XL1 Focus Jumps and Other Observations
an article by Jan Engvald

Shooting anything that is white (flowers, walls, clothes), if I adjust exposure for just no zebra pattern, then it becomes much too dark. Could not see any zebra pattern level adjustment, can Canon service do such adjustment? I would like the zebra pattern to activate at 90-100% video level (at least I think so now).

Wide shots are as sharp as tele shots, as far as I can see on my monitor. However, as expected, f/22 and f/32 are VERY soft. But I found that the camera tells me to turn on the ND filter in those cases, so hopefully I won't shoot with those apertures.

The image in the viewfinder is a bit inconsistent. If I keep my eye exactly in center, then the image is a tad too dark. Moving my eye up left halfway to the edge of the screen, then it is too bright. Moving halfway to the lower right, then all image is mostly gray, no details can be seen. Moving a bit further, and the image becomes negative (light areas become dark and vice versa)!

I definitely don't like that the focusing ring is SPEED sensitive. That is very non human-intuitive. I don't mind that it is a servo lens, but the ring should ONLY be angular sensitive, NOT angular SPEED sensitive. Imagine how many more car accidents there would be if some company made their car steering wheel servo angular speed sensitive!

Backfocus is correct: zooming in and focusing, then zooming out to wide still gives perfect focus. Zooming in again and it is still in focus, so it keeps focus.

But I hadn't shot more than 15 minutes outside when I was hit by the focus jump problem. I was using manual focus (used it most of the time) and had just pressed push-to-focus, then moved my hand to get a steady grip on the lens. It jumped out of focus. Repeated the procedure, with the same result. A third time, and then it worked.

After returning home I just HAD to find out when focus jumps occur and when they do not. Did a lot of testing and found the following (using manual focus):

• It does not occur when fully zoomed in, but on all other zoom settings.
• It does not occur if you only use the focus ring, only occurs when using push-to-focus.
• When it jumps out of focus at the current distance, it always jumps into focus at the distance you previously focused to. This is interesting, it rules out a lot of otherwise possible things, like bad mechanics or random pickup of electrical noise by the focus ring movement sensor.
• The jump occurs exactly when you change the zoom, ever so slightly, after a push-to-focus. It does not happen always, but rather often (about one third of the times in my test).

To verify my tests place three or four objects at different distances, and then repeatedly do: aim at a different object, push-to-focus, turn the zoom ring 1 degree.

Executing the above test I also found that in 1 % of the cases push-to-focus never found focus, at least not within a minute and normally it takes less than a second.

My conclusion after testing is that the focus jump problem is a program error in the camera. Probably some variable that does not get updated in all possible cases when it should. If someone can send these observations to the Canon developers, they should be able to test and find the error. Hopefully old cameras then can be updated with the new and fixed program.

Despite the problems listed above, I still think it will be useful to me. But it could easily have been even better.

XL1 PAL Resolution Tests

Original test pictures: (source is http://www1.canit.se/~thbe/testcard.htm)

After viewing these images, use your browser's "Back" button to return to this article on the Watchdog.

• Upper left
• Upper right
• Lower left
• Lower right

The resolution scales on the test pictures above should be doubled in the pictures below, because I mounted two pictures side by side. Thus, when it says 200 (lines per picture height), the XL1 sees 400, and when it says 250 it is really 500.

XL1 PAL Resolution Test Results

Method: The above pictures were laserprinted (B&W) and put on a wall side by side. Pictures below grabbed via FireWire and the Spark program to a .BMP file. Then converted to .GIF using grayscale only (to avoid dithering) with Lview.

• Zoomed to tele (focal length 44 mm) and aperture f/8.0
• Zoomed to tele (focal length 44 mm) and aperture f/1.8
• Zoomed to wide (focal length 5.5 mm) and aperture f/8.0
• Zoomed to wide (focal length 5.5 mm) and aperture f/1.6

As seen in the pictures above, wide is as sharp as tele, stopped down as well as full open. Only full open tele is slightly softer. It seems my XL1 has a resolution somewhere in the range more than 400 and less than 500 lines. The resolution is probably limited by the optical low pass filter, not by the glass in the lens. The low pass filter removes all staircase apperance of slanting lines. Compare to the Sony VX-1000, which has less effective low pass filter giving a sharper picture but with staircase artifacts ("jaggies").

Below is a series of pictures with a reference and then really small apertures to find out how that affects resolution.

• Zoomed to wide (focal length 5.5 mm) and aperture f/5.6
• Zoomed to wide (focal length 5.5 mm) and aperture f/16
• Zoomed to wide (focal length 5.5 mm) and aperture f/22
• Zoomed to wide (focal length 5.5 mm) and aperture f/32

As can be seen f/32 is really bad (as any lens would be at such a small aperture), resolution only 250 lines! Also f/22 is bad, less than 400 lines. At f/16 resolution is good, above 400 lines.

I think that f/22 and f/32 should never be used. Actually, on Auto mode (Green rectangle) and Av and Manual programs you can never get or even set those apertures. Only Auto and Tv can cause the aperture to be set to f/22 or f/32. If you use those, when the blinking "ND on" sign is shown, turn on the Neutral Density filter and you will avoid apertures above f/16.

XL1 PAL Focusing Accuracy

• Critical focus accuracy: At aperture f/1.8 zoom in, focus, zoom out to focal length 12 mm
• At aperture f/1.8 and focal length 12 mm do manual focus
• At aperture f/1.8 and focal length 12 mm do push-to-focus

Critical focus and push-to-focus produces exactly the same sharpness, the manual focus gives a few lines more resolution, but it is very hard to see the difference. Thus with my XL1, for sharpness reason, any of the methods can be used.

I'm sorry that the laserprinter I used to print the test pictures didn't perform that good (not very good contrast), light and exposure wasn't the best either because I had only one lightsource powerful enough. Please just compare resolution.

XL1 Gradient Banding

You must set your display to 24 or 32 bit color to judge the pictures below, else you will see false banding.

• Smooth gradient at gain -3 dB
• Smooth gradient at gain 0 dB
• Smooth gradient at gain +6 dB
• Smooth gradient at gain +12 dB

The banding is not as severe (in fact I can hardly see any banding) as I have seen on other pictures published. Maybe they had false banding when viewed, or else Canon has been able to reduce the banding. Anyway, this banding is very seldom seen in real word images.

XL1 Summary

It seems my XL1 has fewer or less problems than what I expected after reading about them on Usenet. I will probably return the viewfinder for exchange or adjustment. I would like to get a program update to fix the focus jump problem. If the angular speed sensitivity of the focus ring is done by software I hope Canon will fix it, if it is inherent to the rotation sensing then there is not much hope (maybe a new lens with another design). Everything else is as good as or better than expected. I like it!

Jan Engvald
Lund University Computing Center
Lund Sweden
Jan.Engvald@ldc.lu.se

Back to the XL1 Articles Menu
Thrown together by Chris Hurd