Because of pixel crowding, is the 50D as sharp as the 40D?

I was told about a report in the NY Times saying the 40 D was sharper and the extra pixels in the 50 D are for most purposes not necessary

I do not understand how the reasoning goes that there could be pixel crowding on the 50D as regards the 40D since each one of them has a different size cmos areas in their respective cells. You will have more pixels due to the larger cmos but that's it. As a matter of fact some reports of reviewers state that the 50D's images are NOT sharper than either those of the 40D or the Rebel XSi

Thanks, but it look like if you're after sharpness, the 40 d is a better bet.

George,
Thank you for such a great exposition and explanation of the comparative values between the 40D and the 50D. Great in knowledge and very good in explaining the technical facts.
You have reaffirmed my decision to buy this camera, as I have done so this very morning. I got it with the 18 200 IS and hope that it will be better than just a kit lens, but then it has already strained my budget.
Keep up the good work!

Boris

Thanks. Best report I've seen

"For a true comparison of perceived sharpness and overall noise you need the same portion of the lens in each view, if the 40D image is at 100% the 50D image needs to be at 66.66%. Then you are comparing the output from the same place in the lens."

Since both the 40D and 50D use APS-C sensors, both use the same portion of the lens (assuming use of a full-frame EF or equivalent lens). The 50D sensor is MARGINALLY larger - about 1mm taller - but not enough to significantly affect the portion of the lens that is used. Both sensors record their images from the center 60% or so of a full frame lens. Reducing the 50D image to 66% therefor has nothing to do with the part of the lens being used, only with trying to match the pixel density of the 40D. Viewing both images at 100% is the appropriate way to assess image quality, at least as regards the lens.

The 50D is generally regarded as being equal to the 40D, but not superior, in terms of image quality. In print sizes above about 12"by 18", the added pixel count may produce a higher perceived image quality in terms of detail and clarity, but the 50D offers only a 22% increase in linear resolution, not the 50% one might think from raw pixel count. While that 22% is helpful for cropping, and can help with larger print sizes, it comes at a cost. Those extra pixels add heat, which adds noise. Canon managed to cram half again as many pixels in the same size sensor by eliminating the gaps between the microlenses over each photosite, but the increase in pixel density means more must be done to keep noise under control. For most purposes, the 50D does that well, but the image quality still isn't better... just bigger.

Many thanks. These review have helped make a decision on what is probably the last camera I buy.

Rocky, I appreciate your efforts, but I am sad to think you actually believe everything you wrote above, spend some time and do the math while drawing things on paper and you will identify your errors. You have obviously been affected by the internet misinformation which defies the laws of physics and geometry. Let me help correct the brainwashing you have unknowingly accepted.

First I congratulate you on the two statements where you nailed it.

1) "22% increase in linear resolution". YES! Since we are increasing pixels in both X and Y directions concurrently, the total pixel percentage is the sum of the increase in both axes.

2) "Both sensors record their images from the center 60% or so of a full frame lens." Yes!

Full frame is 36mm x 24mm
40D is 22.2mm x 14.8mm, or 65.3% of FF
50D is 22.3mm x 14.9mm, or 65.6% of FF (0.1mm or about 0.5% larger than 40D)
Essentially equal overall for the portion of a full frame lens that each sensor "sees through".

That's about it, after making statement #2 above you went straight into this bogus claim...
"Reducing the 50D image to 66% therefor has nothing to do with the part of the lens being used".
WRONG!!! It has EVERYTHING to do with where those pixels are looking through the lens. Keep reading for some simple geometry

Quickly followed with the common misconception I was attempting to educate people on "Viewing both images at 100% is the appropriate way to assess image quality, at least as regards the lens."
WRONG AGAIN!!! This is the logical progression from the earlier statement, since the previous was bogus so is this one.

Here is why, its really easy if you REALLY think about it.
Lets assume we are using a Full frame lens that projects an image size of 36mm on a full frame chip, its actually larger but for easy math sake ignore the rectangular angle shape and just look at the X axis.

Remember..
40D sensor produces images of 3888x2592 pixels, (175 linear pixels per mm of image on the chip).
50D sensor produces images of 4752x3168 pixels, (213 linear pixels per mm of image on the chip).

You put the same lens on each camera mounted on a tripod looking at the same center point. You have a decent computer with a widescreen 22" 1680x1050, split the screen in half and at 100% view 840 pixels of each photo at the same time.
Do the math..
40D) 840 pixels / 175 pixels per mm, you are looking at an image from 4.8mm of the sensor, that's 21.6% of the entire photo/lens view.
50D) 840 pixels / 213 pixels per mm, you are looking at an image from 3.94mm of the sensor, that's only 17.7% of the entire photo/lens view. To see the same image view for a true quality comparison, which is from the same 21.6%/4.8mm center portion of the lens, you MUST display 1023 pixels of the 50D image, not 840. Otherwise as I said above you are comparing apples to oranges.

Looking at something artificially blown up 22% larger while comparing it to something that isn't, then saying it doesn't look better, is just plain stupid. There is plenty of that around the world and right here on the good ol' net.

Now I ask you, when you compare the photos you've taken with YOUR 50D (or one your borrowed) to the ones taken with YOUR 40D using the same lens, how do they compare? My 50D takes better quality photos than either the two 40D's I have owned, equal or better than most of the photos I got from my 5D last year. Go ahead live in a cave, my 50D can take pretty good photos in there also!

Thanks again. An interesting debate stirred by a simple question.

I just got the 50D and its a wonderful camera... does it look as sharp at 100% as my 20D... not really, but thats not the fault of the camera. This is the highest pixel density camera released by canon do date and thus it is merciless with lenses... it take faults that weren't resolved by previous cameras and makes them obvious @ 100%. This just means that the manufacturer's are going to have to up their game in the optics department to keep up with their body. Just look at the photozone.de reviews for canon and compare the reviews done with the 50D and 8mp rebel of the same lenses. The lenses cant keep up in resolution on the 50D and any residual CA is magnified since the physical size of the purple fringing on the sensor doesn't change (but there are now more pixels in that same area to resolve it)... you buy the 50D then be warned... you'll be looking at buying some nice glass too... oh, and this camera will push you as a photographer... camera shake will more easily ruin photos (smaller amount of movement is resolved at 100%) and focus is going to be super critical... probably the reason canon broke down and finally put the micro focus adjustment into a non-1D camera... otherwise they'd have to recalibrate a lot of 50Ds and lenses. All this isn't bad though... just means this camera will push you to be a better photographer and will definetely reward you when you get it right =)

EOS 50D, Resolution and Confusion BOB ATKINS

Sometimes a statement taken in isolation can be the truth, but it's not the whole truth, and in fact in isolation it can be misleading. For example... "Stepping out of a plane flying at 5000ft won't hurt you". That's pretty much true, but it doesn't address the issue of what happens a short time later when you hit the ground at over 100mph. That will hurt you!

What does that have to do with photography you might ask? Well, while reading a number of photography forums I've seen more than just a few participants who have been confused by some recent technical reviews of the EOS 50D. They've picked up on statements made in the reviews which are true, but which then lead them down a path of faulty reasoning to an incorrect conclusion. Let's take a look at a couple of those statements.


#1 - With the 50D, if you stop a lens down past f8, you'll see a decrease in resolution
That certainly is a true statement with many (if not most lenses). I wouldn't dispute it for one moment. However some readers seem to have have taken this statement and concluded (or assumed) that (a) Other cameras don't do that and (b) If you want to shoot landscapes at f16, then the EOS 50D wouldn't be a good camera to chose. Both are incorrect and neither follows from the original statement.

The fact is that most lenses peak in resolution in the center of the frame around f8, and that then leads to maximum image sharpness. Some might peak at f5.6, some might peak at f11 but it's more often f8. The reason is quite simple. As you stop down, you quickly reduce the amount of spherical aberration a lens exhibits, and spherical aberration is the dominant aberration affecting sharpness in the center of the field, so as you stop down and spherical aberration diminishes, sharpness increases. However, as you stop down something else happens - diffraction increases and resolution drops. It just works out that around f8 there's a point when the combined effects of spherical aberration decreasing and diffraction increasing are at a minimum and therefore sharpness is at a maximum. It could be f5.6, it could be f11, depending on the lens, but it's somewhere in that range and the difference between f5.6, f8 and f11 usually isn't very big.




Since the lens is sharpest at f8, the image recorded by a digital sensor is also sharpest at f8, so stopping down past f8 lowers image resolution with just about all DSLRs. I suppose if the sensor resolution was low enough, it would totally dominate the final image resolution, so such a camera might not show much change until you got to f16 or even f22, but the point there would be that the resolution would be low at all apertures, and at all apertures it would be worse than with a higher resolution sensor. You aren't really concerned about where resolution peaks, but how high the peak is and how high the resolution is at any given aperture. In all cases the higher resolution sensor will yield sharper images.

So, yes, stopping down past f8 with the EOS 50D might well result in lower image resolution. The important omission is that stopping down an EOS 40D, a Digital Rebel XT, a Nikon D300 or a Nikon D90 past f8 would result in exactly the same thing! Not only that, but the camera with the highest resolution would be sharper at f8. So, for example, the 50D image would be sharper than the 40D image.

Stopping down from f8 to f16 when you need depth of field for a landscape is fine. You will lose a small amount of sharpness in the sharpest regions of the image (see above), but you may more than make up for it with increased sharpness of the area within the now extended depth of field. The tradeoff between maximum sharpness at the focus point and extended depth of field is one you have to make with all cameras, and for all cameras of the same format size (crop sensor or full frame), the tradeoff is exactly the same. The ultimate tradeoff is a pinhole camera where the depth of field is huge - but the image is blurred everywhere!



#2 - You need the best lenses to take advantage of the higher resolution of the 50D over the 40D
So here's another statement that has some truth in it but has lead to some confusion. The incorrect extrapolation that some people make is to then suggest that unless you have the best lenses, you won't see the higher resolution of the EOS 50D, so unless you have a bag of "L" series prime lenses, the 40D will be just as good.

Again that thinking is wrong. I've even seen some people suggest that images from the 50D will be less sharp than those of the 40D with most lenses. I'm not quite sure of the "logic" they have used in reaching that conclusion, but they got there somehow!

Again these "deductions" are completely wrong. The fact is that the higher resolution of the 50D will result in higher resolution, sharper, images than those from the 40D whatever lens you use. Doesn't matter if it's the pretty average "all plastic kit zoom" shot wide open or a super sharp lens like the EF 135/2.0L shot at f8. The higher resolution sensor of the EOS 50D will result in sharper images in both cases. See the 100% image crops above for a concrete example that proves the point.

If the lens was REALLY bad, such as a pinhole lens or a single element plastic lens, or a regular lens stopped down to f64 (where diffraction would drastically blur the image), you might not see any difference between EOS 40D and EOS 50D sharpness, but the 50D certainly would not be less sharp.

I suppose you could argue that the difference between a "normal" and a "premium" lens might be very slightly greater with a higher resolution sensor, but that certainly does not mean that you don't still see most of that difference with "normal" lenses.

So is this the truth. the whole truth and nothing but the truth, without any room for confusion? No, it's not! The above images and analysis use fairly high contrast black and white resolution test patterns. In the real world, detail isn't usually black and white. It may be low contrast green and red, or light grey and dark grey. However, the higher resolution sensor will still yield results which are at least as good as, and usually better than, a lower resolution sensor. There may indeed be cases where you would need a good lens to exploit the added resolution of the 50D sensor when dealing with low contrast detail - but that certainly does not imply or suggest that there won't also be plenty of cases where the higher resolution of the 50D will be an advantage when using ordinary consumer zooms. In fact most of the time that will probably be the case. So while it's always good to have a better lens, you don't absolutely need a better lens to see the added resolution of the EOS 50D over the EOS 40D, and the 50D image should never be worse.


Conclusion
You have to read highly technical reviews for what they actually say, not for what you think they are trying to say. If you're not conversant with all the factors that are involved in lens and image sharpness (and how the two interact), then be careful of reading things into reviews that aren't really there.

Thanks to all. These comments should be condensed into an article with diagrams of lenses, apertures, pixels and coverage. Diagrams should show how the variables add up to, or detract from, resolution, which is what we look for in a newly purchased camera and lens.

Way to go Marcus! Excellent synopsis, all the stuff I wanted to say but didn't take the time. Welcome to the forum.
I just gave you a thumbs up in both entries...