I'm not convinced of the inferiority of the DROID camera a as anything but software. Here are the best specs I could find on this, versus the iPhone camera:
DROID (supposedly)
sensor: Kodak KAC-05020, 1/4", 5Mpixel, 1.4µm² pixel size,
TrueSense color filter, ISO to 3200
Kodak revolutionizes image capture with new CMOS sensor
lens: f2.8, autofocus
iPhone
sensor: Micron
MT9D112D00STC, 1/4", 2Mpixel, 2.2µm² pixel size
standard Bayer color filter, 10-bit ADC
http://www.framos.eu/uploads/media/mt9d112_1__1__01.pdf
lens: 4 Element Plastic, f2.8 (fixed), fixed focus
iPhone 3GS
sensor: Omnivision OV3630 1/4", 3.2Mpixel, 1.75µm² pixel size,
standard Bayer color filter, 10-bit ADC
lens: f2.8 (fixed), autofocus
Now, first, the reason none of these are your real camera. I have a Nikon D70 around here somewhere... hardly a new model. It has a very large 6Mpixel sensor, with a pixel size of 58.7µm². A typical pocket P&S camera will have a sensor around 1/1.8"-1/2.5", with pixel sizes around 4µm², give or take (that's about what you'll find on a 10Mpixel camera). These cameras are also typically using 12-bit or 14-bit DACs, not just 10-bit... more dynamic range. If it's not obvious, sensor size has a huge impact on the sensitivity of the camera... that's the area that collects light for each pixel.
Of course, aperture (the amount of light that can enter the lens) is also important, but all three phones offer f2.8 lenses. That's rather weak for a non-zoom, but obviously, size is critical in a phone camera.
From the chart, it appears that the DROID
should off noticeably worse low-light performance than the iPhone, if not necessarily worse bright light performance. This may be true... it certainly seems to be an issue today. But it may also be based on software.
Kodak makes big claims for their TrueSense technology, which is actually both the sensor and the color filter. Most CMOS sensors use an enhancement mode, NMOS sensor arrary, but Kodak is claiming their use of a PMOS, depletion mode sensor array improves things, by dramatically lowering noise (largely based on pixel to pixel crosstalk). I'm not yet convinced. The difference is that the "majority carrier" in an NMOS device is the electron... that makes sense. But in PMOS, it's the "hole" ... technically, the absence of an electron. It's weird, but it really does work out.
The other thing in this technology is the color filter. Each sensor in any normal CMOS or CCD can only sense in monochrome, not color. To get color, you typically add filters. Dr. Bayer, a guy from Kodak many years ago, came with this is pattern (where each letter is one pixel):
RGRGRGRGRGRGR
GBGBGBGBGBGBG
RGRGRGRGRGRGR
GBGBGBGBGBGBG
and so on. Thus, 1/4 filters are blue, 1/4 red, and 1/2 green... green being visually the more important color. Software interpolates the extra colors from the neighbors to create all RGB pixels. The Kodak sensor changes one green filter to white (eg, no filter), and treats that as uncolored luminance information only. That's a powerful idea... each filter cuts out 2/3 of the light, so changing 1/4 of the pixels to white should effectively give you twice the luminance capture without a big difference in color information.
Of course, software is a big deal here... ensuring the sensor isn't always shooting at ISO3200. According to the photo set I shot last weekend, I'm always at f2.8 (probably fixed there), and I see auto-ISO set at ISO57, ISO65, and ISO200, with exposures between 1/1030 sec and 1/10 sec.
Assuming I got the right info on the sensor, this sensor is also in the Motorola ZN5.
Motorola ZN5 test photos
Those shots might be just a bit better than anything I've seen on the DROID... yet. But maybe there's hope. The noise zooomed in looks of a similar nature (some JPEG artifacts, but other noise, too), but it's WAY more pronounced on the DROID. Maybe this improved via tweaking... one can hope.
