Tuesday, February 3, 2009
D-SLR State Of The Art, Part II
What to know about the latest digital sensors, ISO and image quality
|
Since demosaicing (converting the monochrome sensor image to color) and sharpening require lots of computing horsepower, a case can be made that these are better done in your computer via RAW-conversion software than in-camera—just one benefit of shooting RAW images over JPEGs and TIFFs. RAW-conversion software is improving at least as rapidly as D-SLR technology, including noise-reduction performance. In fact, you can reprocess your older RAW images in the newer RAW converters for even better results than you got when you originally processed them, which is another great reason for shooting RAW images rather than JPEGs.
 | Sony’s DSLR-A900 and Nikon’s D3X share the highest pixel count among “35mm-style” D-SLRs: 24.6 megapixels on a full-frame sensor. The resulting smallish pixels hold the ISOs down from the “speed king” category, but they still produce respectable numbers: a normal range of ISO 100-1600 for the D3X, with a top setting of 3200; and a normal range of 200-3200 for the A900, with a top setting of 6400. While the sensors are similar, the image processing is quite different—the D3X even provides Live View operation. |
Balancing Noise and Sharpness
Noise-reduction methods all reduce image sharpness. Conversely, sharpening an image increases the appearance of noise. So D-SLR manufacturers have to do a balancing act between optimal noise reduction and good image sharpness. The powerful image-processing engines and noise-reduction algorithms do an amazingly good job of this.
Some D-SLRs automatically apply noise reduction when high ISOs are set or long exposure times are used. Some cameras let you switch noise reduction on or off as desired, and some even provide various levels of noise reduction you can use to suit specific shooting needs.
Another benefit of RAW images is that you can adjust both noise reduction and sharpening when you process them in your computer. You also can apply noise reduction and sharpening to JPEGs, but the results won’t be as good because most JPEGs have already been sharpened in-camera and are all lossy-compressed.
Whatever methods you use for controlling noise, we still prefer a sharp shot with noise over a blurry shot without it. If you need to increase your ISO to get an adequate shutter speed, don’t let concern about noise stop you.
Whatever methods you use for controlling noise, we still prefer a sharp shot with noise over a blurry shot without it. If you need to increase your ISO to get an adequate shutter speed, don’t let concern about noise stop you.
| The new Olympus E-30 has the smallest pixels of any current D-SLR, measuring approximately 4.29 microns square. Nikon’s D3 and D700 have the largest pixels among current D-SLRS, each pixel measuring around 8.45 microns square. Not surprisingly, the D3 and D700 provide much higher ISO settings (normal range of 200-6400, top setting of 25,600, compared to a normal range of 100-3200 with no expanded range for the E-30). What’s perhaps surprising is that the E-30 matches the top-of-the-line Olympus E-3’s ISO range despite squeezing 2 million more pixels onto the 17.3x13mm sensor. Again, improved technology makes it happen. |  |
Does Size Matter?
All other things being equal, big pixels are better than small ones, and more pixels are better than fewer pixels. So the big question is: Are more smaller pixels better than fewer bigger ones?
Let’s start with a little math. You can get a good estimate of the size of the pixels on a given image sensor by dividing the sensor’s horizontal dimension by its maximum horizontal resolution. For example, the Nikon D300’s 12.3-megapixel sensor measures 23.6x15.8mm and its maximum image resolution is 4288 x 2849 pixels—23.6mm divided by 4288 pixels equals 0.0055mm per pixel. (For an easier-to-handle number, you can move the decimal point three places to the right to get the pixel size in microns, 5.5 microns in this case.)
Nikon’s D3 has a 12.1-megapixel, full-frame sensor measuring 36.0x23.9mm and delivers a maximum resolution of 4256 x 2832 pixels—roughly the same number of pixels but on a much larger sensor. So, doing the math (36mm di-vided by 4256 pixels) gives us a pixel size for the D3 of 8.5 microns—much bigger than the D300’s 5.5-micron pixels. And indeed, the D3’s image quality is better than the D300’s, in large part due to its much larger pixels. (The D300’s image quality is excellent; the D3’s is simply better.) The D3 also offers much higher ISO capability than the smaller-pixel D300.
Page 3 of 5
2 Comments