Nikon D90 Chroma Noise at ISO 800:

Here's a crop from the original image's middle-left portion of the frame (indicated by the red outline in the original image, shown at bottom), showing an example of the Nikon D90's chroma noise at 800 ISO (it's a near-actual size, high-resolution .jpg at "best" quality--the image is very faithful to the original .tiff):

Nikon D90; Tokina 11-16mm f/2.8
ISO: 800; exposure: f/5.6 @ 1/60th; aperture-priority; spot-metered; -0.7 EV.

Here's the zero-crop, full-frame image, below:

Nikon D90; Tokina 11-16mm f/2.8
ISO: 800; exposure: f/5.6 @ 1/60th; aperture-priority; spot-metered; -0.7 EV.

This shot was taken last night at 8:31PM Pacific Time. This is the period of dusk known as, "magic hour," when the sky goes deep blue, and is also the precise time at which I'll be shooting all of my planned portfolio pieces.

jonnyapple said:
It seems like the real test will be if people know you're lying when you tell them your stuff was shot medium format but it was actually a D3x. ;-)

Good one, jonny!

It seems like the real test will be if people know you're lying when you tell them your stuff was shot medium format but it was actually a D3x. ;-)

adamz said:
yes, it's from my own experience (I've played with all of the Nikon bodies for long enough to have my own personal opinion), if I write about sth I googled I put a reference, if not it's from my own experience.
sorry that I wasn't clear enough, but english is not my mother tongue
anyway, have fun with d700 it's amazing camera

I wasn't questioning your background, I was just wondering if the data was anecdotal, or performed under any sort of standardized procedure, i.e., identical light quantities, side-by-side body comparisons, etc. Also, I didn't mean to deride your ESL status--my apologies.

anjz said:
Thanks for the TIF. Frankly, I think the photo looks pretty good. Personally, I see mostly luminance noise in the crop. I would go and try out a D700 and see if you find much of a difference at that ISO. I performed a little test at my local camera shop between a D700, D90 and Canon 5D Mark II and found. I didn't see much difference at ISO 1600.

Certainly. I should have posted that crop initially. Thanks for your comments, but I'm sure on a large print, that noise would be pretty objectionable to just about anyone.

Well, again, thanks for everyone's comments. I'm not too familiar with the DxO data acquisition/testing methodology, and I'm much more comfortable with real-world image tests. Also, my images tend to have lots of chroma, so generic tests are rarely comparable to what I'll actually be doing with the camera.

I just heard from my other photographer friend who just came from another gallery opening:

1. Everything was printed to 32" x 40," Type-C, chromogenic prints.
2. Everything was shot with a Hasselblad, using a digital back.

While it's certainly worth a weekend rental to see what Nikon's highest-end FX body can do at that print size, it looks like I may have to at least consider going with a medium-format, digital back, which of course would be a rental, and/or a Mamiya 6 x 7 system and drum scans for this project.

Thanks for the TIF. Frankly, I think the photo looks pretty good. Personally, I see mostly luminance noise in the crop. I would go and try out a D700 and see if you find much of a difference at that ISO. I performed a little test at my local camera shop between a D700, D90 and Canon 5D Mark II and found. I didn't see much difference at ISO 1600.

yes, it's from my own experience (I've played with all of the Nikon bodies for long enough to have my own personal opinion), if I write about sth I googled I put a reference, if not it's from my own experience.
sorry that I wasn't clear enough, but english is not my mother tongue
anyway, have fun with d700 it's amazing camera

anjz said:
Would someone help me out? I've been staring at the photo of that guy for 3 minutes and don't see any chroma noise. Maybe someone can describe the location that it exists. I just see some luminance noise where orange meets gray near the upper left-hand corner and generally around the photo, but nothing that probably wouldn't also appear in a D700 as well. I'd be interested in learning what I am missing.

Sorry, it's probably difficult to see in those files. As soon as I get a chance, I'll post some larger crops from the original file. I recently switched everything over to Macs, but I don't yet have a CS5 license for them--all of my Adobe licenses are for PCs (which are all sitting in a closet now). It's pretty noticeable full-size, on a large screen. It's most visible in the large wash of blue light on the cove, toward frame-right.

Here's a link to a full-resolution, uncompressed .tiff (cropped) of the first image:

http://studio460.com/images/noisenocomp-s1.tiff

Recent Convert said:
I hate to dwell on it, but the answer *is* in the link I gave in my earlier post. DXO states its method as follows:
>>>The Low-Light ISO metric indicates the highest ISO sensitivity to which your camera can be set while maintaining a high quality, low-noise image (based on a Signal-to-Noise-Ratio [SNR] of 30dB, a dynamic range of 9EVs and a color depth of 18bits). The Low-Light ISO metric is of primary importance in photojournalism, sports and action photography.<<<

When you look at the rating for the D700, you will find a noise-free ISO rating of ISO 2303, while the D90 exhibits a noise-free rating of 977.

Not at all! Sorry, I was kinda tired when I was trying to look at your link. Thanks for extracting and summarizing some of that data!

adamz said:
read my first post, if You doesn't know that 1EV is 1 stop, than it only confirms for me, go to BASICS
also, chroma abb is not equal image noise, so please don't confuse it

Yes, I know that 1 EV equals one stop. If by "chroma abb," you mean chromatic aberration, I know what that is too. But that's not the point--the image noise is I'm talking about is NOT chromatic aberration.

adamz said:D90 is almost noise less till iso 800, and with good technique even up to iso 1600. d700 will give You max 2 stops better (if You have good technique, if not 1 stop), while d3s will give You app 3 stops better.

Yes, I did read your first post. I just found your post difficult to read, and your statement, " . . . will give You max 2 stops better (if You have good technique, if not 1 stop)," just poorly worded, if clarity was your goal. Also, you gave no source for this data. Is this "2 stops better" from your own, hands-on experience with both bodies?

studio460 said:
Sure, noise can reduced in post, but at the expense of some degree of sharpness. But, I simply asked how many EV faster, the D700 is, over the D90. Does no one here own both?

read my first post, if You doesn't know that 1EV is 1 stop, than it only confirms for me, go to BASICS
also, chroma abb is not equal image noise, so please don't confuse it

heartyfisher said:
I dont think that any one has gone into the detail you need with your specific criteria. I think you will have a look at samples yourself and decide if what you see is good enough. .. would be keen to hear your findings.. http://www.imaging-resource.com/

Thanks, hearty. If I do get a D700, I'll post the data right here on NikonRumors!

shivaswrath said:
the D700 is about "half as noisy" as a D90, from what I can gather from DxO, other picture samples, etc.

Thanks, Shiva. I'll have to try to look at the DxO data again. If the D700 is half as noisy, I would guess that to be possibly +1 EV? The D3 is out of the question since I also have to save money for a digital cinema camera later this year (Panasonic AG-AF100 or RED S35 Scarlet). Also, the image sample noise looks way worse at full-resolution!

By the way, the portfolio pieces will most likely be 100% location work (I hadn't actually planned on any studio photography). But, it's all going to be magic hour stuff with models, so there still will be a lot of saturated blue hues in the image. I like the idea of double-shooting on 6x7. In fact, I may even just shoot the DX images as back-ups (or, like really good "Polaroids"), and try to shoot the 6x7 stuff for the "real" images.

clillja said:
Was I reading the exposure data correctly on the image of the producer? ISO 800 f1.6@1/2000? You could have shot that at ISO 200 at 1/500..? If that's the case, that's a lot of unnecessary noise. If you're going to be "really picky," you can't do stuff like that and expect the camera to make up for it.

Yes, you're right, the ISO could have been set much lower. Those were quick grab shots, "stolen," while doing my real job. I only had a few seconds to take "non-work" images. The ISO was set high for another shot, which I never had a chance to get. The light levels set for this scene were to accommodate our full-sized, 2/3" HD cameras (Sony HDW-F900Rs) which only need about 20 footcandles to reproduce a 70% IRE fleshtone.

Portfolio set-ups would be completely different, shooting under circumstances entirely under my control.

As an aside, broadcast video cameras are commonly rated for sensitivity, so it's relatively easy to compare sensitivities of varying cameras. In videoland, I know that a camera rated at f/11 @ 2,000 lux is exactly one-stop faster than a camera rated at f/8 @ 2,000 lux.

clillja said:
You need to think harder about about noise reduction in post. Yes, it can reduce edge sharpness - but it can be a good tool - and like any good tool - it's adjustable and flexible and you need to learn it and practice with it to get the most out of it.

Yes, I'm sure it is. I haven't ventured into any post work recently, since all of my Adobe CS licenses are all stuck on PCs (and I'm now all-Mac). What's the "killer app" among noise-reduction plug-ins which most people use these days? (I'm ordering CS5 for OS X next week and currently use Aperture for cataloging).

Loooking at the samples - yeah, you're going to be happier with a D700 and more so with a D3S. But - you're a lighting guy, right? You must understand how important your own work is and how much it contributes to the final product? You may not achieve the noise levels you seek without optimal lighting. The camera, ANY camera, will not make up for that.

Was I reading the exposure data correctly on the image of the producer? ISO 800 f1.6@1/2000? You could have shot that at ISO 200 at 1/500..? If that's the case, that's a lot of unnecessary noise. If you're going to be "really picky," you can't do stuff like that and expect the camera to make up for it.

You need to think harder about about noise reduction in post. Yes, it can reduce edge sharpness - but it can be a good tool - and like any good tool - it's adjustable and flexible and you need to learn it and practice with it to get the most out of it.

the D700 is about "half as noisy" as a D90, from what I can gather from DxO, other picture samples, etc.

But for your studio work, I would HIGHLY suggest a D3x, upping the lighting ante, and keeping it close to 100 ISO. . . .you won't get better detail than a D3x right now, and for $7400 now new, it's relatively more of a bargain than at introduction. . .

if you weren't doing studio work, than the D3s would've been my other suggestion. . .

I honestly think the noise is "within acceptable" in the samples you provided - although, I think I might PP more than the next and try to make it right in the end. . .

studio460 said:
Sure, noise can reduced in post, but at the expense of some degree of sharpness. But, I simply asked how many EV faster, the D700 is, over the D90. Does no one here own both?

I dont think that any one has gone into the detail you need with your specific criteria. I think you will have a look at samples yourself and decide if what you see is good enough. .. would be keen to hear your findings.. http://www.imaging-resource.com/

I wonder how accurate DXO's numbers are? For example, the D300s ranks below the Sony Alpha 550 in Low Light ISO (807 vs. 787). Does that pan out in photographs? Seems like the D300s "DXO acceptable" ISO is much lower than the D90 and the D5000 and even a Sony DSLR. Doesn't seem right.

studio460 said:
Sure, noise can reduced in post, but at the expense of some degree of sharpness. But, I simply asked how many EV faster, the D700 is, over the D90. Does no one here own both?

I hate to dwell on it, but the answer *is* in the link I gave in my earlier post. DXO states its method as follows:
>>>The Low-Light ISO metric indicates the highest ISO sensitivity to which your camera can be set while maintaining a high quality, low-noise image (based on a Signal-to-Noise-Ratio [SNR] of 30dB, a dynamic range of 9EVs and a color depth of 18bits). The Low-Light ISO metric is of primary importance in photojournalism, sports and action photography.<<<

When you look at the rating for the D700, you will find a noise-free ISO rating of ISO 2303, while the D90 exhibits a noise-free rating of 977.

Would someone help me out? I've been staring at the photo of that guy for 3 minutes and don't see any chroma noise. Maybe someone can describe the location that it exists. I just see some luminance noise where orange meets gray near the upper left-hand corner and generally around the photo, but nothing that probably wouldn't also appear in a D700 as well. I'd be interested in learning what I am missing.

heartyfisher said:
If its for a book.. I doubt the production quality of the printed pages of a book will be high enough for people to discern the noise you are seeing.

Actually, I would imagine the noise would likely be exacerbated when moved to a four-color offset printing process (dot-gain issues, etc.). Using a video analogy, any noise present in the camera masters' acquisition video only worsens whenever transcoded.

PB PM said:
If you need cleaner shots at high ISO get the D700, they will be better. Not to mention noise reduction software works better, with less detail loss, thanks to the larger pixels.

Yes, larger pixels = better!

adamz said:
studio - I guess You forgot that digital is not film, chroma aberration on Your second photo are easily corrected in PP . . .

Sure, noise can reduced in post, but at the expense of some degree of sharpness. But, I simply asked how many EV faster, the D700 is, over the D90. Does no one here own both?

studio - I guess You forgot that digital is not film, chroma aberration on Your second photo are easily corrected in PP, and lie hearty wrote, production quality of print will not match the quality You see (unless You gonna look at Your "BIG" images with a magnification glass). anyway, get whatever You want and be happy with Your choice

If you need cleaner shots at high ISO get the D700, they will be better. Not to mention noise reduction software works better, with less detail loss, thanks to the larger pixels.

heartyfisher said:
If its for a book.. I doubt the production quality of the printed pages of a book will be high enough for people to discern the noise you are seeing.

By "book," I meant portfolio (models and photographers often refer to their "book," when talking about their portfolios). Also, the eventual chromogenic prints will be quite large, designed for gallery exhibition and sale. I see photo gallery exhibitions around town (Los Angeles) with a few photographers still displaying images shot on 4x5, so my little D-SLR stuff better look damned clean.