Pages: [1]   Go Down

Author Topic: Topaz DeNoise revisited  (Read 1363 times)

David Sutton

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1345
    • David Sutton Photography
Topaz DeNoise revisited
« on: November 22, 2010, 10:28:28 pm »

After reading Mark Segal's article on Topaz Denoise in June I thought I'd look at the latest version to see how far I can push the 5D Mark II. There are a couple of useful videos on YouTube on how to use the software.
Image 1 below is the test shot done with the 70-200 lens at f8 with the flash off to camera right and all other lighting off. The book sitting on top of the bank note was to get part of the background into deep shadow.

It became apparent that when exposed to the right, the mid and three-quarter tones from this camera were fairly good even at high ISO. Above 1600 there was a red colour cast in the shadows which is apparently common in Canon files and which I partially removed in the software and finally took out with the black point eyedropper in curves and then dropped the opacity on that layer down. The ISO 100 shot had some noise reduction in Lightroom 3 and the sharpening left at default, the other shots were converted to tiffs with the sliders zeroed and then sent to Topaz in Photoshop.
Images 2, 3 and 4 below are some crops at 100%

What really interested me is what they would look like in print, so I dropped the exposure on the files a little in LR to match what I saw when photographing, and sent them to an Epson 3800 through Qimage on Ilford Gold Fibre Silk at A2 (16.5” x 23.4”).
Here are the results:
ISO 1600: This print had marginally (but visibly) more detail than the 100 ISO print done in Lightroom. I am fairly confident that if I ran the 100 shot back through LR and added more capture sharpening I could get them to match.
ISO 3200: A lot of photographers have looked at this print and they confirm what I saw on screen at 100%. There is no visible difference between this print and the one at ISO 100.
ISO 6400: If you look very closely you can see a small loss of detail in some parts of the print.

I won't hesitate to go straight to 3200 ISO now. A far as 6400 goes, well in the semi dark I'm not doing macro work. The subject matter that I want to tackle in very low light can take a small loss of detail. Provided I can nail the exposure and avoid heavy cropping, for prints up to A2+ I see no reason not to go for it.
I am also somewhat nonplussed. My experience is that reducing the contraints on any art form does not necessarily produce better art. The technical revolution over the last few years has delivered some technically outstanding work, but it will be very interesting to see what it delivers that moves the heart.
« Last Edit: November 22, 2010, 10:32:22 pm by David Sutton »
Logged

Lightsmith

  • Full Member
  • ***
  • Offline Offline
  • Posts: 197
Re: Topaz DeNoise revisited
« Reply #1 on: December 05, 2010, 01:23:37 am »

I tested a number of NR programs a couple weeks ago along with the new NR function in CS5. CS5 was nearly as good as the best NR in preserving both detail and more important, tonality. DeNoise was outstanding but slow. The latest version is significantly faster to the point where it is usable.

I used a face as it is easier to spot problems with lose of tonality with a face and skin tones than any other subject. Neat Image is also very good but not quite as good as DeNoise. Very glad they improved the processing speed of DeNoise.

Logged

deejjjaaaa

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1170
Re: Topaz DeNoise revisited
« Reply #2 on: December 05, 2010, 01:39:54 am »

Neat Image is also very good but not quite as good as DeNoise.
image w/ Denoise is downsized vs image w/ NeatImage - why ?
Logged
Pages: [1]   Go Up