In an ideal world, we all would replace our digital necessities and toys at least once a year. Digital cameras should be updated annually, of course, if only to keep up with the technology.
But unfortunately, some of us fall behind by a year or more, and as a result are stuck with yesterday’s cutting-edge technology, which can be today’s handicap.
I have a digital camera that I love in most respects. It’s an Olympus C-750 10x Optical Zoom camera. The framing is accurate through the LCD, and the manual controls allow me to control most everything including aperture, speed, and white balance. What this camera lacks is high resolution (it has only 4+ megapixels, relatively low by today’s standards) and camera RAW compatibility with Photoshop. What that means is noise can be too easily introduced into the photos, especially in low-light situations.
But because I haven’t wanted to invest in a new camera, I had to work around these limitations. Here’s one lesson learned.
Turn Off That Noise
Recently I hired a wonderful photographer (Robert Lisak) to take very high-resolution digital portfolio shots of my “mixed-media constructions” — life-sized figure sculptures made out of junk.
The photos are great (to see them, check out my Web site — all are Lisak’s photos except those of “Oil Spirit” and the first “Grandma Lil” image), but when reviewing his contact sheets I realized that I was missing a frontal shot of one of the pieces — my Grandma Lil. Although Rob did take some shots form the front, they weren’t from the angle that I needed; the real grandma was short, so I wanted the camera to slightly look down at the sculpture, not shot from her “eye level.”
Before the sculpture needed to be returned to its owner (my mom), I decided to take a stab at taking the missing photos myself with my Olympus C-750. With sub-professional lighting, backdrop, and camera, I did get a shot from the right point of view (see Figure 1). However, you get what you pay for, and at 100% zoom level, it was filled with junky noise
Since many of us with these ancient cameras (mine is circa 2003) do end up with photos we could use if it wasn’t for that junky noise, I decided to try to “fix” the photo rather than set up the shoot again. Bear in mind that this photo is particularly difficult to fix because the sculpture contains many small wires and screens against a gray background, guaranteeing that it would be hard for a computation to differentiate between detail and noise (see Figure 2).
Using Google, I searched for “photoshop digital camera artifacts filter” and came up with pages of options. My first download was a set of Actions created for Photoshop 6, so they didn’t work fully (I got some error messages). Some other techniques seemed too labor intensive. However, what I did find is the NeatImage Photoshop plug-in filter, the Web site of which shows some really impressive demo images. I decided to give it a try.
Let Me Demonstrate
NeatImage is available for Mac and Windows. I downloaded the Mac demo, the User Guide in PDF, and the profile for my camera. The NeatImage ReadMe file notes that the filter will only work if your noise is uniformly applied throughout your image, and the demo software will only work on 1024-x-1024-pixel images or on the upper left 1024×1024 section of your image. I tried applying the demo in sections, and that was good enough for me to determine that it was worth the purchase price of $39.95. If you have something that needs de-noising (ok, that’s probably wasn’t a word until now), try the demo to see if it will work for you (NeatImage also offers a two-user license for $69.95; for larger-site licenses, contact the company).
The installer adds a new plug-in filter to your Photoshop application (see Figure 3). In Photoshop, open your photo containing the excess noise. Before running NeatImage, I highly recommend that duplicate the image layer first (drag it to the New Layer icon). Now you can apply NeatImage to the duplicate of your image by choosing it from the bottom of your Filter menu (Filter > NeatImage > Reduce Noise). You must begin by choosing a noise profile. You currently can acquire a profile one of two ways. You can build a new one based on your image, or you can download a ready-made profile for your camera in its current working mode, such as ASA 400 (many cameras are already profiled on the NeatImage Web site). Once you do have the profile loaded, if you have a series of pictures shot that include the same noise situation, then you can use the same profile for the shots in the series.
Once you have a profile loaded, you can then preview its results. The preview updates on the fly as you make custom adjustments. After building an Auto profile, I found the default results in most areas were pretty impressive, although there was still some junk mixed in with the details, and there was certainly some loss of detail where junk was smoothed.
The power of NeatImage unfolds as you learn to adjust it to your specific image. The parameters are actually “level” adjustments; think of them as thresholds that tell NeatImage to be more or less sensitive in guessing what is noise and what is an image detail. The key to most fixes is likely found within the “Noise Filter” settings. In my read-through of the manual I learned that in many images the most important aspect to learn was the Noise Reduction settings — in particular, the adjustment of the Y slider.
I ran the filter with different settings to a few duplicates of the image and then combined them using layer masks. I wrote up what I did for this article, and then sent the file and my results to NeatImage to evaluate. It turns out in the case of my image, there was a much better and simpler solution. With Vlad’s professional evaluation of my image, it turned out that the photo had “strong high frequency noise that required some manual adjustment.” I asked Vlad at NeatImage how he knew which settings would work? And he replied that the Windows version of the software gives you more live feedback (see Figure 5), and indeed it does (the Mac version will be updated to include the added features).
Following NeatImage’s instructions, I merely clicked the Default Settings button to reset my adjustments then increased my Noise Filter Settings > Noise Level > High Frequency slider to +75% (see Figure 6). The result was miraculous (see Figure 7).
No More Noise
For the final touchup of this photo, I used only the NeatImage fixed version, with a layer on which I created a smooth gray backdrop for the figure (using Clone Stamp and Airbrush tools and some Gaussian blur), and Levels Adjustment Layers to adjust contrast and lighting (see Figure 8).
If I had trusted that the results could actually be this fabulous from the start, I would have simply spent more time experimenting with the various sliders and adjustments. Obviously, for now those on the Windows platform will experience a bit less of that groping in the dark feeling when experimenting. Nonetheless, with the Preview in the filter being so good, even an expenditure of a half hour or so in the dialog box would have saved me from the time spent to set up the shot again.
The very good news is, if you have a batch of photos shot with the same noise throughout, once you figure out the settings for one, you can simply apply that filter to the next photo with equally magic results, or save the adjustment for another session (see Figure 9).
If you do have that wonderful spontaneous shot never to be experienced again, and that shot is riddled with nasty JPG noise, the value of the NeatImage filter will indeed be priceless (see Figure 10).
Figure 10: Compare the original image (top) with the final version (below).