. Full well pixel capacity can be measured, that's the max charge a pixel can hold. Dynamic range is log2 (full well capacity / read noise).

You can estimate it with a step wedge such as the Stouffer T4110 (and a proper light box, usually using a RAW analysis tool (e.g. rawnalyze) vs. a standard RAW converter.

 

[K]

...what is the correct way to think about DR.

The correct way to think about DR is that Canon doesn't have enough, and Sony and Nikon have plenty. If anyone tries to suggest that the additional DR of Sony and Nikon isn't of all that much practical benefit in most situations, then the correct way to think about DR is that more DR means more stops of shadow lifting, and if you cannot lift the shadows at least 4 stops then your camera is unusable.

So speaketh the disciples of the Holy Exmor.

+1


;D

 

[Lee Jay]

A 12-bit system would provide 12 stops at base amplification if free from noise

I admit I never quite understood this - what exactly is the definition of "free from noise"?

Free from noise in this context would mean that the dominant noise source is quantization. This is essentially never the case with cameras.

 

[Lee Jay]

Dynamic range, in the camera sense, is the ratio between the full-well capacity of the pixels and the noise floor of the system (the noise remaining in a dark frame). There are several sources of noise in a frame not exposed to light and those combine to make the noise floor.

The problem is, this is a lousy definition photographically, even if it is a good definition electronically.

Most people won't consider the noise floor to show any significant detail in their images. In fact, some people can pick out details below the noise floor in some scenes while others won't tolerate any signal to noise ratios below something like 4-8 (which is several stops above the noise floor).

In practice, it's only rarely that any scene requires more than about 9 stops of real, capturable dynamic range to be rendered in a realistic looking way on a regular print or electronic display. Some scenes can require more, however, especially if you don't want it to look natural. For example, the Andromeda galaxy is usually rendered in a highly compressed way with 20 or more stops of DR smashed down into 8. It doesn't look realistic at all compared to how it looks through a telescope (the core is actually vastly brighter than the arms) but it still makes for nice looking images. However, this DR is captured with many frames instead of a single one. Using that approach, DR is essentially only limited by how many frames are captured.

 

[DJD]

IMHO, the best explanation about DR, ISO, and just about every other way to measure a digital camera's sensor performance can be found here.

http://www.clarkvision.com/articles/index.html

What you do with that information is up to you.

Cheers,
DJD

 

[Orangutan]

...what is the correct way to think about DR.

The correct way to think about DR is that Canon doesn't have enough, and Sony and Nikon have plenty. If anyone tries to suggest that the additional DR of Sony and Nikon isn't of all that much practical benefit in most situations, then the correct way to think about DR is that more DR means more stops of shadow lifting, and if you cannot lift the shadows at least 4 stops then your camera is unusable.

So speaketh the disciples of the Holy Exmor.

You live in Boston, yes? Perhaps you could have held your fire until you saw the whites of their eyes.

 

[jarrodeu]

And here I though Neuro was going to say that the Canon cameras are overkill and actually have too much DR. That's why he had his cameras modified to only have 6 stops because anything more is a waste. ;D

Jarrod

 

[Loswr]

...what is the correct way to think about DR.

The correct way to think about DR is that Canon doesn't have enough, and Sony and Nikon have plenty. If anyone tries to suggest that the additional DR of Sony and Nikon isn't of all that much practical benefit in most situations, then the correct way to think about DR is that more DR means more stops of shadow lifting, and if you cannot lift the shadows at least 4 stops then your camera is unusable.

So speaketh the disciples of the Holy Exmor.

You live in Boston, yes? Perhaps you could have held your fire until you saw the whites of their eyes.

That's so old school – today, we carry the fight to foreign shores. :-X