Reading mode: Light Dark

D3 dynamic range?

Go to last post
M

macphistophe

Member
Messages
30
Reaction score
0
Location
US
Could anyone comment on the D3 dynamic range versus the D2x at low iso? Is there a significant performance gain? Are we getting close to the Fuji S5?

Thanks

macphi
 
How about this for cruide math: Thom Hogan says the D300 is about 11 EV, (and by Phil's metrics the D200 and 40D is about 11 EV in RAW) and the general concensus is that the D3 is about 2 stops better. So 11 + 2 = 13.

Another way to look at it: I downloaded the ISO 200 NEF file from hollypapa
http://forums.dpreview.com/forums/readflat.asp?forum=1021&thread=25844645

and fiddled with it in Lightroom. I was able to recover about 1.5 EV of highlights of the bright inside lights, and push the dark lamppost in the lower right hand corner about 4 EV. So that should give a total of 12.5 EV if we call a jpg image 8 EV. Talk about primitive guestimations.

Funny thing, in NX, with D-Lighting, I was able to "recover" all of the highlights in the lightbulbs inside the mall, although I think NX was faking it but the end result looked good. That would be at least 4 EV of highlights, but obviously ain't real or is it?

A final way to estimate the dynamic range: If the system is noisefree, then the DR is roughly equal to the number of bits, i.e. 14. However, this is rarely the case, so drop one EV for noise, and again the magic number is 13! (As an aside, there are cases where with noise present and good averaging techniques, you can actually exceed the number of bits in the ADC in terms of dynamic range. And a noisefree sensor is actually not the most desirable when wanting optimum linearity at the bottom of the dynamic range.)

So that's three different techniques on the back of my envelope giving a DR in RAW of 13. So now we'll sit around and wait for the official "Phil" number.

Carsten Thomsen
 
Oops, I forgot to mention that the Fuji in RAW is about 12 bits, so if the D3 is 13 . . . you should be a happy camper.

I wasn't able to find any D2x RAW DR numbers but would guess about 11.

Carsten Thomsen
 
Carsten Thomsen said:
How about this for cruide math: Thom Hogan says the D300 is about 11
EV, (and by Phil's metrics the D200 and 40D is about 11 EV in RAW)
and the general concensus is that the D3 is about 2 stops better. So
11 + 2 = 13.

Another way to look at it: I downloaded the ISO 200 NEF file from
hollypapa
http://forums.dpreview.com/forums/readflat.asp?forum=1021&thread=25844645

and fiddled with it in Lightroom. I was able to recover about 1.5 EV
of highlights of the bright inside lights, and push the dark lamppost
in the lower right hand corner about 4 EV. So that should give a
total of 12.5 EV if we call a jpg image 8 EV. Talk about primitive
guestimations.

Funny thing, in NX, with D-Lighting, I was able to "recover" all of
the highlights in the lightbulbs inside the mall, although I think NX
was faking it but the end result looked good. That would be at least
4 EV of highlights, but obviously ain't real or is it?

A final way to estimate the dynamic range: If the system is
noisefree, then the DR is roughly equal to the number of bits, i.e.
14. However, this is rarely the case, so drop one EV for noise, and
again the magic number is 13! (As an aside, there are cases where
with noise present and good averaging techniques, you can actually
exceed the number of bits in the ADC in terms of dynamic range. And
a noisefree sensor is actually not the most desirable when wanting
optimum linearity at the bottom of the dynamic range.)

So that's three different techniques on the back of my envelope
giving a DR in RAW of 13. So now we'll sit around and wait for the
official "Phil" number.
Click to expand...
The human eye can only see 10-12 stops as far as i know and the hman eye is still notably better than any camera made.

In the zone system, there are only zones 1 through 9, and only 3 through 8 have any real detail....

So I'm still mystified how you get 13.... ?

For Reference:



--
Zane
http://www.pbase.com/devonshire
Nikon D2x
NAPP Member

'Life is not measured by the number of breaths we take, but by the number of moments it takes our breath away.” ~ Anonymous
 
The DR of D200 is about 8.2 EV, and around 9 EV for 40D according to Phil. I don't know where did you get the data that D200's DR is 11 EV.

And Thom said it loud and clear that there is NO WAY that DR of D3 can surpass that of Fuji S5's.
 
by the engineering definition, DR = full well capacity/read noise, as measured by EJ Martin, who is a professor of physics here at the University of Chicago and quite technologically proficient. This is for base ISO of 200.

http://forums.dpreview.com/forums/read.asp?forum=1021&message=25627719

This engineering definition places the noise floor at zero and is rather liberal. The actual photographic dynamic range would be somewhat less. The DR is about the same as the Canon 1D MII but it is achieved at an ISO of 200.

The entire thread is of interest:
http://forums.dpreview.com/forums/readflat.asp?forum=1021&thread=25658562
--
Bill Janes
 
I'm no expert but the S5 is 14-bit and the Fuji cams have used 14-bit DAC's since the S2. I'm really looking forward to seeing waht kind of DR the D3 is capable of.
--
Regards,
Jeff
 
Carsten Thomsen said:
How about this for cruide math: Thom Hogan says the D300 is about 11
EV,
Click to expand...
Since I'm being misquoted here, I need to step in. The figure I cited is what a sensor company would report in its specifications, which uses a different definition of DR than photographers would. You can't mix and match "DR" values. The way Phil measures it is different than the sensor makers is different than Site A or Site B or Photographer X or Photographer Y. You'd only be able to compare within one tester's regimen, and then only if they were consistent in their testing.
Carsten Thomsen said:
Funny thing, in NX, with D-Lighting, I was able to "recover" all of
the highlights in the lightbulbs inside the mall,
Click to expand...
D-Lighting, and Active D-Lighting, are techniques for re-curving the linearity of the sensor's response, essentially an exaggerated form of local or micro contrast. You don't get "more DR" from them, per se.
Carsten Thomsen said:
A final way to estimate the dynamic range: If the system is
noisefree, then the DR is roughly equal to the number of bits,
Click to expand...
This, too, has been discussed many times before, and the answer is no. The underlying response of the sensor is fixed. You can assign as many bits as you want to record that response. Generally, we're in the stage of design where engineers are still trying to match bit depth to electron count. This was Stage 1 in audio CD technology, which has now moved on to Stage 5 or 6...

--
Thom Hogan
editor, Nikon DSLR Report

author, Complete Guides: D40/D40x, D50, D70s, D80, D100, D200, D1 series, D2 series
http://www.bythom.com
 
Ansel Adams based the zone system not on what a negative was capable of recording, but on what a print was capable of retaining. The real limiting factor on dynamic range is still the media on which we see the final image -- paper, or computer monitor. Pretty much any DSLR around is capable of producing images with more dynamic range than we can possibly reproduce on print or screen.

So why worry about more "useless" DR in our cameras? Because it gives us either the choice of which end to drop (drop the shadows, blow the highlights), or the choice to compress the whole picture and keep everything in range on the final print.

The range of the human eye is something of a red herring here. The way we accomodate a scene with great dynamic range is by scanning it repeatedly, with wider "aperture" (pupil) on some parts and narrower on others. This doesn't translate well to photography.

In any case, there is no medium out there yet that can show anywhere near 11 stops of DR.
 
Dear Thom,

From one "old" audio engineer to another . . . I've worked with high performance A/D converters most of my life, but didn't want an overly technical post. I agree that dynamic range can be defined many different ways (as in the audio world), but what I'm trying to guess, is whether the D3 has better dynamic range than the D300. When I'm working in ballsparks, I believe your 11 EVs for the D300 is a fairly representative ballpark number and Phil quotes the D200 as about 11 EVs (all are with RAW files) at low ISO's and it appears that at the low ISO's the D200 and D300 are about the same. (And yes, I recognize they're not exactly the metrics).

What really interests me, will the D3 be better than the D300 or not. Does it have enough dynamic range so it can play highlight recovery games a la Fuji? If so, it would be a major breakthrough. Also if it went further into the deep blacks, it would be a candidate "one shot" HDR camera. I suspect that is what Nikon is trying to do with Active D-lighting, to map the 11-13 EV's into 8 EV's we can print.

The reports of many of the pros who have shot it seem to indicate that the Active D-lighting succeeds quite well along this goals.

Obviously you can't get away with agressive active D-lighting if the digitized signal is not clean and there are multiple indicatations that that is an important part of the technology that Nikon now is bringing to the table.

So I'll impatiently rest my case and wait for yours and other reviews. Bought your D70 book a while back and greatly appreciated it.

Carsten Thomsen
 
Because that would make my day!

--

Serious hobbyist, very serious gearhead. How serious? 6 months ago I use a 1.3mp P&S, now I use:
D3 (preordered) with 17-35, 24-70, 70-200VR, TC17 II
D200, D40, Fuji S5 pro with 17-55, 12-24, 18-200VR, 50 f1.4
Gitzo GT2540, Arca-Swiss Z1
 
Thom Hogan said:
Carsten Thomsen said:
A final way to estimate the dynamic range: If the system is
noisefree, then the DR is roughly equal to the number of bits,
Click to expand...
Click to expand...
Thom Hogan said:
This, too, has been discussed many times before, and the answer is
no. The underlying response of the sensor is fixed. You can assign as
many bits as you want to record that response. Generally, we're in
the stage of design where engineers are still trying to match bit
depth to electron count. This was Stage 1 in audio CD technology,
which has now moved on to Stage 5 or 6...
Click to expand...
This has been discussed many times before, but some people continue to falsely assert that there is no relationship of DR and bit depth, and they usually use a stair step analogy where DR is the height of the stairs and bit depth the size of the steps. With floating point representation, this analogy would be true. However, with linear integer encoding it is not. If you wish to increase the height of the stairs, you must add more steps (bits), since the size of the step is fixed.

Of course no sensor is free of noise. Shot noise can not be eliminated, but is very low in the shadows and zero in the black frame used to determine read noise. If we use the engineering definition of DR (full well capacity/read noise), the DR would be infinite with no noise. In this hypothetical case, the DR would be limited by the bit depth to 1 stop per bit as Carsten stated. DR can be increased by increasing the full well, reducing the read noise, or both. DR would not be increased by placing a 16 bit ADC in front of a sensor that has a DR of 12, but it might improve tonal gradation and 16 bit ADCs are used in some medium format digital cameras whose sensors are rated at 12 stops by the engineering definition.

A general definition of DR is the ratio of the maximum useful signal to the minimum useful signal. The latter could be limited by bit depth or noise, but the same considerations apply.

In practice, the DR is usually limited by noise. However, the bit depth in a linear integer raw file is absolutely limited by the bit depth to 1/stop per bit. Very few cameras match the bit depth to the electron count (unity gain). For example, the D3 full well is slightly more than 50K electrons and a bit depth of 14 maxes out at 16,383 for a gain of about 3.2 electrons/14 bit data number. See Roger Clark:

http://www.clarkvision.com/imagedetail/digital.sensor.performance.summary/

--
Bill Janes
 
You must log in or register to reply here.

Keyboard shortcuts

f
Forum
m
My threads
Mobile site
About
Editorial content
Cameras & Lenses
Community
All content, design, and layout are Copyright © 1998–2025 Digital Photography Review All Rights Reserved.
Reproduction in whole or part in any form or medium without specific written permission is prohibited.
When you use DPReview links to buy products, the site may earn a commission.
©GPS Media - Guides, Products, Services.
Back
Top