D700 vs D300 sensor noise

[geirrosset1]

<p>Hi<br>

This weekend me and a friend of mine did a shoot together. We used her D700 and the Sigma 24-70mm F2.8 @ ISO200. The setup was: Elinchrom D-Lite4 in Large Octabox camera left, Silver reflector camera right and below and a D-Lite4 bare bulb pointed at a gray paper background. Result is below.</p>

<p>Then yesterday I had to do another shoot. The photos are for a hair dresser competition and they required three different hair styles, we only shot two this weekend, someone had not read the rules. I used the same lights, but this time a D300 @ ISO L1.0, same lens and a black paper background. Result is below.</p>

<p>There is a lot more noise in the background. Is that because of the D300 sensor, because I used a black background or maybe both? If the difference between the sensors is that big I'm getting a D700, but if it is a matter of exposure, which I thought looked right at the time, I'll just have to work on it.<br>

Sincerely<br>

Geir Rosset<br>

Oslo, Norway</p><div></div>

[geirrosset1]

<p>Uploading two images didn't work. So here is the D300 image...</p><div></div>

[pcnilssen]

<p>Geir, it is hard to get a good enough view on screen of the two images you have provided, but I have the D300, and in general, I would be very careful using the L or H ISO-values. You might try shooting in RAW and use a good denoising software if the results are not as you want.</p>

[mariosforsos]

<p>Noise could be due to the file type you were using - for example, if the D700 was setup to shoot RAW and your D300 was setup to shoot JPEGs, then than could potentially account for the "noise" (personally, I see pixelation more than noise).</p>

<p>Also, you do not provide any information about exposure times - a long exposure could potentially produce more noise than a shorter one. It would depend on the length of that exposure.</p>

<p>Finally, you usually get noise at the dark areas - the first image is very, very bright and therefore you wouldn't see as much noise, even in the same settings.</p>

[geirrosset1]

<p>These were both shot at 1/160 (f9 and f13 if I am not mistaken) in RAW 14-bit. As for more noise in the dark areas. It is the light areas of the background of the D300 shot that are "noisy". I'll post some 100% crops. First the D700.</p><div></div>

[geirrosset1]

<p>And then the D300 crop.</p><div></div>

[Kent Shafer]

<p>The more pronounced mottling of the background in the D300 crop seems too large scale to be sensor noise. Could it be that the background's texture, emphasized by the small light source, is more nearly in focus in the D300? f/13 and a shorter focal length on the D300 would provide more depth of field than f/9 and a longer focal length on the D700.</p>

[ArthurRichardson]

<p>The D700 has a bigger sensor area to harbour the approximate same amount of pixel than the D300. This gives the D700 an advantage in terms of better signal to noice ratio. This may very well explain the difference you noticed.</p>

<p>For a proper comparison, you should however the same set up, model and lighting conditions and than swap bodies. This is a bit of apples vs. oranges.</p>

[kevin_delson]

<p>The test images you show are far from controlled.<br>

In one the light is more intense; the other is quite a bit reduced.<br>

Light bouncing off a blonde and then another subject with dark hair?<br>

Not a fair or proper test.</p>

<p>While true the FX sensors will perform better in low light, I don't see this as a problem in your images.</p>

<p>Try a simple test. Shoot a consistent target with a consistent light source.<br>

Shoot ISO, 160, 200, 400, 800. Those should be enough to reveal what is happening.</p>

[ShunCheung]

<p>You should use the same subject, same background, and same lighting if you want to conduct a fair A/B test.</p>

[akshun]

<p>I agree with Arthur, the D300 has 12.3 million pixels vs. 12.1 million in the D700 which definitely has a direct affect on noise.<br>

Id also like to compare which "picture control" settings were used, and if any fine tuning was done to either setting.<br>

i have both camera bodies, and without question , at any ISO, the D700 is far superior to the d300 in terms of amount of sensor noise.</p>

<p> </p>

[photo5]

<p>The D700 has stronger anti-aliasing filtration than the D300, so at ISO 200 it looks softer out of camera, and looks like it has less noise. I prefer the sharper, slightly noisier output of the D300 when shooting at ISO 200.</p>

[richardsnow]

<p>I also suggest shooting at the same ISO. You indicated that you shot at ISO 200 with the D700 and ISO 100 (LO 1.0) with the D300. The HI and LO settings on both the D300 and D700 are noted to be used with caution and in extreme conditions...which this shoot obviously was not as you had controlled lighting.</p>

<p>Both the D700 and D300's native ISO is 200. This will generally produce the best results as there is no signal amplification, (or reduction in the case of the LO settings) involved.</p>

<p>Next time, in addition to the other suggestions made for a controlled test, rather than the "apples and oranges" test you provided, please keep the ISO consistant between the two cameras.</p>

<p>RS</p>

[Jerry_]

<p>Nikon did not make the D300 as a *little* D700. The D700 handles ISO noise better, even with both camera bodies using the same ISO setting (as you have found out.)</p>

<p> </p>

[james_kennedy9]

<p>Isn't there always going to be more noise in the dark areas simply because the signal to noise ratio is going to less in the dark areas due to less photons?</p>

[briany]

<p>I agree with Kent. I think what you're reacting to on the full images isn't noise but rather the mottling of the background, which is more blurred with the D700 image.</p>

[orcama60]

<p>Thom Hogan's book : Complete guide to the Nikon D300 will help you to understand how the D300 works. If you read pages 280 and 290 ( ISO chapter ), you will find out a lot of information that you should take in consideration when shooting with this camera which according to Thom's review, "you can produce nearly noise-free images with exposures of 10, 20 or even 30 seconds on a D300 / D300's" He does recommend shutter before ISO but in case you need it, then to use Long Exp NR in dim situations and probably when using artificial lighting like the ones you used for that shooting. Personally, I do like the photo of the second model that you took with the D300. Perhaps, a bit brighter that you should correct with your software. Have a happy shooting. </p>

[boinkphoto]

<p>We'll see if I skewered for this, but owning the D300, the D300 has a lot of lower ISO noise. This can be particularly seen in areas without detail. This ("base ISO noise") came up quite a bit on DPReview when it was first introduced. I say this also comparing it to both my Fuji S5 and D80 output.</p>

<p>Regardless, it is likely given nearly the same number of megapixels that a FF camera will have lower noise than a DX camera. If you look at the ratio of the number of pixels to the size of the sensor, obviously a FF will end up with larger sized pixels than APS-C. See this:</p>

<p>http://en.wikipedia.org/wiki/Image_sensor_format</p>

<p>to get some sense of of the size difference.</p>

<p>Think of each pixel as a "light bucket". If photons are streaming down on the sensor, just as if rain was falling, obviously the larger bucket will collect more photons over a given time period. The smaller the bucket, the less photons that will hit it in a given period. Thus there is effectively less data from which to judge the actual values. With less data there is more room for error, and that error displays itself as noise (this is also where the term "signal to noise" comes from for sensors).</p>

<p>In some sense you can look at noise as the image processor lacking enough data from a given pixel and having to "guess". Obviously guesses are less than perfect, thus "noise". This also explains what happens with higher ISOs - the image processor is amping up the data it has, which is insufficient, hence more guesses and more noise.</p>

<p>I'm sure there are better and probably more accurate explanations, but I think (hope) that gets the point across. Generally the larger the sensor, the less noise.</p>

<p>One caveat - there are tricks that can be used, whether noise reduction software/firmware or sensor design that can reduce noise additionally. Thus in theory an older larger sensor or even a same sized sensor with less megapixels can yield more noise just because of the lack of technology improvements. However since the D700 and D300 are roughly same generation (and the D700 sensor is notorious for noise control - hence its high ISO support), it's not surprising the D700 beats the D300.</p>

[vineet.rajasekhar]

<p>Hi Geir, </p>

<p>I own a D700 and have used it voraciously for about 10 months now. I've never used a D300. </p>

<p>A lot of my pre-purchase research revolved around choosing between the D700 and D300. I discovered by looking at a number of test shots that at ISOs < 400 or so, there's no visually significant difference in noise. It really only becomes noticeable at the higher end. </p>

<p>If you intend to do a lot of shooting at high ISOs, I can recommend the D700 wholeheartedly, I've spent entire evenings shooting (by mistake) at ISO800 and only discovered the mistake when I looked at the image Metadata. Even at ISO800, you really need to look at the images pixel by pixel to perceive a difference. </p>

<p>Over ISO800, noise becomes noticeable, but very manageable. </p>

<p>Regards, <br>

Vineet</p>

[iwong]

<p>If you have aperture openings as small as f/9 or f/13, maybe you shouldn't have used ISO100 (L1.0) on the D300 either. First thing first, defraction limits will cause your photos to soften up and loss of details will result. The optimal aperture for most lenses tend to be in the f/4 to f/8 range, regardless of DX or FX format. Stopping further down will hit the "rules of diminishing returns" where any gains in depth-of-field will be wiped out by defraction effects.</p>

<p>Secondly, when shot at their base ISO's (200), both the D300 and D700 are remarkably noise-free. I have both bodies and often use them together. I would have a hard time telling which image was shot with which body without looking at the EXIF data, if both were shot using their base ISO. The mottling of your background may be just the texture of the background itself?</p>

[ty_mickan]

<p>Geir, how does the noise show up in the prints that you made? I'd be surprised if that was noticable, and even when pixel peeping, it doesn't detract from the image whatever.</p>

[eric_arnold]

<p>personally, i prefer the d300 shot. if the model is good-looking enough, who notices the background?</p>

<p>the d700 pic might have [slightly] less noise, but it looks overexposed.</p>

<p> </p>

[geirrosset1]

<p>First off. Let me just say I am overwhelmed by the response. No shortage of folks in here willing to help, thank you :D<br>

Ty: I haven't printed them out yet. I will be printing them out on Friday to hand them in. They have to be post stamped on Friday to be eligible for the contest. We'll see what it looks like then I guess.<br>

As <a href="../photodb/user?user_id=628631">Brian Y</a> said, it is more of a mottled look than actual noise. This type of mottling I have found it very hard to get rid of. Even blurring just creates a posterized look in the transistions.<br>

The first image is a bit bright, but these are not the final versions that will be printed. The hair dresser is still sorting through all the shots to pick out the ones that best show off the hair.<br>

As for comparing apples and oranges. I know it was not a 100% fair test. Next week we'll be testing both bodies with the same background and the same lighting.<br>

I'll get back to you with the result of that test.<br>

I did start thinking after posting this though. In the old days of film, the grain of a certain type of film was often considered to add a certain something to the final image. And now suddenly, with digital, "noise" is more often than not considered unacceptable, even though that is the sensors signature, it's grain if you will. There was never a demand that film be perfect, why so now. I think I can live with it, even though sites like iStockphoto seem to frown on even a hint of noise. I submitted a long exposure, night time photo, to iStock because I felt it had real presence, it was rejected due to noise. I asked them if there was any camera body out there that could produce a noise free image in the same conditions, and they said no. So noise is unacceptable, even if it is unavoidable.</p>

[rob_piontek]

<p>You might try different software to convert your RAW. What did you use? Did you adjust the exposure with the RAW converter? With regard to Low 1.0, my experience is that if the contrast of the scene is reasonable, and you're not concerned about losing some highlights, you get lower noise levels by using it. Of course results should be the same as using ISO 200, overexposing a stop, and bringing back in post.</p>

<p>I have always been critical of the D300 ISO performance compared to other Nikon's I've owned. I just didn't think it was much of improvement over what I already had. Still I am surprised by your results...</p>

 

[alvinyap]

<p>I don't own a D700, but on my D300 I can definitely detect this mottling artifact at ISO200, and is very obvious during studio shoots against a dark grey background. Not sure if that helps but yeah. I'm using Bibble 5 for raw conversion btw.</p>

<p>I'd say at base ISO, my now-deceased D200 does not show this mottling effect.</p>