I was pondering the following situation..


Lets say two photographers photograph the same subject at the same distance, using the same telephoto lens (say a 400mm lens), at base ISO.

One is using a cropped-sensor camera (like the D300 or canon D40), and the other is using a full-frame camera.

If the full-frame user crops his/her shot to look the same as the cropped-sensor shot, and they both print at a large size, who will have a better quality print?

I heard some people say that cropped sensor cameras like the D300 have an advantage over full-frame, because a 400mm lens turns into a 600mm equivalent..which makes sense.

But the sensor is also smaller, so does that negate the advantage? in addition, some full-frame cameras have more pixels, so maybe their cropped photos are equivalent?

thanks for your replies... I'm hoping to learn something here.

scott

As usual, it depends. If you're comparing a really top-notch full frame high resolution sensor (aka Eos-1Ds mk III) against a smaller sensor with a very small pixel size and poor noise characteristics, the actual image quality from a cropped print taken with the full frame camera will probably be a bit better. On the other hand if you're comparing a cropped image from the original Eos-1Ds with something like an Eos-40D or Nikon's D300, the smaller (but newer) sensor will definitely deliver better image quality once you crop to the same image size. Since you're using a telephoto lens, it'll be wide open, so the diffraction problems you get with smaller apertures at small pixel pitches won't be an issue.

I believe Popular Photography magazine actually checked this out, comparing the same lens on an Eos-1Ds II and an Eos-20D. Their conclusion was that in this case there was a slight advantage for the 20D once you cropped the 1Ds image to the same subject size.

>> That can only be answered by examining sensors. How many active pixels are capturing the image?

>> Of course, that's not the whole story, because all pixels are not equal - but it's a starting point.

>>Bill

yes, if you only looked at pixel count, you would think you would need a 600mm lens with a nikon D3 to get the same print quality as with a 400mm lens on a nikon D300.. but like you said, all pixels are not equal, hence my confusion on the issue!


>> As usual, it depends.

thanks Geoff.. I guess in my mind I was comparing comparable generation full-frame cameras vs their cropped-sensor "equivalents". and to make matters easy, I thought we would not worry about diffraction issues, although if I remember right, diffraction will be worse on the cropped sensor, correct?

[quote=Geoff Wittig,May 6 2008, 05:27 PM]

" On the other hand if you're comparing a cropped image from the original Eos-1Ds with something like an Eos-40D or Nikon's D300, the smaller (but newer) sensor will definitely deliver better image quality once you crop to the same image size. Since you're using a telephoto lens, it'll be wide open, so the diffraction problems you get with smaller apertures at small pixel pitches won't be an issue. "

There is some confusion here...you talk of cropping to the same "image size". There is no difference in image size whatsoever unless you change the "taking" distance.

If you take a picture of a robin at x feet distance with a 500 mm lens, the image will be, say, 1 inch long on the sensor. It doesn't matter what size the sensor or the camera is - the image will always be 1 inch long. Thus my statement that the main thing that matters is how many pixels the image is taking up.

If you take advantage of a smaller sensor to take the picture from further away, so that the robin's image takes up the same amount of viewfinder space as the bigger sensor does from a closer distance, then you will no longer have a 1 inch image. It will be smaller and require greater enlargement (less quality) than the closer image.

Only if the smaller sensor has more pixels of equal quality PER INCH than the larger sensor, will you have a better result at any given image size. (Ignoring diffraction issues which are not likely a big factor unless you shoot smaller than about f11.)

Hope this helps rather than confuses!

Bill

thanks Bill, I think your answer is pretty straightforward.

if you assume a cropped-sensor camera (D300, canon D40) fills the frame with a robin using a 400mm lens, it will occupy 1 inch of the sensor...it will also occupy 1 inch of the 36mm sensor of a full frame camera (nikon d3, canon 5d), that makes perfect sense

so assumming the pixels are "equivalent" between these 4 cameras, the D300/D40 should produce a better image in this particular situation than a D3 or 5d, since there are many more pixels in that 1 inch area.

but we know that the pixels aren't exactly equivalent - for example full frame sensors are supposed to much less noisy than cropped sensor pixels, but at base ISO this should be less of an issue. but do the pixels store the same amount of information when translated into a raw file ? will they look better when viewed at 100% crop due to greater color detail, dynamic range, etc.? are "upsized" images equivalent? these are just random ponderings....

up to say ISO 800 i don't think you'll notice a difference in the pixels

the other consideration is lens - with a marginal lens like the Canon 100-400, the crop frame has a distinct advantage at long focal lengths

D300 has 178 pixels per mm, that works out to 27 MP for a 35FF sensor; which none have.

The argument comes down to: are more pixels better? At lower ISOs the answer is generally yes.

I believe the 20D has 3504 pixels across, and the cropped EOS-1Ds II image would have 4992 / 1.6 = 3120 pixels across, so it makes sense that the 20D would have a slight advantage.

thanks for going through this exercise with me everyone.

so it appears that -

At lower ISO's and larger apertures it's just a matter of counting pixels. at very high ISO's you may see more noise in the cropped-sensor image due to a higher pixel density. And at smaller apertures diffraction may cause more sharpness loss in the cropped-sensor images.

hey stever,

In this scenario we are shooting with the same lens at the same distance, and cropping the full-frame image down to look the same as the cropped-sensor image.

So in both cases we are not seeing the corners/edges of the 100-400mm lens, so I don't think the quality of the lens is a consideration, right?

scott

Of course you could compare two cropped sensor cameras with the same sized photosites and find slight differences. It's not a question of format since you are proposing cropping the larger format to match the smaller format; it's a matter of the performance of the camera, and in that regard newer generation cameras will outperform older generation cameras and deliver similar noise but with more resolution.

With digital photography we are not enlarging negatives, we are enlarging pixels -- the difference being that unlike a negative, there is no fixed size to a pixel (not to be confused with a photosite, which does have a fixed size). What trips up pixel peepers is that while there may be more noise in a pixel made up of smaller photosites, because there are more of those pixels than those made up from larger photosites they are not being enlarged as much and that means the noise will not be enlarged as much. As an example, my D70 has about the same amount of noise per pixel as my D300 (this varies depending on the lighting); but I have to enlarge those D70 pixels considerably more than those D300 pixels -- and that makes the noise in the D70 pixels more noticeable in the final image.

As for the diffraction, there are two unrelated issues to consider. The first issue is that the circle of confusion becomes more visible the more it is enlarged; this is a function of format, but if you crop the larger format to match the smaller format then you are effectively comparing the same format and not two distinct formats. The second issue is that as the Airy Disk covers more photosites it causes a loss of acuity; this is a function of how close the photosites are to one another, so the higher the resolution of a given format the sooner diffraction impairs that resolution. The most densely packed photosites (90 lppm) become diffraction impaired at about f/11.