It’s been interesting to try and test the 40d against the 5d further.  Since my original comparison shots were posted , I’ve been challenged on a few areas.  1.  I should have used a prime lens and the foot zoom method.  2.  The 70mm setting used on the 40d is not as good as the 112mm setting used on the 5d with the 70mm-200mm f/2.8 lens.  3.  The 40d images seemed soft, was perhaps something wrong with the camera.

So I’ve tested things a few more times, with the following conclusions.

1. 1. 3 may be true. 40d auto-focusing has me a little concerned because using live view at 10x as well as capturing images using both auto-focus and live view 10x, the 40d auto-focus doesn’t seem to be very accurate.  All together I tested four bodies.  Two out of the four bodies I tested didn’t do very well, and a third only marginally better.  The fourth was acceptable, but still wasn’t as sharp as I could get by using live view at 10x magnification to focus.  At this point I am assuming my original tests may not have been in focus.  (I’m not sure I can draw any major conclusions, as the subject matter and very small detail may have some impact and more normal scenes may focus just fine.)
   

2. 2.Live view focusing at 10x zoom is a terrific feature to get spot on focus when you can use a tripod.
   

3. 3.My tests were not taken with prime lenses, but were using two of the more useful and versatile lenses Canon makes.  While the prime lens tests are useful, how well the camera resolves using typical lenses is important, and considering the substantial difference in pixel pitch of the two sensors, this could be a factor.  While using the same focal length may be helpful, a couple of tests didn’t seem to indicate that would make a significant difference.
   

Unfortunately the weather hasn’t cooperated too well.  I wanted to duplicate the test under some cloudy conditions to eliminate the challenge of too much dynamic range where additional contrast may affect apparent sharpness.  If I get a day like that I may shoot the tests again, but it’s time for me to move on.  At this point I believe if you push the detail to a small enough level, the 5d still has a slight edge.  How important is that to you?  You may wish to read my article discussing this (if you dropped straight into this page via google, you can read it here ....5d vs 40d .  Also I have an overview article with images here ... “How good is the 40d?” and a small gallery with those images when using the slideshow at a decent size ... 40d Gallery .

Anyway, on with the last set of tests.  I again set up in my parking lot ... the chain length fence, small pine trees, and sign provide some very small detail and it’s easy to use the same spot each time.  This time I used the Canon 24-70mm f/2.8 lens instead of the 70-200mm f/2.8 lens to control the FoV (field of view).  This puts the 5d at 70mm, and the 40d at 44 mm.  I used a manual exposure, and adjusted the white balance and then the EV to get matching output.  Both files required almost identical adjustments to achieve the images below.  I brought them into Photoshop as smart layers and using the white sign as  a guide I aligned them so they showed about the same area.  No other adjustments were made in the RAW editor.  (One other item of note, this was using a brand new out of the box 40d body, not the one from the original test).

Here are the full images (reduced of course) to show how close the FoV matched.  I have used a small smart sharpen before and after resizing these images down to help keep them visually sharp.  As you can see, I managed to get the FoV very close.  By using the zoom instead of moving the camera, I’ve also kept the same perspective ... the shots are nearly identical.

5d at 70mm                                                                                             40d at @44mm

The fence is the focus point in this scene, and as you can see, there is a large amount of very fine detail to be resolved ... more than average or even typical.  In fact, neither camera is capable of fully resolving the detail.  This to me is the only way to decide if one is better ... push both a little past the breaking point and see how they do.  (One additional note, the 40d was a brand new out of the box camera .. with a large dust spot and hair that you can see quite clearly even in this reduced size.)

I created a photoshop file, with each of the two RAW files as a smart layer.  This allowed me to do quite a few comparisons,  here are the three most useful. All are 250x350 sections at a 1 to 1 pixel ration (100%).

First, this is pretty much straight out of the camera.  There is no sharpening or noise reduction applied at all.

5d                                                                40d

Next, this is using standard capture sharpening in the RAW converter of 25, 1.0, 25, as well as 25 luminance set in NR.  This would be somewhat typical of settings I use when processing a RAW image.

5d                                                                40d

Finally, comparison of using the RAW sharpening (capture sharpening ), and then using smart sharpening with photoshop to try and maximize each image’s sharpness.

5d                                                                40d

Judge the images for yourself.  The final pair were sharpened with different amounts ... I tried to get the best visual results I could with each.  This required a more aggressive setting on the 40d than on the 5D file.  I’m not sure if you can get away with this aggressive of sharpening on printed output in large sizes.

It’s obvious the 40d can capture a significant amount of detail, and with some work it looks good in finished output.  To me the 5d still has an edge and can still resolve the detail better - the texture in the trees is an area I tend to place quite a bit of importance on. Pixel peeping?  Perhaps, but it’s good to know what the differences are if choosing between two cameras.

(back to article)