Choices, choices, choices (in imaging processing)

One of the wonderful problems that has become more apparent with digital imaging with modern optics is high quality images with lots of stars. Back the dark(room) ages with film, it seemed that most images didn’t really pick up that many of the fainter stars. I can make up some really smart sounding reason for why that was (maybe not enough light to activate the film since non-linear film has an activation threshold), but the truth is, I have no idea nor have ever done any real analysis or comparison. But I certainly do capture a lot of stars with my digital cameras. As in, OMG I had no idea that blank area of the sky had so many stars.

This is a problem. Because quite often the stars actually distract from the overall image. And that’s where the choices come in.

First of all, there are always choices in the details. How much to increase saturation, how much to tweak the curves adjustment to enhance the contrast, how to smooth/filter the noise, etc. Which begs the question everyone always asks: is that really what it looks like?

The short answer is “no” but not for the reasons you might think. First, all those cool sci-fi pictures of galaxies/nebulae in the window of a spaceship are seriously not real. That’s because the surface brightness of all those things is too low and our eyes can’t pick up the colors or even much of the light. Sorry, they’re cool effects, but they’re not real either.

The longer answer is “yes, but….” The image is not fake, but it has been enhanced to bring out details we can’t pick out given the eyes nature has given us. Sometimes, what we’re doing is trying to capture the impression we get when we look through the eyepiece of a telescope. The eye and brain are very sensitive and can pick out detail that is quite subtle. The camera doesn’t come with a brain and it just shows you a photon intensity map. If you ever pulled out your camera and took a picture of a gorgeous moon then looked at the picture and thought “WTH?” then you should realize that.

When processing an image, most of the steps involve mathematical transformations/operations that are not at all arbitrary, though there is room for creativity in the sense that I can tweak the parameters used. When looking at the digital noise in an image, how much do I want to average over adjacent pixels to remove/smooth over the random variation of the sensor from pixel-to-pixel? Can I sharpen the image to bring out the detail? Can I adjust the contrast to things that are really there, just not quite visible the human eye? The answer to all of those questions is, of course, yes you can. The artistry is knowing when to stop.

And back to those stars…. What the camera records is all those stars. And sometimes they steal the show when the real, ahem, star is actually not a star. In the recent picture of NGC 2238 (the Rosette Nebula), the nebula is in a rich star field. Enough so that it is distracting and takes away from the gaseous nebula itself. If only you could somehow “dial down” the stars to focus on the nebula itself. Turns out you can.

All of these are the same image. Just processed differently. Here’s the starting point (well, not really, this is after a lot of processing).

So… there’s nothing truly fake about any of those images. But it’s also true that what you see in the image is not what you would see in the night sky. And that doesn’t even get started on wavelength sensitivity. Cameras and eye’s don’t really convert colors the same and then your computer monitor doesn’t even emit the right wavelength of red to match what he cameras originally detected.

The bottom line is that astronomy pictures are a bit of science and a bit of art, crafted to satisfy the imager (me). Well, that’s not quite true. I’m never quite satisfied with the outcome. But now that I know what I could get out of this field, I’m ready to go back and try again.