Canon Powershot S80 Tests

This page started out as a post of some tests I did in December 2005 to evaluate the Canon Powershot S80 camera. The original page has turned into an evaluation of the S80, including a comparison with the A620.

(Updated 2/6/06): Although the tests demonstrate that some samples of the S80 have corner blur, they also show that it is repairable. (See highlighted box below.) Furthermore, I've come to agree with most S80 owners that it is not a serious drawback in most situations.

Perspective: Despite the corner blur shown in Test 1, the purpose of this page isn't to show what a bad camera the S80 is. In fact it's a very good camera. Like any camera, it has strong points and flaws. Some of them - in my opinion - are listed in the comments sections below.

My purpose in doing these tests was to understand a couple of the S80's weak points. When you look at the samples, particularly the corner sharpness tests, please also look at the test poster and keep in mind what a small part of the image is affected. To put it in context, look at real world sample photos in the comparison section. There are also sample shots in the dpreview.com review of the S80.

I saw corner blur in all the on-line S80 samples with short focal length that I've looked at - including the ones on dpreview.com, Steve's Digicams, and even on Canon's own web site. But it seemed like my camera might be worse. So I asked Dell to send a replacement camera. (I'll give Dell an A- for customer service. Brickbats for their horrible "voice mail jail" phone system, but kudos for service once I got a person - the replacement showed up at my door just about 24 hours after I'd requested it! More brickbats for packaging though - the camera box was tossed in a shipping box twice its size, with absolutely no packing material at all! No visible signs of damage though.)

After noticing that the samples on Canon's web site were all shot at f4.0, not wide open, I shot some tests at different apertures - see Test 4. I also tried some exposures and found that the S80's firmware prefers f4.0 in Auto and P modes, and will not use wider apertures unless there is insufficient light.

Update (Feb. 6, 2006): I sent Dell's replacement S80 to Canon for repairs. They replaced the optical unit. This page has now been updated with comparison test crops from the "Repaired S80". See the tests below.*

Summary of tests: The original and replacement cameras had centering problems with the lens, causing corner softness that varied and moved around over the zoom range, with the worst blur not at 28mm as one would expect, but at 42mm. When the replacement camera was repaired, the result was that there was considerably less corner softness which did not vary from corner to corner at a given zoom. As one would expect, it was noticeable mostly at 28mm, and even then, reducing the aperture from maximum to f4 all but eliminated it.

My conclusion is that there is a quality control problem on the S80 lens assembly, causing blur in the corners on some units, The problem can be almost eliminated by replacing the lens assembly.

(Canon's repair was prompt. Using a form on their web site, I was able to obtain a return authorization quickly. They fixed the camera in a week and shipped it back Fedex second day. I received four email notifications as the camera advanced through the process.)

After all this testing, and extensive use of both the S80 and an A620 on a three week trip to Asia, I've decided that I am quite happy with the S80. I plan to put my A620 up on eBay.

For detailed observations and experience with the S80 and A620, see my comments at the end.

(Click on the picture for the original image, 2 Mb, shot at 42mm f4 with the replacement S80. They were all shot at ISO 50, with +2/3 exposure compensation, and no post-processing. Some of the shots were at short range, especially the wide angle ones, so I locked the focus on the upper left corner of the image before centering and shooting. That was probably unnecessary though - I took one picture with center focus at 35mm equivalent to compare, and couldn't tell the difference at 100% zoom. They were shot at fine compression - see Test 2 for a comparison with superfine. My profound apologies about the ratty exposures; I was only interested in sharpness.)

Test 1: Corner Sharpness. There's been a lot of discussion on the forum about the corner blurring on the S80. So I wanted to see how bad it was.

Important: The tests in this section were all taken at wide open apertures. This is a worst case, not a typical one, since the S80 favors a smaller f4 aperture where there is enough light. See Test 4 for other apertures, which reduce the blurring and are more representative of photos the S80 is likely to produce.

My first test was wide open at 28mm equivalent. Here are 100% crops of the upper left corner and the middle, using the original S80 camera.

After seeing this, I wanted to run more tests. (To put the corner sharpness into a bit of perspective, see these comparisons between the original S80 and an A620.)

In one thread it was suggested that S80 corner blurring might be due to an off-center lens on some units. So I looked at all four corners. There was some variation, but all of them showed some blur.

	Repaired S80	Replacement S80	Original S80
	100% crop. 1/60 @ f2.8 (+2/3 exposure comp)* - 5.8 mm (28mm equiv.)	100% crops. 1/100 @ f2.8 (+2/3 exposure comp) - 5.8 mm (28mm equiv.)
upper left
lower left
upper right
lower right

What I see: At this zoom, the repaired S80, to put the best face on it, is consistent in all four corners. But it's not any better overall than the others, possibly a bit worse. The original S80 is blurry on the left side, quite good on the right except for a tiny bit of the corner. The replacement S80 is quite good except for the lower left corner, which is perhaps a tad worse than the original S80.

Of course, that's at the widest end - 28mm - an unusually wide lens for a compact camera. So it should get better at a wider aperture, right? Let's see. On the original S80, 35mm and 42mm looked about the same, so here's 42mm:

	Repaired S80	Replacement S80	Original S80
	100% crops. 1/25 @ f3.5 (+2/3 exposure comp)* - 8.6 mm (42mm equiv.)	100% crops. 1/80 @ f3.5 (+2/3 exposure comp) - 8.6 mm (42mm equiv.)
upper left
lower left
upper right
lower right

What I see: Here the repaired camera is much better. The corner softening is barely perceptible in these 100% crops. The other two cameras, on the contrary, actually seem worse at 42mm. On the original, the blur has moved over to the right side, perhaps indicating some sort of centering problem. On the replacement, the right side has also gotten a lot worse. The left side hasn't changed much.

Testing the original camera, I saw improvement starting at 48mm. There was a little corner blurring at 55 mm, mostly at the bottom. At all longer zooms, the blur was negligible and corner sharpness was just about as good as at the center.

Test 2: Fine or Superfine? These are 200% crops with enhanced contrast to emphasize the compression artifacts.

Compare the compression artifacts, especially around the lettering, and the texture and highlights in the center of the flower.

Most of my pictures are shot on trips where I need to carry along weeks of photos. For that reason, I shoot in fine.

Test 3: Dust Reflections. There has been a lot of discussion of the S80's susceptiblity to dust reflections when using flash where there is dust in the air. One good discussion is here.

I can confirm that it is a problem, the S80 is particularly liable to it, and it doesn't only happen at 28mm. To test it, I went into a walk-in closet, blew dust off the top shelf, and took a lot of flash photos.

Out of 33 S80 photos, the reflections appeared on four. It didn't make any difference whether the zoom was at 28, 35 or 44mm. The room has overhead spot lighting, and the dust was quite visible in both the A620 and S80 LCD displays.

The photo to the right shows the dust reflections. It isn't particularly bad here; there are many worse examples on the discussion forum.

To test the assertion that all digital cameras are susceptible to this phenomenon, I took a number of simultaneous photos with the A620 and S80 side by side, both zoomed to 35mm. Only the S80 showed the dust reflections.

Test 4: Corner Sharpness Revisited at f4. I noticed that the S80's firmware prefers shooting at f4 to shooting wide open. The sample pictures on Canon's web site are all at f4. So I revisited the sharpness tests using that aperture. (All these shots were taken in Av mode, +2/3 exposure compensation, Auto ISO.)

	Repaired S80	Replacement S80	Original S80
	100% crops, 1/25 @ f4.0 (+2/3 exposure comp)*, 5.7 mm (28mm equiv.)	100% crops, 1/60 @ f4.0 (+2/3 exposure comp), 5.7 mm (28mm equiv.)
upper left
lower left
upper right
lower right

What I see: There is a little bit of improvement all around at f4 compared to f2.8. At f2.8, the lower left corner of the replacement camera and all corners of the repaired camera were blurry, but all of that has improved. Really, at 28mm f4 there is not much of a problem.

(I compared each camera at f2.8 and f4 by opening this page in two browser windows and putting them side by side.)

Here are the corners at 42mm equivalent, which was the worst focal length for corner blur in the original and replacement cameras:

	Repaired S80	Replacement S80	Original S80
	100% crops, 1/25 @ f4.0 (+2/3 exposure comp)*, 8.6 mm (42mm equiv.)	100% crops, 1/60 @ f4.0 (+2/3 exposure comp), 8.6 mm (42mm equiv.)
upper left
lower left
upper right
lower right

What I see: The repaired camera is almost crisp; blur is barely perceptible. As for the other two, stopping down from f3.5 to f4 seems to have helped the replacement camera more than the original.

Let's take a look at the effect of different apertures in the original camera. This is the lower right corner, with the original S80 at 42mm.

(At this point I wondered if I was focusing favoring the left side, so I took two pictures, one with the camera right side up and one with it upside down. The right side of the camera was blurrier both times. So the difference is in the camera.)

Next, let's look at corner sharpness throughout the zoom range, comparing the repaired and replacement S80. These comparison photos show two corners, first the lower left, and then the upper right.

What I see: On the repaired camera, the only problem is some softening at 28mm. On the replacement camera, the lower left is a bit soft up through 42 mm, but the upper right is fine at 28mm, then blurry at 35mm and 42mm, and soft through 55mm.

Conclusions: The original and replacement cameras had centering problems with the lens, causing corner softness that varied and moved around over the zoom range, with the worst blur not at 28mm as one would expect, but at 42mm. When the replacement camera was repaired, the result was that there was considerably less corner softness, it did not vary from corner to corner at a given zoom, and as one would expect, it was noticeable mostly at 28mm and even then, reducing the aperture from maximum to f4 all but eliminated it.

Real-life photos: These were shot in Program AE mode, evaluative metering, and I let the cameras pick the exposure. All were taken at 42mm equivalent to stress out the S80's corner problem. Several high contrast subjects were include to see how the cameras handled the highlights and shadows. Click on the images to see the originals (large files).

What I see: I can't really see it at this size, but at double this size or larger the corner defects in the S80 image above are clear, especially blurring in the upper right and fringing in the lower left.

Replacement S80, 1/250 @ f4, 8.6 mm	A620, 1/320 @ f4, 8.5 mm (42 mm equiv)

What I see: (updated 12/16/05) In all four of these pictures, comparing the full size shots (click on the images) the S80 is a little sharper on the left side, but it has blurring all the way down the right side. In the last three shots, I have the same shots taken with the original S80 (full size images, 3 megs+: second, third and fourth). Both S80s show the same right side blurring.

Replacement S80, 1/320 @ f4, 8.6 mm	A620, 1/250 @ f4, 8.5 mm (42 mm equiv)

Replacement S80, 1/500 @ f4, 8.6 mm	A620, 1/500 @ f4, 8.5 mm (42 mm equiv)

Corner sharpness test: These were taken at 35mm equivalent with my original S80 and a Canon Powershot A620. The 620 is 7 megapixel, so its crops cover a little bit more image. Apologies for the exposure differences - this was just supposed to test sharpness.

* Note: The test photos on the Repaired S80 were taken with daylight lighting. The tests on the other two cameras were taken with a combination of artificial and low daylight. So the exposures are different, and some shots are darker or lighter than the ones they are being compared with. It was felt that this was acceptable since what is being tested is sharpness, not exposure. In all cases a tripod and 2 second self-timer delay were used.

Superfine compression	Fine compression	Normal compression
200% crops, 1/80 @ f 3.2, 7.3 mm (35 mm equiv.)

Original S80, 100% crops, 8.6 mm (42mm equiv.)
1/80 @ f3.5	1/60 @ f4
1/50 @ f4.5	1/30 @ f5.6
1/20 @ f6.3	1/15 @ f8

Original S80,100% crops, 8.6 mm (42mm equiv.)
1/80 @ f3.5	1/60 @ f4
1/50 @ f4.5	1/30 @ f5.6
1/20 @ f6.3	1/15 @ f8

equiv. focal	Lower left corner
equiv. focal	Repaired S80, f4, ISO 50, +2/3 exposure comp*	Replacement S80, f4, ISO 50, +2/3 exposure comp
28mm
35mm
42mm
48mm
55mm
73mm
100mm

equiv. focal	Upper right corner
equiv. focal	Repaired S80, f4, ISO 50, +2/3 exposure comp*	Replacement S80, f4, ISO 50, +2/3 exposure comp
28mm
35mm
42mm
48mm
55mm
73mm
100mm


Tv (Shutter Speed) 1/100 - Av (Aperture Value) 2.8 - Focal Length 5.8 mm (28mm equiv.)

Original S80, 1/250 @ f4, 8.6 mm	A620, 1/200 @ f4, 8.5 mm (42 mm equiv)

	Original S80	A620
Lower left corner
Upper left corner
Near center
	100% crops - 1/100 @ f 3.2 - 7.3 mm (35mm equiv)	100% crops - 1/100 @ f 2.8 - 7.3 mm (35mm equiv.)