5D vs. D60 (or 10D) dynamic range question

Hi, all.

 

Has anyone out there tested the low iso (100 or 200) RAW photo dynamic range of the 5D

against the D60 (or perhaps the 10D)?

 

Is there any quatifiable improvement in dynamic range between these cameras?

 

Thanks for any information on this topic.

If I rely on <B><I>what I have read</i></b> the 5D has the better dynamic range of the group. It beats my 20D and I find it's no real slouch.

this might be helpful http://www.dpreview.com/reviews/canoneos5d/page23.asp

I didn't see any significant differences between the 5D and 20D when I tested them. I suppose there might be some technically measurable difference if you did the right tests with the right equipment, but in practice it's not enough to bother about.

 

http://www.photo.net/equipment/canon/5D/

Although I dont see the relevance of this question, being the huge price difference and the 1.6x vs FF Chip formats would be the main deciding factors between the two, I have read many many posts about this very same question, and there are scores that argue that the 5D has a bigger range, and scores others who insists there isnt a difference. Noise on the other hand the 5D is better.

As we all know the photo receptors give 14 bit values. In a perfect world this would be 14 stops. But lots of real world issues NOISE being the most significant eat / obscure the information at the lower end.

 

SO the less noise the more the usable info EQUALS more dynamic range.

 

-- Roger

Thank you for the user comments and pointers so far.

 

In response to the relevance question, I have read numerous messages where folks claim

better dynamic range from whichever DSLR. I have had numerous RAW photos blow out in

high-key lighting (bright sunlight through a window illuminating part of an otherwise dark

room, for instance) that my B&W film handles with relative ease. With my D60, exposing

for highlights gives empty shadows, and exposing for the shadows gives empty highlights.

(Yes, I know about combining highlight and shadow images for increased dynamic range.)

 

So I'd like to know if anyone has figured out if dynamic range really is improving, or if

that's all a bunch of hooey and I should figure it hasn't changed when considering the 5D.

 

Thanks again for the information and opinions!

It's mostly a bunch of hooey.

 

People's subjective opionions are all over the map. Actual tests of Dynamic range don't show much difference and you probably wouldn't expectthem to as they are pretty much fixed by the design of the sensors.

 

Come up with a new sensor design like the Fuji's where you have large and small pixels with different sensitivities and you may increase dynamic range, but you might lose out on other things like noise or resolution - or you might not. Given a regular array of identical pixels which have a linear response to light levels, you're not going to see a lot of difference in DR between cameras, assuming the makers get the electronics right.

"As we all know the photo receptors give 14 bit values. In a perfect world this would be 14 stops"

 

Well, yes, sort of, maybe...

 

The last "stop" of information would have 2 levels, black and white. The next to the last stop would have 4 levels, black, dark grey, light grey and white. These really would not be very usable, even if there was no noise at all (which is impossible). I don't know how many levels of grey you'd need for a "stop" of dynamic range to be usable in a photographic sense, but it's generally more than 4!

The shallowest bits are still useful for providing a hint of texture in the darkest parts of a

scene.

<p><i>As we all know the photo receptors give 14 bit values. In a perfect world this would be 14 stops. But lots of real world issues NOISE being the most significant eat / obscure the information at the lower end.</i>

 

<p>No. You are confusing two different things.

 

<p>The first thing is dynamic range which is the range of brightness that the medium (in this case photo receptors) can represent. Essentially it's the difference between the brightness that gets represented as one-step-before-absolute-black and the brightness that gets represented as one-step-before-absolute-white.

 

<p>Note that the dynamic range is a *range* -- it's a difference in brightness. It says nothing about the precision with which you label or define specific values within that range.

 

<p>Now the second thing is what I'd call tonal resolution. It is exactly the *precision* with which you can specify a particular value of brightness.

 

<p>The fact that most photo sensors output 14-bit values says that the maximum tonal resolution that they are capable of is 2^14 = 16,384 discrete steps. It says *nothing* about the dynamic range that gets sliced into these 16,384 steps.

 

<p>It's perfectly possible for a 10-bit sensor to have a higher dynamic range than a 14-bit one.

 

<p>Think of a ruler. Dynamic range is the length of a ruler, and tonal resolution is how frequent are markings on it. Quite different things.

From my reading. . .95% of dSLR users don't even know what "dynamic range" even means.

 

And no. . .my impression is that even the major review sites are not talking about it because it bores the readers. Hmmm

 

At least the 20D noise levels were improved over the 10D. I was a bit disappointed to see that the 5D not much improved over the 20D.

"Think of a ruler. Dynamic range is the length of a ruler, and tonal resolution is how frequent are markings on it. Quite different things."<p><p>

Brilliant - superb analogy

Jim, according to all available tests, the 20D has almost identical noise as the 10D, but the

20D has much more banding.

 

The '20D has lower noise' statement was from the Canon marketing department, and

somehow it got repeated all over the internet.

14 bits = 2 ^ 14th power.

 

One stop is a doubling of the light. In a 14 bit range you can double the light 14 times = 14 stops - and still have some differentition in the deep shadows and not blow out the highlights. The gradation would be lousy, especially in the shadows.

 

Of course you could adjust things so all 14 bits covered only a range of two in illumination intensity (1 stop). The gradation would be very smooth, but the dynamic range would be lousy.

Almost rigth !<br>

But with Bayer's sensor you don't have 1 photosite != 1 pixel.<br>

So, when Canon say it's a 14bits, did they speak about pixel or photosite and how 2^14 for each photosite result in possible combinaison of pixel ? <br> Did we have (2^14)*3 for each pixel ?<br>

<br>Did the 35d will have 35bits pictures ? :-)

<I>As we all know the photo receptors give 14 bit values. </i><P>

 

We all know? I didn't. <P>

 

Care to elaborate where that came from? Sensor photosites don't "give" digital values.

Maybe you mean ADC bit width? Equivalent ENOB for the sensor's SNR? Please be specific.

"But with Bayer's sensor you don't have 1 photosite != 1 pixel."

 

Time to sink this old tugboat once and for all. Bayer masks are there to enable colorblind

sensors to reconstruct color data, but in fact every photosite's luminance value is used in

the final image.

 

So, yes, in a sensor with a Bayer mask, 1 photosite = 1 pixel in the final image. The color

of that pixel is determined by using information from that photosite's neighbors, but the

brightness is left intact. You can compare images from 6 MP B&W cameras such as the

ones made a long time ago by Kodak and modern 6 MP cameras, and see that the Kodak

does not have a resolution advantage.

my analogy: dynamic range is lacking in bootleg videos, dvds. where you cant see detail in the black, and the lights are blown out to nothing, that is why I never buy bootleg movies.

If, in fact the color data interpolated by the proximate photosites leaves the existing luminance of a specific photosite intact, then why do a number of sites/persons claim that bayer mask photosite interpolation affects the overall sharpness of specific color areas within an image? Are these claims founded? (http://www.outbackphoto.com/dp_essentials/dp_essentials_05/essay.html)

 

Also if the actual R,G and B data from multiple photosites is figured into the final value of each individual photosite, then shouldn't the working dynamic range of each "pixel" value be based on the algorithm used, rather than the technical range of each individual photosite?

I'm not an engineer, but I do know that if one compares monochrome image samples from

older mono 6 MP DSLRs to newer color DSLRs, it becomes clear that the monochrome DSLR

holds no resolution advantage. Contact your local engineer for the reasons why.

 

Another way of figuring out if your assertion is true is by shooting a resolution chart. If you

are correct, a DSLR should have a substantially lower resolution than the photosites per mm

on the sensor, which they do not.