View Single Post
Old 11-17-2007, 09:24 PM
AdrianW AdrianW is offline
Join Date: Mar 2003
Posts: 1,596
Default Re: Neat Image vs Noiseware Community

Interesting. I agree about only applying NR where needed; but I disagree about the resizing bit.

Noise at 100% pixels is often somewhat predictable, due the Bayer filter pattern and interpolation algorithms used by the cameras, so I'd expect the algorithms used in NR are trained to recognise certain patterns as noise. If you resample down then that recognition might not work. Of course I'm only going on the noise that I see in images, and how I'd code to repair it; not from any inside knowledge of the products in question, so it's possible that your approach may actually be best. My gut feeling is that it possibly isn't though.

I'm using NeatImage in case anyone's curious. I just apply to a whole layer at original size, and then use an eraser to knock out any areas where I think NR shouldn't be done. This also has the advantage of making it easier to hit edges accurately (at 3888px wide an edge could be five pixels, but 1400 is could be just one or two, which would be harder to mask). As I say that's just my approach, there are no guarantees that it's the best!
Reply With Quote