Resolution with the D700

<p>Having recently purchased a D700, I ran some tests with lenses I've checked before with film on my USAF test charts to see how it tops out at 100% in RAW @ 200 ISO. Fortunately, the AF is spot on accurate right out of the box. It also matches my reliable F100 and has a slightly tighter envelope around the point of exact focus most likely due to improved AF technology and absolute flatness of the sensor compared to film. Resolution using my 15" computer screen tops out at about 50 lp/mm compared to film where recorded measurements can be as high as 100 lp/mm with the same lenses. Obviously resizing the full image to fill the screen shrinks it down and makes it look amazingly sharp but my question pertains to resolution. Is the perceived resolution more a factor of the screen size or the resolution limit of the D700 at 12 mp?</p>

<p>From what I gather the D3x at 24 mp is reaching the limits of information that can be gathered from the FX size sensor, therefore the D700 will not be exploiting the potential sharpmess available from the FX at the optimum apertures. Film is around 20 mp, so resolution will be higher than the D700 (with the right films). Note that the D3x only realises full potential at some apertures - around f5.6.</p>

<p>Let me ask from pure ignorance ....</p>

<p>if a line pair is defined as a black line next to a white one, then it takes a minimum of two pixels to = one line pair? ( one black and one white pix)</p>

<p>If the D700 = 12 MPxl and that is 4256 x 2832 pix at 36.0 x 23.9mm ...</p>

<p>... in the horizontal ... 4256 / 36.0 = 118 pix / mm = 59 line pair / mm ?</p>

<p>So your '50' measurement is in line with reality ?</p>

<p>Am I close ... help Dr. Bob Atkins!<br>

Jim</p>

<p>For photo.net's D3X review, I did a bunch of A/B tests using the same lens on the D3X and D700. Finally I picked this image as an example: <a href="http://www.photo.net/photo/9308533">http://www.photo.net/photo/9308533</a><br>

There is certainly a difference, as you can read the fine print better on the D3X, but this is fairly subtle.</p>

<p>@Ian: Where did you find that information that ~24MP is the limit for FX sensor?</p>

<p>Don't get me wrong as I'm merely speculating, but if Canon can squeeze 15MP out of an APS-C sensor and FX sensors are still relatively immature, I would assume that there is still more MP available?</p>

<p>15MP on an APS-C is roughly 45k px/mm², 25MP on FX is 27.8k px/mm². If FX achieves the same pixel density that implies ~40MP max. Of course, this doesn't take into consideration scalability or other issues that may occur with a larger sensor.</p>

<p>I guess one of my points was, does the size of the monitor make any difference in limiting the apparent resolution such as a 15" laptop screen compared to a 42" high definition LCD or plasma screen? On my 15" laptop screen, the 50 lp/mm group is resolvable where you can distinguish the individual lines but when the lines get smaller, you can see spruous resolution or what looks like a moire type pattern. To put film in context, my Nikon Coolscan 5000 film scanner can resolve about 64 -72 lp/mm on detail so I guess the D700 isn't too bad at around 50 lp/mm as long as too much cropping won't be needed.</p>

<p>I love film, and I still shoot film frequently, so I'm not a must-use-the-latest-and-greatest film-hating digital fanboy.</p>

<p>That said, I can PRINT larger and sharper prints from a D700 than from any 35mm film image that I've ever captured over the years.</p>

<p>Numerical analysis of line pairs per milimeter is a fascinating exercise - in futility. A digital photograpy is NOT the output of a set of pixels. It's the result of INTERPOLATION of data captured by a collection of pixels. Interpolation algorithms are getting better and better, and the one in the D700 is spectacular.</p>

<p>The same thing happens when we listen to a music CD. We're not listening to 44 thousand 16-bit samples per second. We're listening to the result of all of that sampled data being processed by some very sophisticated smoothing algorithms in the digital-to-analog converter chip. Our digital photos are not Bayer patterns. The Bayer pattern is only a raw material in the process of creating the final image. Detail in a digital photo is partly the product of pixel density and partly the product of inferences made from that pixel data.</p>

<p>Probably not good enough for whatever you may be doing.</p>

<p>Go get an Hasselblad H3D or a Leica S2 or the Nikon D4X, since the D3X is inferior for hi-res.</p>

<p>Compare</p>

<p>Nikon D700, 12 MP, FX size sensor, and you complain about insufficient resolution?</p>

<p>Canon SD990 IS, sensor is tiny - perhaps 5x smaller, but 14.7 Mega Pixels.</p>

<p>As far is pure pixel resolution per millimeter goes , who wins here ? but you would not touch the SD990 pictures after you had tried the D700 camera, perhaps?</p>

<p>.. and of cource you cannot compare picture quality from those 2 cameras, they are in different league. </p>

<p>Think more about picture quality or pixel quality, versus share quantity of pixels, or the resolution on the sensor surface.</p>

<p>Thanks Dan for putting it it context. I figured pinting from the computer will give a better result than from slides or negs which is why I got the D700 given my full frame lenses will work well in FX format.</p>

<p>Alan why would you base the measurement of the camera resolution on your personal look at your personal laptop monitor? Beats me. Do not even try doing this :-)</p>

<p>Capture resolution and display resolution should be kept strictly separate.<br>

Otherwise people with different laptops and different reading glasses would get different resolutions for the D700?</p>

<p>A better way would be to capture an image and draw a line profile across the image of a resolution target in a measuring software like ImageJ. There are more complex ways to do this but even this simple approach gives a reasonable information.<br>

See: http://rsbweb.nih.gov/ij/</p>

 

<p>Alan,<br>

I imagine you are using a 50mm lens stopped down to f/8 shooting in RAW etc. The theoretical resolution of D700 can be calculated as 4256 pixels / (36mm X2) = 59 ln/mm, there is also an optical low pass filter which filters the high spatial frequencies so the extinct resolution is always lower, the best way to measure would be to shoot in RAW format, turn sharpening off and read the marking on the chart when you start to see aliasing (moiré). This indicates you are close to the Nyquist frequency and the extinct resolution of the sensor + low pass filter.<br>

The effect of the low pass filter is like a convolution filter, some of the signal can be recovered by de-convolution or sharpening, so when you apply a sophisticated sharpening filter you might be able to recover a bit of what was lost to the low pass filter but the upper limit is 59 ln/mm for D700. <br>

It should be noted that when you view at 100% (i.e. one pixel on sensor = one pixel on the screen) the measurement will be independent of the your screen resolution. If you use a very high resolution screen however, depending on your eyesight it might be a bit difficult to determine where the lines start to interfere so a regular desktop LCD is best for this test.<br>

Now does higher lpm resolution equate to better overall sharpness and IQ? <br>

The answer depends on sensor size, for the same size for example D700 vs D3X vd 5DII etc. which all have 24X36mm sensors, it does, as long as the optical lens is not the limiting factor. According to the laws of diffraction in optics, every lens has an associated circle of confusion, if the pixel pitch becomes smaller than COC then there will be no benefit in added pixels or higher theoretical lpm. <br>

The other issue is noise, if the pixels are too small (for example Digicams) the signal from each pixel becomes very noisy and cannot distinguish between subtle levels of intensity (contrast) and extinct resolution will be much lower than theoretical resolution especially if detail is low contrast.<br>

Overall in my very subjective opinion based on the prints that I have made you can make good looking prints with D700 up to 18X12 and perhaps larger if subject is abstract, I was never able to get such quality from film. </p>

 

<p>William, 24 mp is not the maximum possible - more is available by my own working - BUT as the mp count gets higher, it becomes more tricky to actualy utilise the extra file sizes due to diffraction and lens problems.</p>

<p>Walter and Irash, thank you both very much for your valuable input! It gives a better understanding of digital format (compared to film) and useful information in further exploration of this relatively new format to me.</p>