D200 v. D300 Image Quality Explored

[tonybeach]

There is already widespread consensus about the better high ISO and AF

performance of the D300; so I will focus on just one aspect of IQ (image

quality) in this thread -- namely, how the cameras fared in a daylight shot of a

static subject at ISO 280 that I recently took within seconds of each other.

Both cameras had the Nikkor 50/1.8 AF-D lens* mounted and were set to f/8,

1/250, auto-ISO and Center-weighted metering at +.3 EC.

 

Here are 100% crops:

 

http://photos.imageevent.com/tonybeach/mypicturesfolder/sharing//Noise%20and%20Detail.jpg

 

 

These were taken from the highlighted portions shown in the image at the bottom

of this post. These crops were taken from the Capture One conversions which

were easier to get similar results from, whereas the NX conversions required the

D2XMODE2 Picture Control setting for the D300 and Non-Picture Control settings

(Color Mode II, Normal Tone Compensation, Enhanced Saturation) for the D200.

Something that is unpredictable when trying to compare the D300 and the D200 is

the tendency of NX to sometimes treat the colors in the files the same and to

sometimes treat them differently depending on the scene.

 

There is something about these two cameras that I was not expecting, and that is

a peculiar tendency of the D300 to expose hotter than the D200 even at the same

aperture, shutter, and ISO settings. However, after dialing both images back (I

always utilize ETTR anyway) the D300 retains all of its highlights the same as

the D200. The differences in the two files I am looking at in this post

depended on whether I was using NX or Capture One. In NX the D300 file was

dialed back one full stop and the D200 file was dialed back .18 of a stop; in

Capture One the D300 file was dialed back half a stop and the D200 file was

dialed back 1/10 of a stop. I would expect that this is indicative of a 8/10 to

4/10 stop improvement in DR (depending on which converter is used), and given

what I see in the shadow noise between the files the usable DR is probably even

more than that.

 

My conclusion looking closely at the files is that the D300 file has slightly

more detail (consistent with its greater resolution), slightly less noise in the

sky (more in the red channel, but less in the other channels and overall less

noticeable), and there is a distinct advantage in shadow noise over the D200.

If you are having trouble seeing what I am seeing, download the file and look at

it in Photoshop because the file is AdobeRGB and loses shadow detail if it is

viewed in a non-color aware application. If you still can't see the difference,

your monitor likely needs to be calibrated.

 

You can argue that the difference of a shot taken under these conditions is not

worth fretting over, and I would tend to agree -- especially if you are

considering raiding your retirement fund to merely upgrade from a D200.

However, there is a marginal advantage in the D300 files which will show up in

larger prints. There are other reasons to prefer the D300 and the quicker AF

can play a definite role in improving image quality; but if you are staying at

lower ISOs, have time to get the focus right, and are not printing over 20" --

then the D200 is a fair compromise against the more expensive D300.

 

*Some may question the lenses I used in this test since they were the same model

but different copies. I have tested these lenses extensively and their

performance is indistinguishable from one another (it is a rare case of absolute

QC consistency).<div></div>

[Matt Laur]

Thanks, Anthony. Curiously, you're making me feel pretty good about sticking with the D200 for a while yet.

[mars790]

I'm actually feeling pretty good about going for the D300. There is more than a slight

advantage, especially in the bottom crop's noise.

 

Thanks for the test Anthony.

[trunfio]

And you made me fell bad about buying the D300! :) Actually, no. I need the AF for sports

and appreciate the gain in ISO. Thanks!

[bruce_margolis]

Thanks for the tests, Anthony. While you don't say it exactly, it sounds to me like you are saying the D300 has maybe a 1-stop advantage over the D200, though I suspect this would be more obvious in lower light situations.

 

I would like to test something similar comparing the D80 v the D300 when I get a chance. The visuals are very helpful for many people.

[photo5]

Between the D80 and D300, with RAW anyway, they are pretty close. With JPG, there's no

contest, the D300 is just far better with straight out-of-the-camera JPGs, which I really

appreciate a lot with this camera. The D80 is no slouch though, it's still fantastic.

[chuck_french1]

I have about 6000 exposures on the D300 in the last month or so. I think it is a much better camera than the D200. I have been shooting mostly with the 200/400 both with and without the battery pack. The focusing ability, the big screen, and the high ISO performance make the D300 a fabulous machine. I am using my D200 as a macro camera with the 70/180 lens and this has worked very well. I started with the D70s and I am thinking of putting the 12-24 on that and I will be covered for everything without having to change lenses very often.

 

I used the 28-70 f2.8 for some high ISO shots at several get togethers over the holidays. It performed magnificently, though I have a lot to learn about getting the best out of the technique.

 

In summary, Hooray for Nikon! Bring on the high megapixel D3 and the big glass you promised months ago

[rombon]

The differences are modest indeed but only if AF is spot on. I find AF spot on with my D300 much more often than with my D200 even with static objects. I think that it is the factor worth considering if the decision to buy or not to buy D300 is on the table.

 

Regards, Marko

[kohanmike]

Makes me more resolute to replace my D70s with a D300.

[walterh]

So do we agree that we are at the point where we find only three relevant differences for image quality between camera bodies in general, including consumer bodies and pro bodies of a wide range in cost?

 

1) ISO performance primarily depending on pixel size and (from now on a minor point) technical development

 

2) Will the "handling" of the camera including viewfinder, AF, dedicated buttons, LCD screen imaging of detail in "chimping" etc. let you get your shot

 

3) Post processing software and skill

 

In other words for still live or similar shooting a D40x type camera is all we need?

 

We all know photographer skill is THE major point but let us not discuss this here :-P

 

And thanks Anthony for your recent longer posts on the subject. I know how much work is required for this.

[andy a.]

To see just how far the D300 (and D3) have advanced over previous Nikon DSLRs, take a look at this chart from Bill Claff (and lots of other relevant information <a href="http://home.comcast.net/~nikond70/">here</a>)

<br><br>

<img src="http://home.comcast.net/~nikond70/Investigations/Photographic_Dynamic_Range.jpg">

<br>

[frank_skomial]

Why D3 crop mode has less dynamic range ?

 

Is it using the same pixels, and size of them ? - perhaps just less of them and only from the center. Where did Nikon goof here ? - by not providing the same quality for D3 DX cropped mode, not the same as for the D3 FX mode.

[photo5]

And I find it rather hard to believe that the 14 bit NEF files don't have more dynamic range than the 12 bit. Like wikipedia says, [citation needed]

[walterh]

Andy can you please point us a bit more specific to the origin of the data?

 

The link you provide gives a whole bunch of interesting links but many are broken links and may be the data you refer to are somewhere lost? The graph is just too interesting not to be able to look at the data :-)

[walterh]

Dave if the noise floor limits dynamic range than a higher bit resolution will do no good for dynamic range it will just be nice to have for post processing. If the maximum count numbers would limit the dynamic range then 14bit should offer more dynamic range over 12bit - sadly this is not the case.

[bruce_margolis]

Say what? Am I reading this DR/ISO chart wrong?

 

It looks to say that the dynamic range of the D50 exceeds all the pre-D3/300 models between ISO 200-1600, including the D80 and D200. Huh?

[walterh]

Bruce yes indeed. Larger pixel size and better readout electronics of the D50. Nice camera if you do not need all the extra controls. The D70 is also lower in noise than the D200, quite noticeable at 100 and 200 ISO especially if you sharpen a lot^^.

[vivek iyer]

Not only the DR, there are surprises when it comes to better UV/IR suppression of the sensor used.

[eric_arnold]

<<Say what? Am I reading this DR/ISO chart wrong?

 

It looks to say that the dynamic range of the D50 exceeds all the pre-D3/300 models between ISO 200-1600, including the D80 and D200. Huh?

>>

 

when i first saw this chart on nikonians.org, the D50 results bugged me out too. i wouldn't be surprised to see a run on whatever D50s are left in stock because of this (especially because they'll AF with more lenses than the d40/d40x). but to put it in perspective, the d50's performance at ISO 200 (its lowest value) trails all the other D-series bodies at their base ISO. btw, the d40x didn't do too shabby in this comparison either.

 

the real loser in this chart, though, is the d2x, which is like, dead last across the board.

 

and while this chart was done by an enthusiast (as opposed to a pro lab), it does show fairly consistent findings, i.e. the DR curves all follow a more or less predictable arc.

[douglas lee]

If still interested/needed, here is the direct link to the above referenced chart:

 

http://home.comcast.net/~nikond70/Investigations/Photographic_Dynamic_Range.htm

 

-Doug

[andy a.]

I'll see if I can find the original nikonians discussion that sparked that image. Bill has written some simple software software using dcraw I believe to examine raw NEFs and do some SNR calculations. He has a strict definition for "photographic dynamic range" which sets an threshold for acceptable SNR and I think takes into account sensor size/density. Further, he provides the software and instructions so people can test their own cameras, and that's how the chart was compiled. I don't think anyone has found a flaw in his methodology yet.

<br><br>

Re: 12-bit vs. 14-bit, remember, you're just increasing sample resolution, not lowering the noise floor or raising well capacity. Any increase in dynamic range may end up just being attributable to better precision in those very bottom bits which lowers noise a tad by decreasing rounding error.

<br><br>

Re: my original link, also see some discussion and elaboration under Collaborations->Dynamic Range Collaboration.

<br><br>

...

Here is one of the original nikonians discussions that I think gives this some context:

<a href="http://www.nikonians.org/cgi-bin/dcforum/dcboard.cgi?az=read_count&om=18129&forum=DCForumID71">link</a>

[tonybeach]

For Bruce:

 

"While you don't say it exactly, it sounds to me like you are saying the D300 has maybe a 1-stop advantage over the D200, though I suspect this would be more obvious in lower light situations."

 

> S/N and usable DR are closely related; so is EV and ISO (on Nikon DSLRs). Regarding noise at ISO 800-1600 under incandescent lighting, below is the D300 at ISO 1600 alongside the D200 at ISO 1600 and 800.<div></div>

[tonybeach]

I do not see a lot to distinguish the other cameras in Bill Claff's DR chart, the differences are marginal and may be attributable to sample variation. We do not know how many cameras were used, nor do we know how many of each model were used. I would bet that if we were to examine crops from a D50 and even a D2x made from RAW files using the came RAW converter, that there would be only relatively small differences compared to what is happening with the D300 and D3.

 

Regarding the measurement of the D3 DX crop mode, I am especially skeptical. Bill states that:

 

"Photographic dynamic range is a practical measure of the ability of the sensor to capture an image that will have acceptable noise at a normal enlargement and viewing distance.

 

This is distinctly different from dynamic range as defined in engineering terms."

 

Clearly he is not measuring the behavior of individual pixels. What is "normal enlargement"? If you enlarged a D3 DX cropped image to the same size as the full frame D3 image (finally, an opportunity to properly use the term "full frame"), would you notice the loss of detail or the increased enlargement of noise more? This also explains why 10-12 MP DSLRs are holding their own against 6 MP DSLRs (viewing them at a pixel level after all involves different assumptions about the degree of enlargement).

 

Anyway, it is a topic for another thread and it is a distraction here to nitpick over D50 or D3 DR characteristics. This was supposed to be a thread about the D200 and D300 image quality. In that regard, discussion about the D50 and D3 DX mode mainly serve to undermine the validity of Bill Claf's results (which may otherwise be largely valid).

[chuck_french1]

I appreciate all the technical work that has been done. From a totally personal and practical point of view I would offer the following comment. I have been shooting from 1600-3200 ISO with the D300, something I attempted a couple of times with the D200 and abandoned because of the lousy quality. I think the D300 can be used successfully as a available light camera in most situations once you learn how to do it correctly. It was my not understanding how to do it correctly that caused the noise, not anything inherent in the D300.

[dcraton]

For Anthony: just a note of thanks for the nice work and post. I know it will be helpful for many.