So, You Get More Noise?

[mwfanelli2]

I have just read a very interesting argument on another site about the "high noise" of newly announced Canon G9 12 MP small-sensor camera.

There are samples on DP Review that run from 80 ISO through 200 ISO. Indeed, the 200 ISO shows some obvious noise in spite of the excellent resolution. Many are complaining, as I often did, about the adding of smaller and smaller photosites.

But here was an excellent point: When you compare two images at the same ISO, you must also compare them at the same resolution. Comparing a 12 MP image to one taken with 5 MP is silly as the 12 MP, noise or not, has much more detail. What you must do is compare 5 MP to 5 MP.

Now, I don't have a G9 hanging around to shoot at lower resolution (!) so I downloaded a 200 ISO image, the full pixel version, from DP Review and downsized it in a photo editor. Not the same thing, but a rough guage. Much to my amazement, the noise was substantially lower, the equal or better than a 5 MP P&S camera.

So, what do you think? Are the higher resolutions really for good lighting and low ISOs? The shots at 80 and 100 ISO are excellent. Should we all learn to take advantage of the lower resolution settings in marginal lighting and higher ISOs? In the film world so long ago, we readily did this without thinking about it.

I am saving up my pennies for a G9, maybe christmas. I am no longer afraid of high noise with high megapixels.

[Sebastian]

I agree with you to an extent. To me, the higher resolution cameras are marketed with the pretense that you can blow the images up larger, or crop much tighter, than you could before.

But then I think about the fact that almost no one I know prints above 4x6, and nobody ever takes the time to crop down to the area of interest, and realize that the whole thing is moot anyway.

I hate marketing spin.

[Photo-John]

I've been saying this for a long time. There is little or no value in looking at noise on a computer screen. You need a real world measure. That's why I've been checking noise for camera tests by making 8.5x11 prints. Noise is relative to the resolution of the image file. So you might view two files on the computer and they'll appear to have the same noise. But if one is a 10-megapixel file and the other a 5-megapixel file, the real noise is actually very different. Higher resolution effectively minimizes noise.