Does Canon 20D have a focusing problem

[Jonathan Wienke]

Your methodology is flawed. If your camera had a focus calibration problem, focus would be consistently in front or back of where you wanted it. The aperture selected is totally irrelevant, it has no effect on AF since AF is always done with the lens wide open and the lens is only stopped down to the exposure aperture during exposure. AF precision with any camera will vary widely depending on the lens, lighting, and subject matter. Given the lens you mentioned only has a f/5.6 aperture at 400mm, the lighting was probably insufficient for AF to lock on as accurately as you'd like. Try a different lens with a f/2.8 aperture in the same lighting, or go outside and retest in sunlight, and I bet the results will be quite different.

[Gary_Berg]

I think you will find that that much variance is within the "specs" of the 20D. Despite how nice it works, the 20D is not a "Pro" camera, and doesn't focus quite that flawlessly.

The fact that you are seeing variance both behind and in front tends to indicate that it is due to the level of accuracy of the focusing mechanism, not the lens itself. The 20D is repeatable in focus only to a certain level.

If the Canon rep will tune the camera for you, I'd take him up on it. I just don't think you are really going to see that much difference.

Now if you are seeing problems on landscapes, the first step is to try using a tripod for all your landscapes, and make sure your shutter speeds aren't so slow as to allow subject movement.

[matt4626]

No

[kevin.illingworth]

Could you show a real life example of a Landscape shot where this variation in focus has caused a problem together with the EXIF details.

Regards Kevin

[Terry]

I purchased my 20D in September and all went OK until I began using the Canon 600mm F4 IS lens. The camera constantly front focused using both the AF or MF mode. My 100-400 IS appears to be fine.

By accident last week I was talking to a press photographer friend who, having purchased a 1d Mk11 had the same problem with all his long lens when using his camera. After a lot of discussion with Canon UK, all the equipment was returned and after several false starts, the gear was "re-calibrated" and now works fine.

Intersestingly, when a friend tried his 600mm IS on my 20D camera the focus problems did not occur. However, when I then used my lens on his 20D camera the same problem occured. This appeared to indicate that the lens, not the camera was at fault.

I was sceptical, but contacted Canon UK and arranged to drop of my camera and lens and wait in the car park after a 500 mile  round trip to get the gear sorted out prior to leaving for Poland on a wildlife photo shoot this coming Saturday. The re-calibration worked out perfect and the two items now function together as they should. I was informed whilst at Canon, the company may have to consider re-designing the lens mount on a number of their digital cameras to resolve this unusual problem. I have yet to try my 600mm lens on a 35mm film camera body to establish if the lens still functions correctly with a SLR??

To check the problem with the 20D, or any other Canon DSLR is easy, just photograph a calibration charts place at 45 degrees to the horizonatal camera plane, whilst setting the camera apperature wide open, in my case f4, from the lens minimum focal distance, focus on the centre point of the chart. If the camera is "front fucusing" as mine was, the lines marked +1,2,3, (closest to the lens) at the near side of the charts centre spot will be sharp.

If the image shows that between +1 / -1,-2,-3 are sharp, your equipment is OK and does not require calibration. If you try this out use a high shutter speed,

Terry

[moh@sbcglobal.net]

I recently purchased a canon 20D. I noticed what appeared to be a focusing problem with my landscape pictures.

I then conducted the following test which showed a problem with the Auto focus. This was also verified by the pro-photo store I frequent and also Canon rep. Canon rep offered to "tune" my camera and lens!

The focusing seems to be off by 1-2" in front or back depending on the F-stop.

Could someone tell me if I am doing something wrong.

Test condition:

Body: Canon 20D
Lens: Canon 100-400L @ 400
Object: Flat Metal ruler, laid on the ground
Distance : about 8 ft camera to ruler. Camera hight about 5 feet.
Focusing point: center focus only
IS: off
Tripod: Yes
F-stop: various
ISO: various

[moh@sbcglobal.net]

Measurement was done with plenty of daylight. I made the same measurement with 70-200L at 200 @ F=2.8 with similar results.

Results have been repeatable.

[JeroenM]

the first "focus attempt" the 20D gets pretty close but when you then focus again (same spot, duh) it gets even better.
my experience. don't take my word on it.

[James Hill]

"Despite how nice it works, the 20D is not a "Pro" camera, and doesn't focus quite that flawlessly."

What other functions are allowed to be less than perfect because it's not a "Pro" camera? Are the apertures slightly off? Shutter less accurate? Lens mount not quite perpendicular to the sensor? Does Canon have an official policy on how poorly cameras are allowed to function based on price? "Ah, this thing's only pro-sumer, we can let the focusing kind of suck."

[Gordon Buck]

Very interesting questions and test.  Am I understanding the problem/question correctly?

Testing ability to auto focus at a distance of about 8 feet with a 400 mm lens and finding that focus achieved is actually sometimes 1 to 2 inches in front of the target but other times 1 to 2 inches behind the target?  An error of, roughly, 1 inch in 100 inches (1%)?

How are you determining the actual point of exact focus?  

Isn't minimum focus distance for the 100-400 about six feet?

How does auto focus error compare to manual focus error?

Really, really not wanting to reopen the depth-of-field can of worms but, with that setup, isn't depth of field also about an inch or two at eight feet?

[Ray]

I haven't noticed any consistent misfocussing with my 20D with the exception of one particular lens, the Canon EF-S 10-22mm. I discovered it whilst comparing the lens with my Sigma 15-30mm at 15mm. The Sigma at first appeared much sharper. Then I discovered the Canon lens was not autofocussed on the test chart, which was a bit odd because this fine Sigma lens of mine once fell out of my vest and bounced down a rocky slope.

Further testing confirmed that the 10-22 always misfocussed with autofocus within the range of about 1 to 5 meters, whether it was a test chart or anything else. At infinity it was okay and closer than 1 meter it was better.

I didn't try the ruler test but I've discovered that a test chart with progressively finer lines allows one to judge very accurately when the lens is focussed. I'm not sure what the correct term is, but at some point on the chart, if the distance is in the ball park, very pronounced aliasing artefacts or chromatic aberration become apparent when the lens is exactly in focus. Slightly out of focus, they disappear.

Refocussing several times (ie. half pressing the shutter several times) caused the focussing to alternate between better and worse, but it was never spot on without manually focussing. The Sigma was always spot on, using autofocus.

Of course, I've returned the lens. But supplies are short in Australia and I haven't yet got my replacement.

It did occur to me that the fairly frequent reports that this lens is a bit soft might be due to a slight autofussing problem with the lens design that some owners aren't aware of.

For what it's worth, my Sigma at its wide end was very marginally sharper than the Canon in the middle of its range, and better in the corners as one would expect.