New test results -- where's my problem?

[waple]

I've been getting, what I think, are very poor results with my equipment and I recenly saw a test where someone took a photo of money on the wall. I just took a ton of shots to test where my problem may be.

I took three shots in MF at the lens' widest, and then three AF shots. Then I stopped the lens down by two and did the same thing. After that, I went through and found the sharpest photo of each set of three and compared the MF to the AF shots. These are a few of the results.

MF is always on the left and these first 5 tests are with a Canon 40D. I had an Alien Bees B800 fire for light and to hopfully freeze any movement. All shots are at 100iso at 1/125 and on a tripod and used a 2-second timer.

First lens is a Sigma 105/2.8 Macro and you can clearly see the MF shot is much better. However, is it as sharp as it should be?



Next, a Canon 50mm/1.8 wide open:



To be fair, some shots that were stopped down did look as good or better in AF than the MF ones, like here with the same lens:



Now with the 70-200/4 IS (IS was off), the results were terrible, I think. This is the lens that has made me concerned about quality lately. Here it is open wide. I think the MF one doesn't even look good. Can someone post a similar money shot with this lens so I can see what it should look like? I don't know, maybe this IS considered good.



This next one was from 16 feet away at f8. It's a litter harder to see the difference, but the AF one is a little worse. Again, is the left shot considered sharp?



This last shot is comparing an AF shot from the 40D vs an AF shot from a 30D. Sorry I didn't get them exactly aligned right, but it's close enough to tell.



So what do you think? Do I possibly have a body problem? Does someone have similar test results to show what a dollar bill should look like?
Thanks
--
www.AllizDogStudios.com

[waple]

Alrighty, I have more pics. Same basic premise: MF on left, AF on right, all shot from the same place with the same lighting with the same settings. This is very telling, I think. All shots here are the Canon 70-200/f4 IS at f8 and 1/250 and I focused on the curl in the upper right. It was the center of the lens but I cropped and moved it over.

First, the 30D MF looks pretty darned good but the AF version... not so much:



The 40D MF again looks a lot better than the AF. Looks like it may be a lens focusing issue:



But wait, When looking at the sharpest 30D MF shot vs the sharpest 40D MF shot, the 40D is clearly not as sharp. Could this mean that the 40D itself is also bad?



And just for kicks, here are the two BEST MF shots from both cameras. Not too good.



So it looks like I probably need to send in both the 40D body (which is still under warantee at least) and the 70-200 lens. Am I correct?
--
www.AllizDogStudios.com

[E-Man]

Wow - those results show some pretty clear differences in focus accuracy. I don't have any of the lenses that you do, but it makes me want to try the same comparison with my new 24-105L to see if there are any differences between manual and auto focus.

I am also surprised by the differences in the 30D and 40D indicated in your last pairing, though the focal lengths are different.

[richardpcrowe]

The 70-200mm f/4L IS lens, although a stellar performer in most cases does have an Achilles Heel; it doesn't seem to do well at around 200mm at the closest focusing distances. I don't know if Canon has solved this problem.

The macro lens will always give you better images of a closeup flat field like the ten dollar bill than will a conventional lens. That is just the nature of the beasts!

I would definitely send in your 40D for a checkup and would probably send the 70-200mm along with it. It is best to call the service center first.

I use the Canon Service Center in Irvine, CA.

Canon USA, Inc
Factory Service Center

15955 Alton Parkway
Irvine, CA 92618
USA
949-753-4200

This is a 100% crop of an image at close to minimum focus shot using AF at 200mm with my 70-200mm f/4L IS lens. ISO 400 with a shutter speed of 1/750 second and f/6.7. It was hand-held and a very quick shot while I was photographing my dog. It is sharp dut not the knife-edge sharpness I usuallly expect from my 70-200mm f/4L IS lens. I shot the image with my 30D. I will test it today, 6/27/08, more formally using a tripod and both my 30D and 40D cameras.

Note: when shooting at 200mm using the closest focus of the 70-200mm f/4L IS lens, the depth of field is extremely narrow. At f/6.7 the depth of field will only be .03 feet or just a little over 1/3 of an inch. Using f/4, the depth of field would be only .02 feet or less than 1/4 inch. When hand holding a camera, it would be very difficult not to move the camera enough to mess up the focus.

[swampler]

OK...I just had to try the same test. Here's a shot of my (20) dollar bill...no complaints here!

30D and 70-200 f/2.8L at 200mm and f/2.8, ISO 400, handheld.

[Naturalist]

When testing lenses did you make sure that there was no filter attached

[swampler]

Doug,

Good point; however, it looks like the OP has a problem with AF, regardless of filter or not, since the MF is better than the AF. My shot was without any filter.

[E-Man]

I just did a similar set of shots with my 40D and two lenses. All photos are with a tripod and 2s timer. The first is from my 35-135 USM lens (which looks pretty bad).

The next two images are the 40D with a 24-105L. The first is autofocus, the second is manual focus. The improvement over the 35-135 is tremendous, but I'm also surprised that the manual focus turned out better than the auto focus.


Does this indicate that a calibration of the lens or the camera is necessary? Or do I need to do more comparisons between auto and manual focus with the 24-105?

Eric

[Kristy]

This is interesting to read... I've been having AF problems and am trying to decide if I am the problem or not...

Thing is that my AF does't seem to focus where I put the focus point..... And I'm not sure if it's my body or my lens.... I'm inclined to think it's the body because I get a similar result from 2 different lensed.

Anyway, I have zero faith in the Canon AF system at this point... so I enjoyed reading through these examples.... Confirms my thoughts.

[E-Man]

After some additional testing of my 40D and 24-105L lens using the PDF file at Home, I looks like I have been getting some front focus. My pictures look "nice" but something has always seemed "off" to me concerning its focusing performance.

I tested the 24-105 at each marked focal length (24, 35, 50, 70, and 105) at f4.0 with mirror lockup and 2s delay on a tripod. At each focal length, I made three exposures using auto focus, forcing the camera to auto focus again for each exposure. Then I did the same thing, but with manual focus and used the 10x display on the LCD. This provided 6 exposures for each focal length. All exposures were recorded RAW and viewed at 100%.

The results were very clear and consistent. The camera/lens combination produced front focus for each and every auto focus exposure at all focal lengths. The amount of front focus varies somewhat from about +8mm at 24mm FL gradually diminishing to about +4mm at 105mm FL. I also see some CA on vertical lines (but not on the horizonal lines) beginning at about 50mm FL (very minor), becoming more pronounced at 70mm and being very obvious at 105mm (still only on vertical lines, not on horizontal ones).

Wondering if it is the camera or the lens, I tried this again using my 50mm f1.8 II at f1.8 and again at f4.0. Same results, but no CA at all with the 50mm lens.

Two overall results:
1) On my camera and lenses, auto focus consistently produced front focused exposures by an average of about 6mm using this test chart, and

2) I was consistently able to produce a better manual focus than the camera could using auto focus. I guess this is a function of my first result.

These results aren't really surprising after looking at the many exposures I've made and doing lots of reading here and across the web. The front focus problem doesn't seem to be "way" off, but its off by enough for me to notice. Its too bad that I'm taking the camera on vacation with me in 2 weeks, or I'd send both the 24-105 lens and the body in for calibration right away. The 50mm is old enough and inexpensive enough that I'm not going to bother doing anything with it, and given the consistency of the results across lenses, it may very well be the body that needs a little tweaking.

Just thought I'd share my results.
Eric

[E-Man]

Originally Posted by Kristy

Anyway, I have zero faith in the Canon AF system at this point... so I enjoyed reading through these examples.... Confirms my thoughts.

While my first reaction was similar to Kristy's, I've been thinking about this a little bit more lately. Essentially, I want to make a comparison between my 40D/24-105L combination and some scientific laboratory equipment that I used to use in a former life. In the lab, we had scales (formally called a balance) that were capable of weighing substances with resolution and accuracy to 0.000001 grams (for example, one of the hairs from my head weighed 0.002470 grams). These scales were expensive, high-precision pieces of equipment.

While the price tags of these two devices (high-precision scale and high-resolution camera) are not directly comparable, the same principle applies. That expensive and high-precision scale was useless until it was CALIBRATED with a known set of CERTIFIED WEIGHTS. I remember the calibration guys visiting us every month with a set of weights: 1g, 5g, 10g, 50g, 100g, 300g and probably a few others. Each weight was placed on the scale and adjustments were made until it produced accurate results across its operating range (much like calibrating a zoom lens at each of its marked focal lengths). This process happened about once per month for the scales (they received frequent use). Expecting this level of performance from the scale "out of the box" is unrealistic in any setting.

I would suspect many musicians would make similar claims about their instruments.

Based upon what I have read and learned, I suspect that my camera is capable of some really stunning performance, but first needs to be calibrated for optimum performance.

If I wasn't planning to take it on a family vacation soon, I'd have sent both the camera and the lens in already...

Eric

[JDArt]

These are extremely interesting tests you've done. At the very least I would be sending the results to Canon, along with the camera, and asking that the autofocus be adjusted so that it performs on a par with the manual focus. You may even have to send in the lenses.

One ancillary question I have is whether the micro-adjustment features of the latest 1-series cameras would really allow you to perform such mico adjustments yourself.

[Danlphot]

Originally Posted by Kristy

Thing is that my AF does't seem to focus where I put the focus point..... And I'm not sure if it's my body or my lens.... I'm inclined to think it's the body because I get a similar result from 2 different lensed.

I HATE using a camera with multiple focus points. Just set it up to use the center focus point and see if you are happier with the way it works.

Dan

[E-Man]

Originally Posted by JDArt

These are extremely interesting tests you've done. At the very least I would be sending the results to Canon, along with the camera, and asking that the autofocus be adjusted so that it performs on a par with the manual focus. You may even have to send in the lenses.

One ancillary question I have is whether the micro-adjustment features of the latest 1-series cameras would really allow you to perform such mico adjustments yourself.

This is what I was thinking, too. I'll prepare a CD with the test images for them to see, but I think I'll leave out the 50mm lens results and have them put both the 24-105 and the 40D through their paces for calibration. I'll just tell them that I don't know whether its the camera or the lens that is causing the trouble.

I would think that adding the micro adjustments to the 1-series bodies would eliminate a great deal of work for the cust service center for calibration. While I haven't read the specifics of this feature, if would be great if it could remember settings for 4 or 5 different lenses.

[Max@Home]

Originally Posted by Danlphot

I HATE using a camera with multiple focus points. Just set it up to use the center focus point and see if you are happier with the way it works.

...but in that case do NOT FocusLock-Recompose at small DoF, as then AF errors will be more rule than exception...

...€0.02...

Kindest regards!

Max@Home