Which body gives best signal to noise ratio

[ed_hurst]

<p>Hello all,<br>

I have recently started shooting a lot of stars and am loving the D800E. However, I can't help wondering whether one of the other recent bodies with fewer, larger pixels might achieve lower noise but lower resolution. Of course shots from such a body have to be enlarged more for any given output size. So my question is this - for any given output size, which body produces the best results in terms of noise when shooting at ISOs between, say, 1000 and 6400? I am not concerned with the lowest pixel level noise, but the lowest possible level of noise apparent at the output size.</p>

<p>All advice gratefully received.</p>

<p>Best wishes,</p>

<p>Ed</p>

[bgelfand]

<p>I suggest you start by going to the DPREVIEW site, the review of the D800, the noise page to be specific, and compare the various models you are interested in.</p>

<p><a href="http://www.dpreview.com/reviews/nikon-d800-d800e/18">http://www.dpreview.com/reviews/nikon-d800-d800e/18</a></p>

<p>The page allows you to compare various camera brands and models at various ISO setting.</p>

<p>Then you can go the <em>Popular Photography</em> magazine site and look at their reviews of the various models you are interested in. They quantify the noise figures:</p>

<p><a href="http://www.popphoto.com/gear/2012/07/camera-test-nikon-d800-dslr">http://www.popphoto.com/gear/2012/07/camera-test-nikon-d800-dslr</a></p>

<p>Their noise figures are from DxO Analyzer software. Here is their test procedure></p>

<p><a href="http://www.popphoto.com/reviews?page=0,0">http://www.popphoto.com/reviews?page=0,0</a></p>

<p> </p>

[mike_halliwell]

<p>...or, <em>obviously</em>, go to the source of the info....DxO!</p>

<p>http://www.dxomark.com/index.php/Cameras/Camera-Sensor-Database/Nikon/D600</p>

<p>This is the page for the D600, but they are all there. Look under the Measurements tab....then the options are <em>which SNR</em> you want. If you hover your cursor over one of the dots on the graphs it gives the actual reading. Enjoy!</p>

[bruce_rubenstein]

<p>Pixel size isn't everything; there's also quantum efficiency.</p>

[keith_b1]

<p>There's a handy comparison chart at this link: http://home.comcast.net/~NikonD70/Charts/PDR.htm#Ideal_FX<br>

Note: I cannot personally testify as to the veracity of the info contained therein.</p>

[keith_b1]

<p>There's a handy comparison chart at this link: http://home.comcast.net/~NikonD70/Charts/PDR.htm#Ideal_FX<br>

Note: I cannot personally testify as to the veracity of the info contained therein.</p>

[bruce_rubenstein]

<p>Those graphs are for dynamic range, which is not S/N ratio.</p>

[pete_s.]

<p>Go here:<br>

http://www.dxomark.com/index.php/Cameras/Camera-Sensor-Ratings/%28type%29/usecase_sports<br>

Top spot go to D3s, then D600 followed by D800E, D4 and D800. Then comes the Canon bodies.</p>

[studio460]

<p>Pete said:</p>

 

<blockquote>

<p>Top spot go to D3s, then D600 followed by D800E, D4 and D800. Then comes the Canon bodies.</p>

</blockquote>

<p>Very interesting. I hadn't seen that particular DxO data in that manner. I was just thinking of selling my D3s since I thought its ISO performance was comparable to my D800E in some real-world tests I did recently (I'll have to re-test under better control to confirm my findings).</p>

[mike_halliwell]

<p>If you wade into the actual SNR graphs on DxO, you spot some interesting 'discrepancies' between manufacturer claimed ISO and measured ISO.</p>

<p>That difference is verging on -2/3 EV on the D800, particularly at the high end. I'm not saying Nikon is worse or better than anyone else, but taking the manufacturers figures 'blindly' may not be helpful. I think Ralph's view about doing your own tests is the way to go.</p>

[studio460]

<p>Yes, I had previously put a lot of faith in those DxO numbers. However, I shot some different tests recently (not the ones shown below) which clearly showed differences in the ISO response between the D3s and D800E at "same" ISO settings.</p>

<p>Below are <em>other</em> tests I did a while ago (whose goal was to determine if an image acquired in the D800E's crop-mode was as good as an uncropped D7000 image--the D800E's crop-mode image was visibly better than the D7000's full-DX frame's image). I'm not 100% sure if these are apple-to-apples comparisons (as I said, I need to re-test), but I up-sampled the D3s image to D800E resolution (which makes more sense to me than down-sampling the D800E image to D3s-resolution) for the sake of this comparison.</p>

<p><img src="http://studio460.com/studio460/D3s-6400.jpg" alt="" /><br /> Nikon D3s @ ISO 6400 [up-sampled to D800E-resolution]</p>

<p><img src="http://studio460.com/studio460/D800-6400.jpg" alt="" /><br /> Nikon D800E @ ISO 6400 [at native D800E resolution]</p>

[studio460]

<p>Here are the full-frame images from the above 100% crops:</p>

<p><img src="http://studio460.com/studio460/D3s-700.jpg" alt="" /><br>

Nikon D3s @ ISO 6400</p>

<p><img src="http://studio460.com/studio460/D800-700.jpg" alt="" /><br>

Nikon D800E @ ISO 6400</p>

[Andrew Garrard]

<p>It is odd that DxO seem to believe a lot of cameras are significantly overestimating ISO. Either there's a lot of marketing fluffery, or possibly they're testing with a lens on the camera and not compensating for the difference between f-stops and t-stops. (I'm guessing at an explanation which is generous to the manufacturers. That would explain a relatively constant error, not the D700's larger disparity at ISO 25600, for example - though since ISO 25600 on a D700 is just ISO6400 underexposed by two stops and boosted digitally, I can't explain why it behaves differently.) I could believe that the spectral response of the D800 is not the same as the older cameras (I think I heard they changed the filters), so I wouldn't expect an exact match.<br />

<br />

I am nervous about trusting the single figure summaries from DxO - "high ISO performance" meaning "the ISO at which the noise exceeds a value which we've decided is acceptable on a particular input". The graphs are more informative, but a single "this camera is <i>this</i> good at high ISO", not so much. (That's not their fault - if they're asked for a single number comparison, it's no worse than any other, it's just that the story is more complicated than can be expressed in this simple form.)<br />

<br />

Incidentally, note that the 1Dx is appreciably ahead of the other Canons, whereas all the recent full frame Nikons are very similar. This may mean that other Canon sensors in future cameras will catch up as the technology rolls out. (Actually, at high ISO, the Canons are pretty competitive - it's the dynamic range at low ISO where there's a big gap.)</p>

[mike_halliwell]

<p>The ISO graphs also <em><strong>seem</strong></em> to show that the lowest ISOs of many Nikon cameras is underestimated, ie Nikon says 50, experiments say 80 etc. This shows itself as something to do with base ISO and the next one up, being the same in reality. Weird or what???<br /> _______</p>

<p>I (almost) wonder if this is why some people need to dial in -2/3EV at lowest ISOs on their D700s? (ISO 100 actually = ISO 162)</p>

[studio460]

<p>I would really love to hear any comments on the above 100% crops. Which one looks "better?" And, why?</p>