A simple control for reducing Highlight Noise Reduction.

suggest a way to improve Neat Image
Post Reply
peterb
Posts: 5
Joined: Sat Mar 15, 2003 7:32 pm
Location: Taby, Sweden

A simple control for reducing Highlight Noise Reduction.

Post by peterb »

The company PictureCode has come with a new program called Noise Ninja. It uses a wavelet technology to reduce the noise. It was anounced today at http://www.steves-digicams.com/diginews.html It gives very good results. One sample picture from the demo with the name daisy.jpg is a yellow flower with a dark green noisy background. It is taken with a Canon 10D at ASA 3200.

I tested with Neat Image and the profile on the Neat Image webpage for Canon 10D, 3200 ASA and JPEG. The noise reduction was about the same but the center of the flower got much more blurred.

I tried with different filter settings but could not reduce the blur without getting more noise in the background.

Then I changed the Noise Profile. I set the upper 7 slides for green to zero and reduced the upper 7 slides for the red to one third of their positions. Then the blur of the flower disappeared and the result was about the same as with Noise Ninja. There are some red noise left in the yellow but I think the settings could be improved a lot.

It looks as Neat Image tries to reduce the noise to much in the lighter parts to get optimal results - or perhaps the used profile is not good for that picture. The picture can have been manipulated in some ways.

In this picture the details are in the highlight parts. There should be some simple way of reducing the noise reduction in the highlight parts. Now it is very time-consuming to do that.

The Noise Ninja also uses a Calibration chart with a lot of colors. Perhaps that makes a difference - especially if a lot of unlinear processing of the picture is done in the camera.

Regards,
peterb
NITeam
Posts: 3173
Joined: Sat Feb 01, 2003 4:43 pm
Contact:

Post by NITeam »

Peterb,

Thank you for the news.

My guess is that the profile you used was not very well matching the particular image you processed. That's why you saw too much blur in some areas. If you built the profile from the very same image (which I understand may be not possible with a small crop) the results should be better. This is just an example of how a 'generic' profile for a certain model of a camera may be not as good as a particular profile built for this specific camera you use. Every camera is different, don't forget this.

Hope this helps.
Vlad
Post Reply