Akvis Noise Buster
|
||
| Always skeptical, ever hopeful, we downloaded the trial version of Akvis' Noise Buster and went to work on some photos. Tingles of anticipation turned to promise and we forged on - with benefits.
Noise remains the most consistent problem in digital photography, and is especially present in high ISO shots, as well as every consumer model on the the market today. Of course if you're just printing 4x6 snapshots from a 12 megapixel point and shoot then you probably won't notice how bad the noise really is and go on your merry way. For those who favor enlargements the problems expand with the size of the print, and exponentially I might add. We found that Noise Buster worked best when using separate channels (Red, Blue, Green) for selective filtering rather than running it over the composite (RGB) image. Adobe's Photoshop CS2 also favors this approach, and Noise Buster installs as a plug-in to CS2 so we used it that way. As an outdoor photographer noise reduction is a mixed blessing. I'd like my skies to be clear and blue without smearing the foliage or leaving halos around the horizon. There's no one technique that works all of the time so I use a combination of selections, masking, working in layers and channels to achieve what minimal improvements I can muster. Noise Buster proves to be a very handy tool and does a better job than the native noise reduction in CS2. In fact by simply viewing the file sizes on the images below you can see that CS2 applies a linear filter to the image. Compare the file size of the crops blue sky and pink clouds: First Pass = 109 kb and 109 kb; Second Pass = 78 kb and 78 kb; respectively. So there's really no discrimination as regards the detail in the photo, just a general scrubbing of the image. |
|
|
| Noise Buster not only does a better job of scrubbing out the noise in the blue area of sky (upper left of photo) you can also see the difference in the pink clouds. Just download the crops and zoom to 400% and compare the CS2 and Noise Buster images. Further compare the file sizes: First Pass = 75 kb in the blue corner, and 88 kb in the pink clouds; Second Pass = 42 kb and 62 kb respectively. Why is that difference important?
There's really nothing in the blue sky but noise. In fact from the angle of the sun the light at that point is polarized. So there's no detail to add to the file size, just noise. But in the crop of the pink clouds there's much more detail and thus more information therefore the file should be larger. Not so for CS2. The file sizes are the same for each pass. Conversely the images processed by Noise Buster do show an appropriate difference in file size, and that leads us to believe that Noise Buster does indeed apply more discrimination to the subject matter and works in the manner described on their website. When working on composites, or the full RGB image, Noise Buster faired less well by comparison in the the default or auto mode, and tended to smear foliage. This required considerable tweaking, cutting the smoothing back to near zero, and basically a lot of trial and error before a satisfactory result was achieved. This is no doubt due to different amounts of noise in the respective channels. So if you are applying the same amount of filtering to all channels you are bound to overdo one. This makes the Standalone Home version that we evaluated less attractive. The first photo chosen for representation (sunset) offered enough detail for evaluation, and yet also offered a a plain area in the sky to evaluate the actual amount of noise where no detail was present. In actual practice you would need to protect the water from the noise reduction filter while smoothing the sky, and then reversing the mask to work on the water where the detail lies. We didn't bother. Yet it is easy to see that Akvis has an edge and works selectively to remove noise rather than just scrubbing the entire image. The Water Lily offers more detail and complexity. You may note that Noise Buster does suppress some minor detail in the petals, but then our focus during filtering was the center of the flower. Most of the noise was in the blue channel which also provides shadow detail in the petals. The workaround is to employ masking techniques which would also apply to CS2 though it tends to save detail over noise removal. In the end it's a matter of preference, and you will always sacrifice some detail when filtering noise. Please yourself, it's your photo. Our conclusion is that Noise Buster works very well, and though we pitted the home version against a professional grade program it appears to offer an advantage especially if you can work in Channels, and/or deploy masking techniques to assist in selective filtering. We found that setting the Smooth level to 5 or less is necessary when filtering foliage, and it is better to to filter twice at a setting of 2-3 than once at a higher setting. It's not a bad idea to nibble away at noise, regardless of the program you're using, and Noise Buster is no exception. Give it a Try. |
||
|
|
||
First Pass Using Channels for Selective Noise ReductionUpper Left Corner of the Image |
||
Original Image |
Akvis Noise Buster |
CS2 Reduce Noise |
 132 kilobytes |
 75 kilobytes |
 109 kilobytes |
|
|
||
Clouds with Highlights |
||
Original Image |
Akvis Noise Buster |
CS2 Reduce Noise |
|
|
|
|
|
|
||
Second Pass Using Channels |
||
Original Image |
Akvis Noise Buster |
CS2 Reduce Noise |
|
|
|
|
|
|
||
Original Image |
Akvis Noise Buster |
CS2 Reduce Noise |
|
|
|
|
|
|
||
Water LilyCS2 Top Row and Noise Buster Below100% Crops ~ Noise Suppressed by Channel and Smart Sharpening Applied Equally |
||
 |
|
|
|
|
|
Original Images |
||
|
||
Akvis Noise Buster |
||
Freeware 200949 Reasons Home Page |
||
