The S100's lens covers a useful range of 24-120mm (equivalent), which is far from the most versatile of any current raw-shooting compact camera, but significantly more useful than the S95's 28-105mm (equivalent) zoom. The lens is optically stabilized, but while it's fast at the wide end (F2.0) the maximum aperture is limited to a rather less-useful F5.9 at the longest telephoto setting. This is an inevitable consequence of the S100's pocket size - camera with faster lenses at the tele end, such as the Olympus XZ1 or Fujifilm X10, are rather bulkier.

Reported max aperture vs focal length (vs S95)

While the S100's lens maintains a fast maximum aperture of F2.0 at wideangle, it's relatively slow at the long end (F5.9). However these numbers have to be interpreted in the light of the fact that the zoom range is extended in both directions compared to the S90/S95, from 28-105mm equivalent to 24-120mm equivalent.

The table below compares the reported maximum apertures across the focal length range between the S100 and S95; the numbers were obtained using the lens control ring to operate the zoom in steps. From this we can see that the S100's lens is in fact only about 1/3 stop slower than the S95's across the cameras' shared range (note that the aperture is reported to the nearest 1/3 stop, aside from the telephoto end).

Focal length (35mm equivalent)
* At 105mm equivalent

Lens range

The S100's lens is both wider and longer than the S95's, and here you can see the difference that extra wide and long coverage makes in normal use.

S100 Wide (24mm equiv) S100 Zoom (120mm equiv)
S95 Wide (28mm equiv) S95 Zoom (105mm equiv)

Looking at the numbers alone the difference between 24mm/28mm and 120mm/105mm might not seem like much, but as you can see in the examples above the S100 is able to capture a significantly wider field of view at the wide end of its zoom than the S95, and offers a small but useful boost in zoom at the long end.

Sample variation

Sample variation is a fact of life, sadly, and as we've already reported, we have used several S100s during the course of our testing. Working through our standard studio tests we found moderate to serious optical issues with three of them - significant enough, given the high quality of samples that we'd seen elsewhere, to give us pause (and to delay the process of completing this review).

We have now used five cameras, with lenses that span a range in terms of optical quality. The best sample offers good, uniform sharpness in all environments in which we used it, and the worst gives noticeably soft results on one side of images taken in the critical environment of our studio, but delivers perfectly acceptable results in 'real world' shooting.

Studio Scene (JPEG, ~75mm, f/4.5, ISO 80)

In this table we are comparing two cameras, Cameras A and B, in our standard studio scene. This scene is shot at a specific subject distance (roughly 1 meter) and within a tight range of focal lengths (in this case 76mm and 74mm equivalent, respectively).

In our studio test scene (shot at a distance of about 1 meter) both cameras exhibit acceptable sharpness in the center and lower left of the frame. However, camera B's lens is noticeably soft on the lower right corner.
Camera A
Camera B
Center crop (100%) Center Crop (100%)
Lower left crop (100%) Lower left crop (100%)
Lower right crop (100%) Lower right crop (100%)

As you can see from these samples, Camera A gives decent, if not spectacular sharpness across the entire image area, whereas Sample B gives excellent sharpness in the middle, acceptable sharpness at lower left, but poor sharpness at lower right. From examination of this image alone, most people would probably conclude that Camera A is 'good' and Camera B 'bad', but things aren't quite that simple...

Real world scene (JPEG, ~24mm, f/4.5, ISO 80)

This scene was shot with the same two cameras, at 24mm (equivalent) and an aperture value of f/4.5. Focussing was set from the wooden docks in the middle distance.

In the real-world at the widest focal length both cameras have captured excellent images that are virtually indistinguishable.
Camera A 24mm
Camera B 24mm
Center crop Center crop
Lower left crop Lower left crop
Lower right crop Lower right crop

As you can see, at this focal length, and in a 'real world' scene with more depth, both cameras give basically identical image quality. Even at 100% on screen we'd struggle to distinguish results from either camera, both of which give excellent results. The corners are a little soft compared to the center from both samples, partly as a result of in-camera distortion correction, but image quality remains perfectly acceptable even here.

Real world scene (JPEG, ~85mm, f/5.0, ISO 80)

Near the far end of the zoom range both camera A and B are still acceptably sharp in the center but in terms of corner sharpness the two cameras have swapped compared to the studio scene. Camera B, which had a soft corner in the lower right of the studio scene is now sharper than camera A in the same corner.
Camera A 85mm
Camera B 85mm
Center crop Center crop
Lower left crop Lower left crop
Lower right crop Lower right crop

Out in the real world at 85mm (equivalent) - pretty close to the focal length at which we shoot our studio scene - there are differences between our two sample cameras but the differences are subtle. Looking at the crop from the lower right of this image, Camera B - which gave poorer results in this area in our studio scene - has the edge over Camera A. Whereas Camera B gave soft results towards the lower right of the studio scene, it is camera A which appears slightly decentered on the right in this image, and by contrast, Camera B is slightly softer at lower left.

These images should serve as a perfect example of why you shouldn't read too much into a single set of samples, taken in a single environment. If you do a lot of copy work at relatively close range, you'll likely be unhappy with Camera B. For everyday shooting, however, it will keep you perfectly content - and remember that Camera B was one of the worst cameras of the five that we looked at, judged on its performance in our studio scene. Even the absolute worst camera that we used (again, judged by its performance in our studio at a focal distance of one meter) gives excellent performance in pretty well every other environment in which we used it.

We're not letting Canon off the hook though - we perform exhaustive critical studio testing precisely because it shows up aberrations that might not normally be evident. The level of sample variation that we've seen in the various cameras that we've used - however unimportant in normal use - is unusual in our experience of such a high-end camera, and definitely counts as a Bad Thing.