Trevor
0
- Joined
- Jul 17, 2009
- Messages
- 4,386
- Points
- 113
As far as noise goes: I find it to be better with cmos sensors in dark conditions, but worse in fully lit ones. It's by no means terrible, but if you take a picture in brought daylight (iso 100, f/8, 1/500s) the cmos sensor produces some noise in, for example, the blue of the sky.
I think my knowledge of CMOS vs. CCD's in the context of P&S cameras is limited by the fact that nowadays I don't take note of anything but full-frame PJ/sports bodies.
Here's an image shot with a Canon S90 (which has a CCD sensor) that produces grain in the blue sky:
http://www.dcresource.com/sites/def...anon-powershot-s90-photo-gallery/IMG_0298.JPG
I think it's more of a pixel size issue than a CMOS vs. CCD issue.
The Canon S90 has a sensor in it with an area of 43.3 square millimeters, compared to 864 square millimeters of a full frame camera. That's like cramming 200MP onto a full frame sensor! That's a LOT of pixels... SMALL pixels.
Then, beyond that, you have a Bayer filter over the sensor. In a CCD or CMOS (as opposed to film or that Foveon nonsense) sensor, each photosite can only read one color. Check out the image below:
As you can see, all color elements arrive at each photosite, but only one channel is allowed to hit the sensor. This leaves you with a three component tiled pattern: 50% green (green + luminance), 25% red, and 25% blue. Each final pixel you see in your JPG or RAW file is the product of a processed called demosaicing where the Bayer mosaic is processed and a final image pixel is generated from sampling at the nexus of four of the photosites - 2 green, 1 red, and 1 blue.
So, when you take a photo of something that appears so uniform, you're going to run into grain in a P&S for two reasons:
- There are only so many photons to go around. In a given time, one pixel might see more blue photons than another. When your pixels are so small, this problem is compounded.
- Only 1/4 of the pixels on your sensor even pay attention to blue light - plus the green pixels which will also see uneven distribution.
When you consider both of those limitations, the presence of a grain pattern in something like a blue sky isn't really all that surprising.
I'm not sure why the iso value cannot be lowered below 100 (even when set to auto) on this camera though - it could be a technical limitation of the sensor design.
Base ISO is basically just reading the values off the sensor without amplification.
My 1DX will do ISO 100 - 51,200. So, if you bumped the base ISO down to 25 and used the same amplification hardware, you'd lose two stops at the high end of the ISO range. So the final range would be 25 - 12,800. If you kept 51,200 as the high, it would end up being overwhelmed by read noise.
If you want to shoot in bright light with a fast lens, ISO 100 and a shutterspeed of up to 1/8,000 is typically just fine. A friend of mine rented an 85 f/1.2 and we took it out to shoot in broad daylight - I could usually get away with a high shutterspeed. But when I couldn't, my friend had ND filters on-hand - because when your bag doesn't contain any sub-$1,000 lenses, not spending some money on a few nice ND filters is pretty inexcusable.
Some camera bodies offer an "L" mode (just like "H" ISO modes) where the image is digitally brought down, but most people don't bother with that for the above reasons - and on grounds that it is digital, not analog.
Trevor
Last edited: