View Single Post
Old Wednesday 10th December 2008, 12:36   #19
Common; sedentary.
Tannin's Avatar

Join Date: Jun 2003
Location: Ballarat, Australia
Posts: 1,559
Sorry Peter, I should have expressed that more clearly.

Take the same picture of the same bird with the same lens (poor little thing must be really wishing we would pull out those nails holding it on the perch by now) with a 40D and a 50D.

I did a fairly lengthy post on exactly this the other day in another place a bit closer to home than here - - it includes illustrations to make things clearer but I think you have to join up to see them - that's free if you want to. Anyway, I'll try to summarise.

Take those same-lens, same-everything shots and crop out the background, leaving just the bird. Let's say you do a medium-hard crop and wind up with a 4.0MP image from the 40D. The exact same crop will give you 6.0MP with the 50D image.

Now, resize the two images to any size you please - doesn't matter which size, so long as they are the same, and so long as you are not making it so small that you can't see any difference - i.e., don't take them down to the size of a postage stamp or even a postcard. (You can even up-size them if you want to, just so long as you make the final images the same size.)

The 50D image will always win. It probably has slightly more noise per-pixel (though opinions differ on that, some say it doesn't), but it also has a 50% advantage in the number of pixels you started with after cropping and before resizing. The more pixels you have to play with, the less contribution each pixel makes to the final image, and the less noise you end up with. Noise, remember, is random variaton between pixels. When you combine pixels, you reduce the noise. The 50D combines 50% more pixels to make the same (e.g.) 800 x 600 image. For the 40D to match it, it would have to have way less noise per pixel, enough less to outweigh that 33% handicap.

(50%? 33%? Yes, those are the correct figures because 6 milion is 50% more than 4 million, but 4 million is 33% less than 6 million.)

If, on the other hand, you decide to show the whole bird with the 40D image and only part of the bird with the 50D image, then of course you are going to get a different result! This is where these silly "50D has more noise" arguments come from - if you do a 100% crop on both images, you are [b]not[/i] looking at the same thing at all! You are looking at 100% of the bird on the 40D shot, and only 66% of the bird with the 50D shot and your comparison is meaningless.

Roy, we all crop as much as we have to. Some of my published shots are 100% crops - not too many, but a few. I'm not for one moment saying that a 100% crop for artistic reasons is a bad thing (though of course it's better to avoid that if you can). The point here is that if you take the same shot from the same place of the same poor little bird with the same lens, and for artistic reasons you need to take a 100% crop from the 40D image to get your finished image, you will only need to take a 66% crop from the otherwise identical 50D image. Or, if the bird is so small in frame that you need a 100% crop from the 50D, then to print it at the same size from the 40D image you need to take a "150% crop" - i.e., enlarge the 100% 40D crop from (say) 640 x 480 to something like 800 x 600! And the result, of course, will be well below the quality of the 50D result either way.
Tannin is offline  
Reply With Quote