We've got a lot of discussion around high ISO performance of the D800 vs other cameras - D700, D3, D4, etc. I've got a different perspective and expect the D800 will outperform the D700 and D3 in terms of noise at high ISO levels. In fact, DxO has indicated the sensor scores are likely to be the highest ever measured - but they have yet to perform the hands on review. http://www.dxomark.com/index.php/Publications/DxOMark-Reviews/Nikon-D800-Review
DxO represents the standard for camera and lens testing. Dxo has developed not only a methodology for testing, but also a methodology for comparing different cameras with widely varied numbers of pixels. The approach they use normalizes around a common print size - a 30x20 cm or approximately 12x8 inches - about 8 megapixels. That means images from high megapixel cameras are downsized more than lower megapixel models. In the case of the D800 you have approximately a 4:1 reduction while the D700/D3 have a 1.5:1 reduction.
The other factor is more obvious - it's the characteristics of the photo sites and the relative size of those photo sites. Essentially we have a large pixel pitch approach used by the D3/D700/D3s and D4, and we have a small pixel pitch approach used by the D300/D300s and D7000 - and now the D800.
In the case of the D3x vs. the D700/D3, the low light performance of the individual pixels is much better on the D700/D3 due to the 50% larger pixel pitch. But most of that benefit is lost due to downsizing to the standard image. The D3X ends up with an ISO score of 1992 vs. 2302 for the D700 - or about 1/3 stop advantage to the D700. The D300 score is 679 - approximately 1.5 stops worse than the D3x with similar pixel pitch on both.
But the D7000 fares well. In spite of still smaller pixel pitch than the D300, the D7000 has an ISO score of 1167 - nearly 1 full stop better than the D300.
Putting it all together, that means the next generation sensor improves ISO performance by approximately 1 stop, and the larger megapixels downsized to a standard print pick up 1.5 stops on the D3x and the proportionately larger megapixels of the D800.
Collectively, this means the D800 will be approximately 1/2 stop better than the D700 and D3 in terms of ISO performance - largely because of the large megapixels.
The D4, of course, shares the technology of the D7000 and D800, but has a larger pixel pitch, so you can expect the D4 to outperform the D800 by a half stop. And that will make it a full stop better than the D3 - approximately the amount measured and reported by Bill Claff.
#3. "RE: Just to clarify Eric..." In response to Reply # 2
<Downsizing is not the same as cropping. Photographers using the DX crop or cropping to 15MP after the fact should expect about the same noise performance as the D7000.>
Jim - that is correct. If you have a 15mp DX crop, you will negate the advantage of the larger megapixels and will have something very close to the D7000. That's still a fine camera and nearly a full stop better than the D300 in terms of noise.
#4. "I make it 2/3 stop..." In response to Reply # 3
From the D300s at least. DxOmarks rates the low light ISO of the D7000 at 1167, the D300s at 787. What my eyes couldn't tolerate is the 25% smaller image (.70/.93 vs. D300s) - but that's another topic.
#6. "RE: D800 High ISO performance" In response to Reply # 5
San Francisco, US
Thanks Eric, very interesting, useful information. I found some high ISO images from the D800 on the Mansurov's site, if anyone is interested. I have no connection with the site, but these are the first images I've seen that didn't come from Nikon's website. My take is that the images support what Eric says.
#7. "RE: D800 High ISO performance" In response to Reply # 0 Thu 09-Feb-12 03:28 AM by PerroneFord
I'm confused by something here...
>Collectively, this means the D800 will be approximately 1/2 >stop better than the D700 and D3 in terms of ISO performance - >largely because of the large megapixels.
It is generally held that the D3s is about a stop better than the D3s. Perhaps that is overstated and it's merely a half stop better. I don't know. If the D800 is purported to be a half stop better than a D3, that would mean it's roughly equal to or a stop behind the D3s. Ok, fair enough, but then you offer this....
>The D4, of course, shares the technology of the D7000 and >D800, but has a larger pixel pitch, so you can expect the D4 >to outperform the D800 by a half stop. And that will make it >a full stop better than the D3 - approximately the amount >measured and reported by Bill Claff.
If the D4 is only better than the D800 by half a stop, does this mean the D4 has equal or less performance than the D3s?
#8. "RE: D800 High ISO performance" In response to Reply # 7
You've got a couple of typos and I may have added to the confusion by being conservative and excluding the D3s.
The D3s is a half stop better than the D3 and D700 due to tweaking performance via software. The sensor itself is the same but it performs better through Nikon tuning the software.
When you start talking 1/3 to 2/3 of a stop difference, my rounding may or may not work.
The D4 is probably a half stop better than the D3s, but DxO has not tested it yet. It still has not been delivered. It could turn out to be more - 2/3 stop or a full stop. The technology as illustrated by the D7000 which has been tested is a full stop better than the prior generation, and a half stop better than _s updates.
Even if high ISO performance of the D800 is the same as the D3s, the color rendition and dynamic range are significantly better.
#9. "RE: D800 High ISO performance" In response to Reply # 8
Vancouver (WA USA not Canada), US
Your predictions for the D800 are in line with mine. If you look at something like my Photographic Dynamic Range (at my site) then the D800 should be a little more than 1 stop better than the D7000, just above the D3 and D700 for the high ISOs and comfortable below the D3S and D4. The low ISOs are harder to predict. They will probably tail off but it's less clear where; I'm guessing at about 9.5 on my chart (between the D700 and the D4).
#12. "RE: D800 High ISO performance" In response to Reply # 0
Vancouver (WA USA not Canada), US
DxO represents the standard for camera and lens testing.
Unfortunately, they do not represent the standard for sensor testing (as opposed to camera and lens). They do not report read noise, gain, full well capacity, or other well established sensor characteristics. These values could theoretically be recovered from their SNR curve data, but that data is thoroughly overcooked. Too bad, because they to seem to have good protocols.
Regards, Bill P.S. - Inquiries to DxOMark regarding these issues go unanswered.
#14. "RE: D800 High ISO performance" In response to Reply # 0
Great post Eric. Thanks.
Of course now I think the choice for me, between the D4 and the D800 is tougher, since their high ISO/low noise characteristics are so close. (LOL)
If I only did sports or wildlife (flying birds) moving larger animals, etc. the choice would be an easy D4. If I only did travel, architecture, macro, weddings, the choice would be an easy D800. Unfortunately for me, or fortunately as it's lots of fun, I do travel, architecture, events, news, only an occasional wedding as I really like to avoid them, and all kinds of wildlife, with some sports.
This is getting tough and I've got to make a decision soon, as if I'm going for the D800 (can't justify both right now) I want to cancel the D4 order sooner than later, and definitely before it's shipped to me.
#15. "RE: D800 High ISO performance" In response to Reply # 14
Ned, I am somewhat in the same spot. I generally shoot two bodies for sports. Right now that is 2 D3s. I had intended to upgrade them both to a pair of D4's. I had been using my D7000 as a daytime sports camera, and a studio/fashion camera. So now I am somewhat baffled as to how to do my upgrades.
Buying 2 D4's is one option. But it's going to be one HECK of a stretch to buy 2 D4's and a D800. The D3s is meeting 90% of my needs right now. So my current thinking is to buy 1 D4, keep the D3s in rotation, and buy a D800. I'd probably let the D7000 go, but it's not going to fetch me much money so that's up in the air.
This is a mess. Not what I wanted at all. Good problem to have, but still a problem.
#16. "RE: D800 High ISO performance" In response to Reply # 15
Perrone, maybe the mods will move our posts over to another topic, as this is kind of an offshoot, not really right on target, but not far off either.
Anyway, I was now thinking of sticking with the D700 (My poor man's D3S) and get the D800 with grip. Then, with the rest of the cash purchase some other equipment I could use for the event work I seem to be doing more of these days. I'm thinking dump the 2nd D700 (maybe keep it for a short time longer), and definitely dump the D200 at this point.
#19. "RE: D800 High ISO performance - specific performance ranges" In response to Reply # 0
Maybe some details on DxO's measurements would help. Camera performance is highly dependent on what ISO levels you use in practice.
As DxO measured the low light ISO, they calculated the D3s at .4 stops better than the D3 and .6 stops better than the D3x. The D3 at introduction was .15 stops better than the D3x. But at high ISO levels, I interpret the advantage of the D3s to be greater - nearly a full stop.
For context, the other overall scores for color and dynamic range were the same for the D3 and D3s - slightly below the D3x. But that's not the full story as the D3x advantage comes from low ISO performance where it is substantially better. At higher ISO levels the color and dynamic range scores favor the D3/D3s.
The detailed measurements for the D3s show improved performance over the D3 at almost every ISO level - with the difference in performance greatest at high ISO levels (above ISO 1600). The D3x excels at low ISO levels with far superior performance at lowest or base ISO - minimal noise and exceptional color and dynamic range. At ISO 200 and beyond there is little difference until you get to ISO 1600 and above where the D3s shines.
What this means is there are optimum ranges where each camera has an advantage. The high megapixel cameras perform better than anything on the market at base ISO - ISO 50 to ISO 200. With all these cameras there is an increase in noise and a dropoff in color and dynamic range as ISO increases - and they are pretty similar from ISO 400 to ISO 1600.
The D3 initially had small advantages over the D3x at higher ISO levels, but with the D3s the camera apparently was re-tuned to achieve significant improvements at high ISO levels.
What this means for the D800 is that you can expect fantastic color and dynamic range at base ISO. But as you move into normal and high ISO levels the larger pixel pitch of the D4 will prevail. As for comparison with the D3/D3s/D700 - it appears the only place the latter might outperform the D800 is at high ISO levels - above ISO 1600 or even ISO 3200.
Using the D7000 as a reference point for the pixel and processor performance in the D800, we should see a significant step of improvement (I'd estimate a full stop at all levels). The D7000 reduced the pixel pitch from the D3x, but was able to generate substantial improvements in dynamic range, color, and high ISO performance to nearly match the D3x in those areas. That means really good things for the D800.