Thanks gml1 for dealing with facts, rather than opinions and speculation, and for taking the time to share. FYI, the hierarchy of your results are the same as mine, at least as far as I can see, even though you used a totally different setup. So there is some consistency.
And so much for the theory that says that greater resolution not visible.

Yeah, you guys are right.

I'll take this 1 pixel image...


Over this 54 pixel image, any day of the week!

gml1 wrote:
Here's a 150% crop from the 50D image compared to the uprezed versions of the same image from the 5D (top-right), 450D (bottom-left), and the 40D (bottom-right).
This is how the differences will look like on 75x50 inch print @ 96ppi:
http://i.pbase.com/o6/17/716817/1/104037501.th7Ri0Bo.1.bmp


You are obviously a fool who refuses to admit you are wrong.

The 18-200 IS lens OUTRESOLVES the 50D at the center between 18 and 24 mm :
see here. And this comes from a reliable lens tester.

So, there goes your MYTH about sensor outperforming lens resolution.

Note that the poor edge performance of the 18-200 (expectedly) is mostly due to the 10x zoom range. Looks like the Canon 18-200 IS is behaving like the Nikkor 18-200 VR.

Me thinks Imaging Resource uses the 50mm 2.5 macro (at least according to the exif, which also says 70mm FL -?-) for all of their still life shots (regardless of what Canon camera is tested). Every time the same lens for the same test shots (as far as I can tell).

The 50D house shots were shot with a 24-105L @ 28mm (according to the exif).

BTW When you click on the exif info under the images, you are presented with a list of all the applied in-camera settings.

Both the 50mm 2.5 macro and the 24-105L aren't particularly bad lenses. But I wouldn't want to conclude anything based on these comparisons (yet).

Besides that, differences between 12MP and 15MP images are very small (if any at all) in small to medium sized prints. You have to print REALLY LARGE to see a distinction here. But even when pixel-peeping at 100% I doubt there will be a BIG difference visible.

EMC 2 wrote:
davenfl wrote:
Much as I like Imaging Resource this test proves absolutely nothing and most certainly does not support your conclusions. Who started this nonsense about the current lens being outresolved by a 15mp 1.6 sensor. There simply is no such thing in reality, please.

Agreed 100%

Agreed 100% again!

dcains wrote:...Anyone know the wavelength of visible light?

Yes:

380-420 nm — violet light
420-440 nm — indigo light
440-500 nm — blue light
500-520 nm — cyan light
520-565 nm — green light
565-590 nm — yellow light
590-625 nm — orange light
625-740 nm — red light

Note that 740nm is much smaller than even the smallest, most densly packed pixel on any DSLR camera.

Maybe we need a working definition of what "out-resolving a lens" should mean.

When I put a 17-85mm on a 10D using ± 20mm fl at f/8 photographing something say 50cm away I will get most of the frame in quite sharp focus.
Do the same using a 40D, and from about 50% and out of center the image becomes very soft.

DOF is equal, so it is not that...
Lens is the same, so not that...
Focus done manually, so not AF...

So why the differences in IQ and attaining sharpness and image detail?

thw2 wrote:
You are obviously a fool who refuses to admit you are wrong.


More of the same: here's a 100% crop of the 50D (middle) vs upscaled crops of the 40D (left) and 450D (right).
These crops are equivalent to looking at 48"x32" @ 96 ppi prints.

EltonTeng wrote:
Shouldn't we be viewing these 150% crops from like 15-20 (or pick a number) feet away?

Tough room. We're comparing 150% crops now.

See above for 100% crops.
Btw, the bigger the distance, the smaller the differences that the (human) eye can see.

gml1 wrote:
Again, look at the difference between the 50D (15mp) and the 450D (12) - well, there is none.
Both have slightly more detail than the 40D but the point here is that beyond 12mp on a CROP camera, you only get empty magnification (plus noise) but no more detail.
Remember, these crops are equivalent to 48"x32" @ 96 ppi prints.


I don't know what lens you used but "PRODUCT OF ENGLAND" at the bottom of the image appears noticeable sharper. All the finer prints are in fact. You have to remember the jump from 12 to 15 is a very small gain.

gml1 wrote: Look at the difference between the 50D (15mp) center and the 450D (12mp) on the right - well, there is no difference.

The point here is that beyond 12mp on a CROP camera, you only get empty magnification but no more detail.

I don't know what screen you're using, but to MY eye, almost every instance of type (words) on the center crop is clearer (has more detail) than in the right crop. (Compare, for example, the center-vs.right views of any words below the bottom of the beer glass.)

Looks like 12mp is not in fact "outresolved by" the lens used, at least not in this case.

I'll take the 15mp camera, thanks, and you can take the 12mp camera!

gml1 wrote:
EltonTeng wrote:
Shouldn't we be viewing these 150% crops from like 15-20 (or pick a number) feet away?

Tough room. We're comparing 150% crops now.

See above for 100% crops.
Btw, the bigger the distance, the smaller the differences that the (human) eye can see.


You need to get your eyes checked because I can see a difference from your examples. The 50D shows more resolution than the 40D or XSi. It is easily seen in the finer detail.

Jim

Jonathan Wong wrote: "PRODUCT OF ENGLAND" at the bottom of the image appears noticeable sharper. All the finer print is, in fact. You have to remember the jump from 12 to 15 is a very small gain.

I agree. If the three crops above "prove" anything, it's that

a) There is indeed a perceptible difference going from 10 to 12 to 15 megapixels, and

b) as might be expected, the differences are harder to see as the percentage increase in sensor size gets smaller.

I'll believe that a sensor is being "outresolved" by the lens when I can't tell the difference between output of two sensors when one is twice as large as the other, not merely slightly larger as 15mp vs. 12mp is.

RalphJ wrote:
Jonathan Wong wrote: "PRODUCT OF ENGLAND" at the bottom of the image appears noticeable sharper. All the finer print is, in fact. You have to remember the jump from 12 to 15 is a very small gain.

I agree. If the three crops above "prove" anything, it's that

a) There is indeed a perceptible difference going from 10 to 12 to 15 megapixels, and

b) as might be expected, the differences are harder to see as the percentage increase in sensor size gets smaller.

I'll believe that a sensor is being "outresolved" by the lens when I can't tell the difference between output of two sensors when one is twice as large as the other, not merely slightly larger as 15mp vs. 12mp is.


Of course we are talking pixel peeping galore here... 100% crops... These differences will be VERY HARD to detect even in bigger prints. Still, they are there

Regarding the comparative shots of the lager bottle I'm not going to join in the bun fight about which is sharper. Thing that strikes me is how damn good the Rebel image is. Perhaps when I retire my 40D I should have a good look at the 500D or whatever Rebel is current at that time.

On a per dollar basis the Rebel look remarkably fine to my eyes!

David Baldwin wrote:
Regarding the comparative shots of the lager bottle I'm not going to join in the bun fight about which is sharper. Thing that strikes me is how damn good the Rebel image is. Perhaps when I retire my 40D I should have a good look at the 500D or whatever Rebel is current at that time.

On a per dollar basis the Rebel look remarkably fine to my eyes!

I'm known as a massive 450D fan and I'm not surprised by the results at all. If you want the 50D's features then fine, but for image quality and particularly for value, the 450D/Xsi is just stunning

David

RalphJ wrote:
gml1 wrote: Look at the difference between the 50D (15mp) center and the 450D (12mp) on the right - well, there is no difference.

The point here is that beyond 12mp on a CROP camera, you only get empty magnification but no more detail.

I don't know what screen you're using, but to MY eye, almost every instance of type (words) on the center crop is clearer (has more detail) than in the right crop. (Compare, for example, the center-vs.right views of any words below the bottom of the beer glass.)

Looks like 12mp is not in fact "outresolved by" the lens used, at least not in this case.

I'll take the 15mp camera, thanks, and you can take the 12mp camera!



The 40D and 450D have to be uprezzed more than the 50D, which will penalise them more. Given this fact, the 50D crop is only barely sharper, but appears to resolve no more detail than the 450D especially and is delivering bugger all improvement over the 450D at least.

Pixel Perfect wrote: The 50D crop is only barely sharper, but appears to resolve no more detail than the 450D especially and is delivering bugger all improvement over the 450D at least.

Is "bugger all improvement" an Aussie idiom?

I do heartily agree with whichever poster above said the 450D performance is impressive: A few months ago I was really torn about whether to buy a 40D or a 450D. In the end I decided that the 40D felt much better to hold; the grip was just too small on the 450D. So I bought a 40D even though at the time the Rebel IQ (at least at lower ISOs) was arguably better than the xxD-series IQ.

gml1 wrote:
Again, look at the difference between the 50D (15mp) and the 450D (12) - well, there is none.
Both have slightly more detail than the 40D but the point here is that beyond 12mp on a CROP camera, you only get empty magnification (plus noise) but no more detail.

That's only because the 450D has a lighter AA filter than the 50D.

It has NOTHING whatsoever to do with lens vs sensor resolution. If an ultrazoom lens like the 18-200 can outresolve the 50D, it tells you we have yet to reach the limit.

The XS/1000D has as good resolution as the XSi/450D (go download some RAW files from Imaging Resource). Does this mean that there is no point in going beyond 10 MP on a crop camera? Perhaps, you should just stick to your Nikon D200/D80/D60 camera then.

gml1 wrote:


More of the same: here's a 100% crop of the 50D (middle) vs upscaled crops of the 40D (left) and 450D (right).
These crops are equivalent to looking at 48"x32" @ 96 ppi prints.

Heh! Out resolving a lens is not a bad thing. Having more pixels to work with is good too. I like it.

I cropped the Mas Portell bottle tops of each and did a common work flow I do and found the 50D superior for what I do.
I re-sized them equally in small steps and the 5D text in the cap started with the halos immediately. As I continued it just got worse as the others held on very well. I suspect some sharpening on the 5D shot rather it be the OP or the in-cam. Perhaps I'm wrong.

And it may all be due the the test shot methods of the OP also. Different focal lengths and metering modes to name a couple. Definitely not a controlled laboratory test.

This is a tough crowd for sure. When you step up to the plate here you better be on your best game. I am not one of the heavy hitters for sure.

Nothing I have read anywhere so far had deterred me from getting one.

Daan B wrote:
Me thinks Imaging Resource uses the 50mm 2.5 macro (at least according to the exif, which also says 70mm FL -?-) for all of their still life shots (regardless of what Canon camera is tested). Every time the same lens for the same test shots (as far as I can tell).

The 50D house shots were shot with a 24-105L @ 28mm (according to the exif).

BTW When you click on the exif info under the images, you are presented with a list of all the applied in-camera settings.

Both the 50mm 2.5 macro and the 24-105L aren't particularly bad lenses. But I wouldn't want to conclude anything based on these comparisons (yet).

Besides that, differences between 12MP and 15MP images are very small (if any at all) in small to medium sized prints. You have to print REALLY LARGE to see a distinction here. But even when pixel-peeping at 100% I doubt there will be a BIG difference visible.

The 50 macro is one of canon's sharpest lenses, the 24-105 doesn't even come close to it's performance. If the 50d had the 50 macro, I'm betting you would see more of a difference.

gml1 wrote:
Here's a 150% crop from the 50D image compared to the uprezed versions of the same image from the 5D (top-right), 450D (bottom-left), and the 40D (bottom-right).

This is how the differences will look like on 75x50 inch print @ 96ppi:
http://i.pbase.com/o6/17/716817/1/104037501.th7Ri0Bo.1.bmp


The way I judged the images:
5D = best; 50D not worth the upgrade from 450D; 40D = worst.

dhphoto wrote:
If you want the 50D's features then fine, but for image quality and particularly for value, the 450D/Xsi is just stunning

David

Yeah, too bad the AF in AI Servo on the Rebel is so dismal. Kept mine for 1 week before selling it.

I will take the superior AF of the 50D along with all the other better features any day.