Chromatic aberation

By far the biggest problem that the S100FS faces is chromatic aberation (CA). Fringing of one sort or another has been something of a traditional problem with FujiFilm cameras but the S100FS produces some of the most dramatic results we've seen from such a high-end camera from such an established brand. This is a big dissappointment because it is by far the most obvious image quality flaw and is utterly unneccesary (it could surely be removed in-camera using the Real Photo Processor).

We found often disturbing levels of CA in the majority of images we took. This led to us taking the unusual step of subjecting the S100FS to a simplified version of our lens tests. (Which is no simple task given its range), in order to characterise the problem.

The results are shown below, and offer a comparison with the Nikkor 18-200mm VR which is probably the best performing of the small number of super-zoom lenses available for DSLRs. This data is not comparable to the charts in our lens tests because we have hadto re-calibrate the vertical scale in order to stop the S100FS's results going off the chart.

The left-hand edges of the graphs represent the performance at the center of the lens and work their way to the edges. The Nikon lens result includes an extra data point because it was tested on a camera with a wider, 3:2 aspect-ratio sensor.

At 28mm equivalent focal length on the S100FS and 27mm equivalent on Nikon's lens, it's clear that the Nikkor is controlling CA much better, though the red channel creeps up towards the edges, resulting in red/cyan fringes appearing towards the edges of the frame. The Fuji has relatively high levels of both red/cyan and blue/yellow CA, resulting in green/magenta fringeing that is more noticable than either on its own.

Our tests showed that CA improves as the zoom extends: it is still very visible at 35mm equivalent, tolerable at 50mm equivalent and settles down to perfectly acceptable levels between 100mm and around 150 mm.

Here we see the results at 110mm equivalent on the FujiFilm and 105mm equivalent on the Nikkor. Again the Nikkor outperforms the S100FS but it doesn't try to cover quite such a wide focal-length range (27-300mm equiv, rather than 28-400mm equiv.), and costs more, just as a lens, than the complete S100FS. Add the cost of even a cheap DSLR and total will be more than twice the cost of the FujiFilm.

Unfortunately, from 200mm equiv. onwards, blue/yellow CA takes off spectacularly. It proceeds to get worse all the way up to the lens's extreme at 400mm. The above image shows the performance of both lenses at 300mm equiv. and demonstrates why we had to recalibrate the vertical scale to accomodate the S100FS's data.

These figures highlight a number of problems. The first is that the lens performs least well at both ends of its zoom range. This is not uncommon in zoom lense designs but is quite a drawback because most people shoot most ofen at one extreme of their zoom or the other. The 'sweet-spot' in which the S100FS performs well is in the 100-150mm equivalent range, which accounted for around 15% of the 480 real-world images we shot with the S100FS.

The other problem is that at some focal lengths (around 200mm equiv. for example) at which the CA is non-linear and rises dramatically at the corners of the frame. This makes it more difficult than usual to remove.

All of which is made all the more sad because the resolution data generated from the same tests showed that, while the performance at the two ends of the zoom are merely 'good', the resolution delivered between about 35mm and 300mm is superb. Click here to see the 28 and 50mm equivalent figures.

Real-world examples

To give an impression of what this all really means for your photos, here are some examples taken at both extremes of the zoom:

100% crop 28mm equiv, F8
100% crop 400mm equiv, F8