After hearing so many questions and complaints about noise at higher ISO, I wanted to get a better idea of how bad the noise is (or isn't). For me, there is a noticeable improvement over my previous camera, so that is the comparison I'm interested in. Others want to compare the A100 to other dSLR cameras, but I am not doing that comparison here. (Maybe one day.)
First, an ISO 400 photo from my DSC-V3.
Now for the A100.
The A100 seems to underexpose a bit more as the ISO goes up. I'm not sure if this is generally true, but I believe I have read that others have also noticed this behavior. I increased the ISO 800 photo +.3 EV just to keep the brightness in a reasonable range, and it was still darker than the photo from the V3.
To try to keep the comparison fair, I downsized the A100 photos to about 80%, in order to match the pixel count of the V3. (Averaging the pixels will naturally reduce the noise somewhat. I don't consider this "cheating" the test -- the A100 has more pixels and it's fair to use them. Viewing both at 100% would unfairly show the A100 photos in a greater magnification, in effect magnifying the noise relative to the other camera.) I then cropped similar sections from each photo.
However, I also used DRO+ in the A100 settings, which, in theory, should enhance/brighten the shadows, which could increase the noise level. Also, due to the different design, the depth of field is shallow on the A100 using the wide aperture, in the attempt to gather more light and have a faster shutter speed.
Even at ISO 400, not only is noise visible, but there is lost detail due to noise reduction, in the V3 photo. It's a usable photo, but the A100 photos are better -- even the ISO 800 one! I think this is to be expected, since the sensors in DSLR cameras are generally much larger than in almost all non-SLR cameras.
At the same ISO, with the 28mm, f2.8 lens I used, the A100 had a much faster shutter speed.
(Just for another quick comparison, I tried the kit lens at about 28mm [f5], and it had a shutter speed of 1/13 with a +.3 exposure at ISO 400. So, speed-wise, using the kit lens is similar to the V3.)
While I'm sure this is not a perfect test, I wanted a somewhat real-world situation where one would want to use higher ISO -- lower light, indoors. What I didn't realize with the original test was that exposure, by itself, really doesn't tell you anything. The manufacturer sets the gain on the amplifier for the signal from the sensor to approximate a standard, but it can vary. So, evaluating which camera is "faster" in shutter speed is largely irrelevant, except as a point of comparison. If you can match the exposures to get a fair comparison, you then have to decide which result is noisier. What this test doesn't show is precisely how much faster the shutter speed of the A100 can be given a result that has similar levels of noise.
What this test does show is that the A100 does have a faster shutter speed with superior results, this is proof enough for me that the A100 has a significant advantage over even higher-end non-SLR cameras. Add Super Steady Shot, and the large low-light advantage is even greater. The main negative to the DSLR is that it may be harder to get everything you want into focus with a wide aperture because of the more shallow depth of field; change the aperture, and you start to lose the speed advantage. Stick with the kit lens, and you also lose the speed advantage, although the ISO advantage is still there, as well as the additional quality from the larger sensor. Getting a lens with a wider aperture is a big plus for low light photography, though, as you can always use a faster shutter and/or lower ISO (for more quality). Given sufficient light, the differences between cameras are not going to be as great.
One line of thought is that the A100 has little noise reduction at high ISO in order to preserve details, while some other DSLRs (and most point and shoot cameras) will perform noise reduction (with inevitable loss of some detail) automatically. Even from this simple example, you can see that along with the noise, a lot of detail is still preserved -- I was (barely) able to make out the number printed on the blue pamphlet in the ISO 800 photo (in the original copy). For best results, use RAW for higher ISO in order to have more control over noise reduction in the computer, as well as making it easier to add saturation, etc., as desired.
I'm pleased with the ISO 400 performance, and I have a hard time even seeing the noise in the ISO 800 photo at a glance, looking towards areas of solid color for a noise pattern. I think I can use ISO 400 without worry, and will probably reserve using ISO 800 for when it's necessary, but I see no need to obsess over the additional noise. Setting ISO to Auto, the A100 will only use ISO 100 to 400, and not choose 800 (at least for later versions of the firmware), so I may end up using the Auto setting most of the time and enjoy the responsiveness of the camera.
Latest changes March 29, 2009
Original version created July 24, 2007