A big "noise" test with 10 Nikon D-Slr's

[AndyE]

Hi Nikoncafe members,

I posted yesterday my first pics with the new D40X.

One photo with my dog, taken at ISO 1600 raised many questions about the relative ISO performance of the D40X to Nikon model x or y (esp. D80 and D200). (here is the post)
So I started yesterday night a comprehensive comparison of high ISO noise with all Nikon cameras I could find.


The Setup

To be able to compare every photo with every other one, I decided to take them indoors.
A black background, a black table and some uninteresting stuff at the table (as proof of lack of creativity ).
I mounted a AF-S 200mm/2 VR on a tripod to facilitate and easen the changes of camera bodies.
It also helped to keep the same frame between all bodies.

Lightning was set up with a Profoto D4 generator, one head and a 3x4' softbox.
The D4 is nice as I can adjust it over 8 f-stops and it doesn't change color (reduced my WB setting effort)


The Contenders

All digital Nikon bodies I could find (sorry for the D70 and D100 folks - these models were out with friends)

• From the old boys club - The D1, D1H and D1X.
• With the D2X and D2H we had 2 sporty guys.
• With the D200 and D50 two well established work horses came in.
• and last but not least. The new kids on the block. The D80, D40 and D40X.

The Process

All bodies had been set up to:

• WhiteBalance = flash
• Timer delay = 2 sec
• Lowest ISO setting
• Noise reduction = off
• File format = JPEG, large, fine
• used a Pocket wizard to connect to the flash unit
• Aperture = f11
• Shutter = 1/125 sec (it really doesnt matter)

I used the following settings on the Profoto flash unit:

• ISO 100 = 600Ws
• ISO 125 = 525Ws (all the best from the D1X ...)
• ISO 200 = 300Ws
• ISO 400 = 150Ws
• ISO 800 = 75Ws
• ISO 1600 = 38 Ws
• ISO 3200 = 19 Ws
• ISO 6400 = 19 Ws (and aperture set to f16)

With those bodies which provide noise reduction, I did a second run. Starting at ISO 800 all the way up. Saved separately.


Where are the files?

I have posted all original and unprocessed JPEG files at www.pbase.com/andrease/noisetest.
Please be aware about the 260MB, if you want to download ALL JPEG's.
To access the individual original files, please check first the "Original" tab under the pbase picture before you download.
All smaller versions (medium and large) are heavily compressed by the pbase.com servers (with all possible artefacts)

There is a simple scheme for the filenames:

D1-ISO-0200.jpg = D1 with ISO 200
D2H-ISO-Hi2-NR.jpg = D2H with ISO 6400 (1600+Hi2) plus Noise reduction = On


Results

Disclaimer: I had only ONE body per camera type. There can be differences between samples.
Your milage might/will vary. If there are any errors, please raise them. I try my best to fix them.

There is progress - undeniable.

So, what is now the state of affairs with ISO 1600?
Looking at the original files, I would argue that there is a strong connection between release date of a specific camera and high ISO noise.
Independent of the price level. The later it is released, the better the performance (generally speaking).

and yes, the D40X is better than the D40 (and the D50).

I've received many requests for a comparison to the D80 and D200.
What shall I say? Ok, let it out.
According to these pics, the D40X is better than the D80, which in turn is better than the D200.
Please download the original files to check for yourself.


That's the new one - D40X, ISO 1600



People are most interested in these two

D200, ISO 1600


D80, ISO 1600



Same class

D40, ISO 1600


D50, ISO 1600



The Professionals

D2X, ISO 800 + Hi1 (= 1600)


D2H, ISO 1600



Old guys

D1X, ISO 800 + Hi1 ( = 1600)


D1H, ISO 1600



...this is the way ISO 1600 looked like 8 years ago: D1, ISO 1600



All the best (and enjoy an extended pixel peeping session at the link provided above)

Andy

[Paul aka MontyDog]

#1064 - You have an error in your SQL syntax;

[Rick Waldroup]

Thanks Andy. That was great. I shoot a D2H and a D1x and I've always known that the D2H was noisy, especially in darker areas or underexposed areas. When I got my D1x, I was impressed right away by what I thought was less noise and truer skin tones. Your test, at least to my eyes, confirms that. I like my D2H for the speed, etc. But my D1x has become my everyday walk around camera.

Thanks once again.

[Tosh]

Wow! Great work, Andy. Thanks. I'll be studying these when I get home later tonight.

Do you have any sense of whether the cameras' noise performances would follow suit in Raw images?

[AndyE]

Originally Posted by Tosh

Wow! Great work, Andy. Thanks. I'll be studying these when I get home later tonight.

Do you have any sense of whether the cameras' noise performances would follow suit in Raw images?

Thanks, Glenn.
Don't know if the ranking would be the same with RAW files. These are plain-out-of-the-camera JPEGs.

Cheers,
Andy

[AndyE]

Originally Posted by Paul aka MontyDog

Have a very Happy Easter Andy

Same to you Paul.

All the best,
Andy

[Czechman01]

Thank you for taking the time to go through all of this effort. Something like this is long overdue and I personally thank you for the info.

However, instead of these full frame, reduced size, images it would be much more helpful if you took 100% crops of small areas and put them next to each other so we can get a quick overview of each.

Then a link to the full size images for those wanting more details.

[AndyE]

Originally Posted by Czechman01

Thank you for taking the time to go through all of this effort. Something like this is long overdue and I personally thank you for the info.

However, instead of these full frame, reduced size, images it would be much more helpful if you took 100% crops of small areas and put them next to each other so we can get a quick overview of each.

Then a link to the full size images for those wanting more details.

Thanks Woody, appreciated.

I did not want to provide 100% crops for the following reason:

Covering cameras from 2,7 MP to 12,2 MP isn't that easy. If it would be true 100% crops, it would be considerable different in size and influencing the perceived quality. Scaling them to same size wouldn't be true crops and again impacting the perceived quality.

Cheers,
Andy

[Czechman01]

Megapixel size doesn't matter.

A properly exposed flat gray surface will reveal noise levels, so a 200 pixel square crop of a flat solid color object like a gray card would give a quick overview of noise performance on an apples-to-apples comparison.

It wouldn't be an in-depth analysis, just a quick overview.

Originally Posted by AndyE

Thanks Woody, appreciated.

I did not want to provide 100% crops for the following reason:

Covering cameras from 2,7 MP to 12,2 MP isn't that easy. If it would be true 100% crops, it would be considerable different in size and influencing the perceived quality. Scaling them to same size wouldn't be true crops and again impacting the perceived quality.

Cheers,
Andy

[Beezle]

Thanks for the effort.

The D40x is looking interesting.

I really wonder what the next pro body will be like in this regard.