I must confess. Although I love my Nikon Coolpix P5000 and the images that it produces, the small sensor associated with P&S cameras cannot generate the silky smooth images one can get from the larger DSLR's. And, although I’ve been happy with the output from P5000, it produces an extremely noisy image (even at ISO 100) compared to the above. So I typically resort to post-processing my digiscoped images with Photoshop CS to enhance image quality. I’ve been extremely pleased with my results, but have noticed all too often that my images, when pixel-peeped, still retain an annoying amount of chroma (color) noise in the form of red and green pixel splotches in blue sky or in the dark feathers of birds. After doing a bit of research on the net I came to the conclusion that I need a more aggressive noise reduction software package than Photoshop can provide alone.
Initial searching convinced me that Noise Ninja was the package to get, so I downloaded the trial version from their website. It retains almost full functionality, but places a grid over processed images until the license is purchased. I spent a better part of a day playing with it, and was simply amazed at how well it performed at removing chroma noise from feathers and sky. I was able to produce sharp, low-noise detail in my subjects while generating smooth backgrounds with no noise. Very impressive! Suddenly, the old way of Photoshopping my images looked somewhat disappointing by comparison.
Ready to pull the string on a license I decided to search a bit more on NN to find out what other photographers liked about it. One thread polled users to vote between NN and NeatImage. I was surprised to find that most everyone seemed to be familiar with both, but now prefer a package called Noiseware Pro. Intrigued, I downloaded their trial version and decided that I should do a comparison. So, for the next couple of days I compared Adobe PS CS, Noise Ninja, and Noiseware to see which package gave me (personally) the best results. I downloaded the trial version of Neat Image, and found it to produce output very similar to Noiseware Pro. But I haven't played with it enough to figure out the correct settings to compare w/ the previous two. So my opinion of it will have to wait.
I pulled a couple of images from my Coolpix P5000 and a few from my Nikon D300 so that I could evaluate the packages on both P&S and DSLR. I performed a magnetic selection in PS to highlight the foreground, then I copied the selection to a separate layer. I then copied that layer into separate copies of the photos – one for Noise Ninja and one for Noiseware to process. This way I could selectively sharpen the same foreground and smooth out the background. I would then flatten the images to see how well foreground/background images blended.
For Photoshop workflow I used an Unsharpen Mask (Amount=50%, Radius=2.0 pixels, Threshold=0 levels) on the foreground, and a Gaussian Blur (Radius=1.0 pixel) on the background.
For Noise Ninja, I used all default settings and changed only the ‘Luminance’settings (Strength=5, Smoothness=5, Contrast=10, USM Amount=150, and USM Radius=1.0) for the foreground. The background got the following settings: 10, 10, 10, 0, 0.4.
For Noiseware I used the default settings with the following settings on the foreground:
Noise Level (Luminance=0, Color=0)
Noise Reduction (Luminance=0, Color=100%)
Detail Protection (Luminance=0, Color=0)
Detail Enhancement (Sharpening=5, Contrast=0, Edge Smoothing=Normal)
For the background I used the following: Noise Level (0,0), Noise Reduction (100,100), Detail Protection (0,0), and Detail Enhancement (0,0,Normal)
More experienced users may find fault with these settings, but I was merely trying preserve as much detail and reduce noise as possible in the images I processed. I found that Noiseware could easily produce over-contrasty images, while Noise Ninja tended to generate more high-frequency noise (tinier-grained noise). Hopefully this makes sense when you examine the images.
Rough-legged Hawk
I digiscoped this bird last May at Pt. Mouillee a week before (sadly) I found its carcass below the same tower (a very late migrant that must’ve been ill). The Original image from the Coolpix P5000 was processed using Photoshop CS, Noise Ninja, and Noiseware Pro. At 200% image size, side-by-side comparison showed pixelation and chroma noise in the Original. PS added more detail to the bird and reduced background noise, but chroma noise is evident in both bird and sky. NW removed chroma noise from the bird and sky, and improved bird sharpness just a touch. NN removed chroma noise but showed more graininess in the bird detail.
At 100% image size, NW and NN produced very nice results, with NW getting an edge in image cleanliness. At 300% image size, PS (at right) showed significantly more pixelation and chroma noise in feather detail while NW effectively removed both. When comparing NW with NN at 300% both removed chroma noise from the feathers, but NN (right) showed a bit more pixelation that could be removed with a bit more tweaking of the software package. The edge in this comparison goes to Noiseware Pro.
Indigo Bunting
This bird was digiscoped w/ the P5000 in late June. Although it’s a bit on the soft side I wanted to see how the software handled ‘greens’ and ‘blues’. Here are final versions of the Original, Photoshop, Noise Ninja and Noiseware Pro. At 100% image size side-by-side comparison produced the following: The Original (top left) is not bad, but shows a bit of chroma noise in the green background and in the bird. Noise Ninja removed chroma noise, but left an annoying edge artifact in the bird after layers were blended. Noiseware showed no artifacts and produced a nice clean image while PS shows significant chroma noise and grain in the bird. Edge to Noiseware.
Forster’s Tern
This flight shot was taken with the Nikon D300 and Sigma 400mm f/5.6. The Original and PS versions of the image show significant chroma noise in the sky, which was effectively removed by both NN and NW. Comparing NN and NW at 100% it appears that NW shows a bit more contrast, but otherwise it’s a toss as to which is better.
Black Tern
Finally, I wanted to see how the packages handled contrasty, severely backlit images, or gray-day pics, like this Black Tern flying directly at me. The original has been already Photoshop’d, so I also wanted to see what
NN and
NW could do to clean up PS. When compared side-by-side at 100% the tern shows a great amount of noise, which is expected on a gray day. The camera has to try to create a gray sky from red, green and blue pixels, and the resultant images therefore show many of these pixels. Both Noiseware and Noise Ninja did great jobs at reducing this noise, with NN producing a slightly cleaner image at the expense of perhaps a bit of detail. Again, it appears to be a toss-up.
ConclusionI’d be happy with either package. Both Noise Ninja and Noiseware perform wonderfully as Photoshop Plug-ins and improve on anything that I’m able to do with PS alone. Both NN and NW are great at eliminating chroma noise and producing smooth backgrounds that makes Canon and Nikon images so easy to look at. NW tends to produce higher-contrast images so the slider should be used with great care. NN tends to generate a bit of a watercolor effect with excessive noise reduction, and resulting images can look a bit overprocessed. I found that I preferred Noiseware overall as it seemed to handle all of the imaging situations I threw at it w/o problem. I would encourage anyone else considering these packages (and Neat Image) to download the trial versions and compare them for yourself. I would welcome any additional comments anyone has to offer, and will publish them as long they’re in good taste. Thanks for playing!
A number of digiscopers have commented that they seem to get better images when the scope is slightly defocused, thus allowing the camera to find the right focal plane. A couple of Birdforum threads have addressed this issue (1), (2). Since I sometimes have issues trying to keep focus in the field, I decided to perform some experiments under controlled conditions to determine the limits of 'my' digiscoping setup in terms of focusing/defocusing. I emphasize 'my' because I suspect that results could be completely different for other scope/camera setups.
