General Electric G1 Digital Camera Review
Read a digital camera review of the General Electric G1 on DigitalCameraInfo.com.
Testing / Performance
**We tested the GE G1’s color accuracy by photographing an industry standard GretagMacbeth ColorChecker chart. The ColorChecker chart is made up of 24 color tiles that sample colors from around the spectrum, focusing on commonly photographed colors such as green foliage, blue skies, and flesh tones. We then ran these images through Imatest, which compared the ideal colors of the chart with the actual colors the camera reproduced. The modified chart below shows the camera’s reproduction in the outer portion of each color tile. The inside square within each tile is the ideal color of the chart corrected for luminance. The small vertical rectangle is the ideal.
The G1 was most accurate about ½ of a stop underexposed, which is why the small rectangle looks brighter than the outer square. Looking at the inner square compared to the outer square, you can see several of the colors are off the mark. This color error is quantified in the graph below, which shows the actual colors of the ColorChecker chart as squares on the color spectrum, and the colors the camera reproduced as circles. The line connecting the squares and the circles corresponds to the color error.
The graph shows significant color error in blues, greens, and yellows. Blues are often shifted to enhance skies, while yellows and greens remain more accurate. Unfortunately, the GE G1's 9.19 mean color error is not very good, so unless you like the way the colors are shifted, you will not be able to shift them back to their original shade.
White Balance ***(3.89)*
*We test white balance accuracy by photographing the ColorChecker chart under various types of lighting: cloudy outdoor, fluorescent, and tungsten, using both the Auto white balance setting and the appropriate white balance presets. We also shot the chart using the G1’s flash, but it lit the chart so unevenly that we couldn’t score the white balance confidently. The Auto setting in the other three types of light was not very accurate. It was mediocre in fluorescent light, poor in outdoor cloudy light, and downright terrible in tungsten light.
*Like its Auto setting, the G1’s white balance presets performed horribly. Both the cloudy and fluorescent settings performed worse than the Auto setting under the same light. The tungsten preset was just as terrible as the Auto setting, but in the opposite direction; it turned images blue instead of yellow. Stick to using Auto white balance with the G1 and hope for a miracle.
**Still Life Sequences
***Click on the thumbnails below to view the full-resolution images*
Part of the first-ever line of GE cameras, the G1 carries solid specs for today’s camera market, including the 7-megapixel sensor. We put its imager to the test by photographing an industry standard resolution test chart at various apertures and focal lengths. We ran the images through Imatest to determine how much detail it could resolve. Imatest reports resolution in terms of line widths per picture height (LW/PH), which corresponds to the number of alternating parallel black and white lines the camera could discern before they began to blur.
Click on the chart above to view the full-resolution image*
*The G1 did not fair well in this test. Its sharpest image was shot at ISO 80, using an aperture of f/4.3, and a focal length of 19 mm. The G1 resolved 1475 LW/PH horizontally with 18.3 percent undersharpening, and 1385 LW/PH vertically with 38.5 percent undersharpening. Most point-and-shoots generally benefit from slightly more aggressive sharpening to make the photos look good right out of the camera.
Also worth noting is the amount of lens vignetting apparent in the image. Lens vignetting is when the edges and corners of the frame are darkened, as you can clearly see in the test chart image, which was uniformly lit. This suggests a poor optical system and will require users to manually brighten the corners of photos to correct it.
******Noise – Auto ISO (1.21)
We shot the ColorChecker with the G1 set to Auto ISO, and ran it through Imatest to calculate the noise. The G1 chose an unusual ISO of 99 under our bright studio lights. Usually, shooting at an ISO around 100 will yield low noise, but in the G1’s case, noise was still very much abundant. We measure noise as a percentage of the image that is drowned out by it, and even at ISO 99 this value was 1.76 percent. This is a poor score.
**Noise – Manual ISO ***(2.63)*
**We also evaluated the camera’s noise levels over its entire ISO range. The graph below plots the G1’s sensitivity settings on the horizontal axis, with the corresponding noise levels on the vertical axis.
Most remarkable about this graph is that noise dips very low at ISO 800. Because ISO is a measure of sensitivity, it is natural to have higher noise levels at higher ISO. Such a significant drop in noise after ISO 400 points to considerable noise reduction being applied. This is confirmed by viewing an ISO 400 image compared to one shot at ISO 800, as shown in our still life image chart above. Notice that the stripes on Randy’s jacket are visible in the ISO 400 shot, but are almost completely obscured by smoothing in the ISO 800 image. This is the tradeoff when noise reduction is applied; noise levels are lowered, but sharpness is drastically decreased. The high noise levels at low ISO suggest that without noise reduction applied, noise levels at ISO 800 and 1600 would be astronomically high. The GE G1 has stretched too far to include high ISO sensitivities, and this is evident in its very poor score. **Low Light (7.29)
**We dimmed our studio lights to test the limits of the G1’s low light performance. We shot the ColorChecker chart at light levels of 60, 30, 15, and 5 lux, and ran the images through Imatest to gauge the color accuracy and noise levels. These light levels correspond to a room softly lit with two lamps (60 lux) down to very low light that requires squinting (5 lux). All shots were taken at ISO 1600.
Noise levels were quite low in low light, but are strongly smoothed over, greatly reducing detail. Color accuracy isn’t great, but didn’t drop too far at low light levels. At 5 lux the camera could not expose properly without going into Slow Shutter mode. In its slowest long exposure - 2 seconds - the camera overexposed at 5 lux. Clearly, the G1 has limits in low light.
We also tested long exposures up to 30 seconds, which you can see in the graph below. All long exposures were shot at ISO 400. Noise levels remained essentially the same for all long exposures, with perhaps a little decline after 10 seconds, which hints at increasing noise reduction. However, color accuracy dropped significantly on long exposures, from a mean color error of 13.7 at 2 seconds, to 20.9 at 30 seconds. The slow shutter option is a fun addition to the G1, but don’t expect to take nice looking shots with it.
Dynamic Range ***(5.51)***
We test dynamic range, a measure of the total tonal range of a camera, by photographing a backlit Stouffer test chart. This chart consists of a long row of rectangular tiles, each a slightly different shade of gray, varying from brightest white to darkest black. The more tiles a camera can expose, the better its dynamic range. We tested the G1 at every ISO sensitivity to see how the dynamic range maintained over the entire range. The results are shown in the graph below, with dynamic range plotted in units of Exposure Value (EV).
The G1’s dynamic range at ISO 80 is only about 6 EV, which is quite poor. After falling steadily with increased sensitivity, the dynamic range takes a jump back up at ISO 800. The reason for this rise in dynamic range is because of the noise reduction being applied that we noted in the noise section above. Noise levels significantly limit dynamic range by making it harder to distinguish different tones. Smoothing over noise allows some more tones to show through, but with the tradeoff of reducing sharpness. In other words, noise reduction is good for dynamic range, but bad for resolution. The G1 ends up with a mediocre dynamic range score that is saved by the good dynamic range at high ISO. **Speed/Timing***Startup to First Shot (6.3)
*The G1 took 3.7 seconds to start up and take its first shot. If you're trying to capture a quick shot, this is an eternity. *Shot-to-Shot (5.3)
*The G1 will not shoot a burst in Best Quality mode. It will, however, shoot continuously in Full Quality mode, where it will take 2MB shots every 0.7 seconds. The G1 can shoot continuously until the card is filled, or can be set to only take five shots before stopping. However, this is tricky to set on the camera, and counterintuitive. When the "Cont. Shot" menu is set to "Off", and the drive button is set to burst, the camera will shoot continuously. When the menu is set to "5 Shot", the camera will only take five shots. The last option, the "5 Shot (Last)" mode, shoots continuously, but only records the final five images.
*The G1 takes 0.1 seconds to shoot when the shutter is already held halfway down and prefocused, and 0.6 seconds without being prefocused.
*It takes 2 seconds to process 3.5MB shots taken in Best Quality mode at ISO 80.
Bright Indoor Light – 3000 lux
We tested video noise levels and color accuracy under bright studio lights in the same manner we tested the still images above. In bright light, the G1 had extremely poor color accuracy, with a mean color error of 22.2, though the saturation level was good, at 104 percent. Noise levels were quite good, however, at only 0.47 percent.
*Low Light – 30 lux
*We also tested video performance in low light, where the camera performed worse. Color error was still very high at 21.9, and noise levels jumped to 2.41. The footage was also extremely underexposed, as you can see below. *Video Resolution
*Similarly to our still resolution test described above, we shot video of our resolution test chart at 1700 lux. The G1 resolved 193 LW/PH horizontally with 33.6 percent undersharpening, and 505 LW/PH vertically, with 94.1 percent oversharpening. These extreme sharpening levels indicated the image artifacting apparent in the video. Sharp edges had bright lines next to them, where there shouldn’t have been any. This is the second worst video resolution performance we've seen this year, just behind the Olympus 770SW. (100 percent crops)*
*To get a look at how the G1’s video mode handled motion, we stepped outside of the lab to shoot moving cars and people on the street. The overall frame looked very hazy and washed out, with ugly smearing caused by highlights. Yet the worst of it was the motion. Aside from making moving objects look very jerky, even when in the middle of the frame, the exposure changed so much that the video seemed to blink and flash, possibly even dropping some frames. This headache-inducing effect makes the video barely watchable, unless you have a large jar of Tylenol nearby.