Casio Exilim EX-FH20

In our rigorous set of lab tests, the FH20 consistently underperformed. While rarely doing so poorly that we felt like throwing it out the window, in every test bar white balance, it scored worse than most, if not all, of the cameras we compared it to. In some cases it was only a small difference, but in others it severely disappointed us, especially when compared to its brandmate, the FX1, which uses a different sensor. Overall, it was a sub-par result that does not speak well for the image quality you'll get from the FH20.

Color (6.01)
To test how well the Casio Exilim EX-FH20 can record color, and how accurately it does so, we rigorously test the camera in our labs. We shoot the industry standard GretagMacbeth color chart under 1700 lux illumination. We run the resulting images through Imatest image analysis software. This produces the chart shown below, which demonstrates color accuracy. For each square, the outer rectangle represents the captured color, the inner square is the ideal color corrected for luminance, and the inset rectangle is the actual color from the original chart.
 

The GretagMacbeth color chart shows the FH20's good performance with purples and greens

This information is displayed in another fashion in the graph below. Here, the ideal color is shown as the square, the color the camera captured is in the circle, and the angle and length of the line between the two represents the shift in hue and saturation from the ideal.

The FH20 had trouble with blues and oranges

Resolution (5.72)
In the common parlance, resolution is often considered equivalent to the megapixel count of a camera. However, many other factors affect the actual resolution of your final image, including lens optics and image processing. To test a camera's image sharpness capability, we shoot a standardized chart and use Imatest to analyze these photos. The result is expressed in line widths per pixel height (lw/ph), a measure of alternating black and white lines in a selected area.

The industry-standard resolution chart that we use for testing

Once again, the FH20 didn't perform up to expectations. It offered a mediocre 1724 lw/ph horizontally, and 1343 lw/ph vertically. This puts it below the entry-level SLR Canon XS, a number of cheaper, older ultra-zooms, and its big brother, the F1.

White Balance (10.06)
White balance adjustment is the way a camera adapts to differing hues of light cast by different sources. For instance, the color of the light that comes from an incandescent bulb is different from a fluorescent tube, which in turn is different from the sun. A camera usually has multiple ways of dealing with this phenomenon, and we test two of them. Firstly, there's auto white balance, where the camera's sensor attempts to read the lighting situation and adjust accordingly. Second, there are a series of presets associated with most common light sources, which the user selects manually. In testing both of these approaches, we shoot under varied light sources, photographing the GretagMacbeth chart, and run it through Imatest to see how accurately the camera captures the chart's whites and grays. Keep in mind, differences in the images you see below are highly exaggerated; you will not see this much discrepancy when actually photographing with this camera.

Auto White Balance (8.82)
The FH20 impressed us by having a strong automatic white balance. This was especially noticeable under flash illumination, and indirect/cloudy sunlight. However, the camera struggled a bit under tungsten illumination, which is a common problem.

Preset (11.30)
The white balance presets were also highly accurate. The camera did very well under fluorescent light and indirect sun, and better (if not exactly great) with tungsten compared to the auto result. There was no preset for flash, though.
 

Exaggerated White Balance Errors (Presets)

Fluorescent Preset WB - Fluorescent Illumination

Cloudy Daylight preset WB - Daylight illumination

Tungsten preset WB - Tungsten illumination

The EX-FH20 shoots well using white balance presets, but even without manual intervention the automatic system works well. This is boon for point-and-shooters who don't want to have to fiddle with their settings.

page title="Noise and Video"]
Noise
Noise is the appearance of digital static in your image. It can strike at any time, but becomes far more noticeable at higher ISOs and across large areas of a single color. We test noise levels in two ways. First, we shoot everybody's favorite GretagMacbeth chart at 1700 lux, and with every available ISO, and analyze the resulting images for noise levels. Secondly, we set the camera to Auto ISO, and let it decide the best setting to use.

Noise – Manual ISO (7.26)
The noise levels on the FH20 started about average, just over 1% at ISO 100, and rose quite dramatically to 2.75% at ISO 1600. While not a huge noise level, it's pretty high, and your high ISO shots will definitely suffer for it.

As you can see from the graph below, this score is only slightly below our comparison cameras.

Auto Noise (1.82)
Unfortunately, the FH20 did substantially worse in our auto noise test. This is as much an exam of the camera's smarts as it is of its noise level performance. Since we shoot in well lit conditions, the camera can quite safely shoot at ISO 100 or 200, which would lead to low noise levels and a good score. The FH20 decided that ISO 400 was an appropriate setting which, coupled with the slightly above-average noise levels for the camera, gave it an abysmal auto noise score.

Low Light (6.68)
A camera's ability to function in low light is important, as it's rare to find yourself in a situation with perfect illumination. Our low light testing proceeds down a twofold path, first shooting at successively lower illumination levels, then shooting long exposures at a fixed low light level, analyzing color and noise results for both.

As you can see above, we sampled from 60 lux, about average household illumination, to 5 lux, the brightness cast from a single candle flame. The FH20 functioned well under 60-lux illumination, maintaining roughly the same color accuracy we saw in full light. Likewise, it handled itself pretty well at 30 lux, but at 15 and 5, it promptly freaked out, and drove its metaphorical car off the cliff, hitting us with low color accuracy and high noise levels.

We also shot long exposures at ISO 400. Ideally we run tests ranging from 1 second to 30 seconds. Unfortunately, the FH20 can't shoot beyond 8 second s at ISO 400, so our testing procedure was shortened. We also had a problem with the FH20 blowing out the image, even under such dimly lit conditions. However, with a bit of tweaking, we achieved the result shown below.

Once again, a slightly underwhelming result from the FH20. The noise levels started out quite high, and improved a little over the course of the test. Still, the FH20 manages to almost, but not quite, live up to what we're comparing it with, only outperforming the pricier EX-F1. It's not horrible, but it could be better.

Casio Exilim EX-FH20 Low Light Scores

Still Life
To demonstrate the color accuracy and noise performance you can expect to see under normal conditions, we shot these pictures of our lovely couple and Rosie the Riveter and friends under ordinary fluorescent lights, using the camera's automatic mode, at all available ISOs. Click for the full size images, but be prepared for slow downloads, since the files are large.

Video Performance (6.60)
While the FH20 can shoot fast enough film to please every David Attenborough in training, we only tested video performance at the highest resolution and standard speed. As we'll discuss further, the high-speed modes had a tendency to be dark, horribly focused, and generally disappointing.