Panasonic DMC-G1

We were impressed overall with the imaging performance of the Lumix G1 in our battery of lab tests. The camera's color accuracy and resolution were both first-rate, conceding nothing to the traditional SLRs we used for comparison purposes. There are issues with image noise, and white balance performance was mediocre. Howeverlow light results were impressive, maintaining good color accuracy even as the room dimmed to single-candle lighting levels.

Color (11.47)
The Panasonic G1 includes a response to a problem we've dealt with for some time: the gulf between color accuracy and color attractiveness. For many consumers, a camera that accurately reflects the colors in real life is less desirable than one that boosts flesh tones to make them look healthier, delivers bluer-than-blue skies and autumn leaves straight from a a calendar page. For our purposes, on the other hand, we want to start out with a photo with colors as close to what we experienced as possible, and we'll add whatever level of boost we find desirable in software. The G1 accommodates both audiences by way of the Film Mode feature, offering a range of settings from Standard (the default, vanilla choice) through Dynamic, Nature,Smooth, Nostalgic and Vibrant. For our testing purposes, in pursuit of color accuracy, we used Standard mode to shoot a GretagMacbeth color chart under studio lighting, at all available ISOs, then analyzed the resulting images using Imatest software. Two charts produced by Imatest are particularly revealing. In the first, shown below, the color captured by the camera is shown in each outer rectangle. The small inner rectangle at the right shows the original color of the GretagMacbeth chart. And the larger inner rectangle shows the GretagMacbeth color adjusted to match the luminance of the photograph.

As shown here, the captured color values are mostly very close to the original chart colors.

The G1 did exceptionally well in our color test. The squares on the top row, including the two at top left meant to indicate fleshtone reproduction accuracy, are particularly close, and there are few areas of significant deviation. This can be seen clearly in the next chart, which indicates color accuracy in a more schematic diagram. In this case, the colors captured by the camera are shown in the small circles, the original chart colors in the squares, and the length of the lines connecting the two reflects the difference between them: shorter lines, more accurate color.

The short lines in this chart reflect the G1's excellent color accuracy.

Resolution (10.44)
The megapixel figure in a spec sheet is a poor indication of the actual image sharpness you'll see in a photo taken with a particular camera. The quality of the image sensor, not just the number of tiny photo receptors squeezed onto its surface, is vitally important, along with the way the sensor input is processed. Our resolution testing teases out the actual performance characteristics of a camera by shooting a standard resolution test chart under controlled lighting conditions, at several distances from the chart. We then analyze the resulting images using Imatest to determine sharpness as measured in line widths per picture height (lw/ph), a good indication of the point where you cross over from nice, crisp reproduction of tiny detail into a mushy grey mess.

Actual-size snippet from a G1 shot of our resolution chart

The best resolution score for the G1 was 1923 lw/ph horizontally and 2010 lw/ph vertically, a very good outcome. This result was achieved with the kit lens at its widest-angle setting, but the resolution figures remained high across the 3x zoom range. Imatest indicates that through most lens settings the images are slightly undersharpened, which is fine: oversharpening can be a problem when it produces artificial jaggedness.

One deviation from our standard test procedure should be noted here. We ordinarily shoot our SLR resolution test with noise reduction turned off (we'll do the same with point-and-shoot cameras when possible, but this control is often lacking in point-and-shoots). In the case of the G1, the user can throttle noise reduction up and down by changing Film Mode settings, but it's impossible to turn it off completely. We shot in Standard mode with noise reduction at its minimum setting.

Resolution scores for the G1 stand up nicely to the competition. The difference between the Rebel XSi and the G1 might be visible in very large prints, or when blowing up a particularly small area of a shot to full-size, but overall the fluctuation here doesn't reflect a major performance variation.

White Balance (5.66)
If you're reading a book in the shade or reading it under a fluorescent light, the page still looks white, even though the color of the light is actually very different. While our brains are very good at making this compensation when looking at something live, different shades of white in a photograph are still going to look wrong if the camera can't compensate for the light source discrepancy. That's where whitle balance comes in: it's the digital processing step that tries to recreate the colors our brains tell us we saw at the time the photo was taken. To accomplish this difficult task, most cameras provide two options. There is an automatic setting, where the camera analyzes the light and attempts to adjust accordingly. And there are manual presets, allowing you to tell the camera what kind of light you're shooting under and eliminate the automatic system's guesswork. We test both variations, shooting a standard color chart under a variety of lighting conditions and analyzing the results for color accuracy using Imatest.

Auto White Balance (7.19)
The examples below are based on the bottom line of the GretagMacbeth color chart, which consists of a white square on one side, black on the other, and a series of four neutral gray patches at different densities between. Imatest analyzes photos of the chart and produces these exaggerated images to highlight the deviation from the ideal, Note that these are exaggerated: you wouldn't see this level of inaccuracy in your actual photos.

As seen here, the G1's automatic white balance system did exceptionally well on one light source that's often a problem area: fluorescent light. Otherwise, the flash results are acceptable, but shooting under tungsten (i.e., incandescent light, like ordinary household bulbs) and in the shade caused an unpleasant color shift.

Preset (4.12)
The G1 doesn't have a preset for fluorescent, which is unusual: many cameras we test have several fluorescent presets, reflecting the different types of fluorescent bulbs on the market. Then again, considering how well the automatic system handled fluorescent lighting, it's tough to complain. On the other hand, using the flash and shade presets actually resulted in worse scores than the automatic white balance system achieved. Only the tungsten illumination results, which were pretty bad using auto white balance, were improved by choosing the preset.
 

Overall the G1 white balance performance isn't awful, but it is certainly an area that could use improvement in Panasonic's next Micro Four Thirds model.