Sensor sizes seem to fall into two distinct size categories: large DSLR sensors and a second class of much smaller sensors found in compact cameras. Most serious compacts, those with full manual control, seem to have sensor sizes at the upper end of the group of small sensors.
The diagram below illustrates the relative sizes of the two sensor groups. The cropped DSLR sensor group is shown in orange tones, the compact sensor group in blue tones.
As evident in the diagram, there is a big gap in sizes between the largest compact sensor, the now rarely used 2/3” (8.8 x 6.6mm) sensor, and the smallest DSLR sensor, the 4/3 (18.0 x 13.5mm) sensor. This gap is much larger than the size difference between individual sensors in each of the two groups.
Sigma’s DP1 has used a sensor in the first group of large DSLR sensors. The price of having a very large sensor in a compact body is a not very bright, nor versatile, fixed 28mm lens. All other serious compacts to date (late May 2008) use the much smaller sensors typical in compact cameras. Except for Ricoh’s GRD I/II, all these cameras feature versatile zoom lenses that are at least fairly bright at the beginning of their respective zoom ranges. Because their sensors are small, noise and noise reduction remain an issue in all but the best lighting conditions.
The question is why there isn’t an intermediate size of sensor for serious compacts: a size that fits between the DSLR and compact sensor sizes. These sensors could offer better image quality than those sensors in the compact group while they wouldn’t be so large as to preclude zoom lenses. An intermediate size would be 12.0mm x 9.0mm, which I’ll call the “S/C” sensor. This hypothetical sensor has approximately half the area of the 4/3 sensor and about twice the area of the 2/3” sensor. This sensor should allow for reasonably compact lenses and camera bodies.
Making a sensor for just one class of camera or model seems feasible if we take Panasonic’s LX2/D-LUX 3 as an example. Its unique 16:9, 8.8mm x 4.8mm, was only ever used in that particular camera model. As of May 2008, Fuji’s 2/3” Super CCD HR sensor is also only used in one model: the FinePix S100fs.
So perhaps a new intermediate sensor size could provide a better balance between image quality and a compact form factor.
Read More......
When the Rebel XT was released with 10MP, there was an immediate reaction in the DPReview 10D/20D/30D forum. How could Canon put a 10MP in the entry level body and leave the 30D stranded with just 8MP? In time, it became clear that the extra 2MP adds very little to resolution while the decreased pixel size, or rather sensel size, adds a touch more noise at high ISOs. Needless to say Canon puts a number of capabilities in the X0D series that don't make it into the Rebel series. Higher frames per second in burst, better autofocus, more metering modes, larger, brighter viewfinders, and more robust shutter mechnisms to name just a few. Yet with each generation, the feature gap between these classes shrinks. With the original Digital Rebel, Canon literally handcapped the camera in software in order to preserve 10D sales. The latest Rebel gets spot metering, a sophisticated AF system, larger OVF than before (though still mirror rather than prism), and even highlight tone priority! All this plus 12MP (the 40D has 10MP) and contrast detection AF in live view! Thus the 40D user outcry at having been jipped in the megapixel race was expected and even somewhat understandable. What's interesting is that just as multiple threads are popping up in the 40D forum, with one new 40D user actually crying about it, prospective Rebel XSi buyers are complaining about the pixel packing in the Rebel forums. If we assume that the Rebel Xsi and 40D have similar sensor technologies and equal quality low pass (AA) filters - and neither of those is truly a safe bet at this point - then a couple conclusions follow. In good light, with a good lens, the Xsi should be capable of capturing a touch more detail than the 40D. In low light at high ISO, the 40D ought to fare a bit better. Neither of those advantages will be very significant, and the difference in dynamic range corresponding to such a small change in sensel size is negligible.
Assuming Canon hasn't put a crappy AA filter in the Rebel XSi, it should be pretty sweet for focal length-limited shots of distant birds in flight. If light is good, an XSi user should be able to get significantly more pixels on a distant bird than a 1D III user with the same lens. The particular lens that springs to mind for this application is the 400mm f/5.6L, which should pack enough resolution to take advantage of each one of those little pixels. Click here for a great illustration of the "crop factor" at work, and keep in mind that the advantage shown for the 20D compared with the 1D II on that site should be even greater with the XSi compared with the 1D III. Highlight tone priority is another welcome addition that should come in handy with bird photography, where white feathers often present an exposure challenge. I think the XSi is exactly what I've been waiting for in a wildlife camera.
Read More......
