Preparing images for stock – getting rid of noise and banding
For the past few days, I’ve been doing the least interesting part of shooting for stock. I’ve been preparing images for the agencies that represent me.
Every agency has their own set of requirements and standards when it comes to identification, labeling and quality. Some are more strict than others.
When I started submitting to one of them, I got a wake up call.
What was easily acceptable at the others could not pass the muster at this one. At first, I thought they were being ridiculous and their concerns were unwarranted but I’ve come to realize that their high standards have made my images better.
Aside from the obvious sensor dust that needs to be retouched, the highlight and shadow details and the color corrections that always need to be made, there were other more problematic things that I had not noticed before.
When you post photos on the web at 72dpi and 640 pixels wide, flaws are easily hidden. Bad camera technique will never show. Agencies on the other hand look at all their stock images with the possibility that they might go billboard or poster size or in high quality magazine with excellent reproduction.
We’ve all seen the dramatic, powerful skies that are typical of HDR or graduated neutral density filters. They look impressive. Most of them would never make it through the front door of the larger agencies.
To get the dark, dark skies that we all like, usually means selecting the sky with the lasso tool and increasing density and contrast with curves or slapping on a graduated filter.
Both techniques can ruin a photo.
Two things can occur. Using curves to darken a sky often creates banding. Look carefully at which have a smooth transition from light to dark. Now apply curves to them. Usually banding occurs. Instead of the smooth stepless shift from light to dark, if you look closely, you see small steps or differences in density.
If you don’t have the shadow detail on the raw image when you darken it, your photo editing will create jumps in density from dark to darker. For me this is a no go. Banding means a rejected image and I have to rework it and resubmit.
The ever popular graduated ND filter creates another problem. Noise.
Because light is prevented from reaching the sensor where the filter has blocked it, the resulting area of landscape becomes underexposed. Under exposing means no detail and it also means NOISE. Too much noise means a rejected photo.
All is not lost. There are ways around both problems. To eliminate the banding you need extra exposure information in your sky detail. The same goes for noise.
This is why I think HDR is far superior to using filters or curves to darken skies. By over and under exposing the sky in my HDR shots, I eliminate banding. I also decrease the noise in the shadow areas although a little noise will sometimes remain.
There are various filters and plugins for Photoshop to reduce noise. The most popular is probably Noise Ninja. The pro bundle which works as a plugin in Photoshop sells for $70 US. It’s popular because it does a great job. It’s easy to use and offers far more control than is available in Photoshop.
As good as Noise Ninja is, my favorite is the Boundary Noise Reduction filter by Colormancer. It sells online for $65 US.
Before I go any further, let me explain that I have no ties with either company.
I bought and paid for the plugins with my own money and get nothing in return from them. They aren’t even aware that I am writing about them. What I do think is dishonest and misleading is what goes on at other websites. It’s a “you pat my back and I’ll pat yours” situation. Amateur and semi-pro photographers will try and sell you about how good the product is when in fact, it really isn’t. You figure out which blog I am talking about. There’s no shortage of people looking to get recognition and free publicity on the net.
Back to the banding and noise. Both programs are excellent for noise reduction. In my experience, Boundary Noise Reduction is slightly better but could just as easily live with Noise Ninja.
Here’s a handy little tip for reducing noise in your skies. That is the place where it’s usually most noticeable. Flat, even areas usually show noise readily as opposed to photographs with lots of detail. I usually don’t need the noise reduction in other areas of the image.
I copy the complete image to a new layer in Photoshop. I then apply the filter and watch the preview to see the effect of the filter. Too much noise reduction and the image becomes blurred. Too little and the noise is still there. It’s kind of a Goldilocks thing. Not too much, not too little.
When I get the effect of the noise reduction I want, I apply it to the top layer.
With the top layer I now add an image mask, then with a large, softly feathered black brush, I paint over the area that I don’t want the filter applied to. If you use a black brush on the image mask, you are removing the top image where you paint and the bottom image shows through. By varying the density or opacity of the brush you control how much or how little image you reveal underneath. If you go too far, change the brush to white and you can undo what the black brush has done.
When I get the effect I want, I flatten the layers and the result is a smooth, noiseless sky with everything else remaining crisp and sharp with no noise reduction introduced elsewhere.
Image masks are a great way to control your photo in ways that a filter never can. Once you get the hang of it, it will become second nature.
If you look carefully at your photos at 1:1 magnification you’re bound to see the flaws that I talk about. I think that the majority of dramatic photos that we see on the net would reveal banding excess noise at larger sizes. That’s not a bad thing if you only ever intend to display at 640 pixels wide but he banding and noise will show up at prints 11×14 and larger.
Refining your technique while shooting and processing will make for better prints and better raw images to work with. That makes processing them so much easier and more time can be spent creating outdoors instead of being tied to your computer and fixing preventable flaws.