For many years now I have been a visual artist working for the European Southern Observatory (ESO), where one of my missions, among others, is processing images taken by ESO and the NASA/ESA Hubble Space Telescope. This image has been literally one of my ‘biggest’ challenges yet.
This spectacular image from ESO’s mighty VLT Survey Telescope (VST) shows the Cat’s Paw Nebula (NGC 6334, upper right) and the Lobster Nebula (NGC 6357, lower left). These dramatic objects are regions of active star formation where the hot young stars are causing the surrounding hydrogen gas to glow red. A zoomable version of this giant image is available here.
The instrument that captured this picture creates a large mosaic, in simple words it was put together by stitching a lot of tiles. Sometimes the edges don’t blend perfectly, and Dr. Pixel has to come to the rescue.
What is it about?
Here you can slide between the original VST image before and after it was cleaned. The mosaics with their different intensity where a huge problem. To clean this with traditional methods (clone, dodge & burn…) in a reasonable amount of time appeared to me like a hopeless case at first glance. I had to find a better way. The solution? I will now try to explain.
Before we start I would like to talk about a question that hear very often. How real are the colors in astro images?
Telescopes observe at different wavelength which can be translated into a certain color. As you can see below, a 600 nanometer filter should be colorized orange. But there is also subjective perception.
A correctly colored image can look unpleasant to the eye. So we have a little freedom in choosing a certain color range within the orange in order to balance the result (a bit towards red or yellow). If we want to produce an RGB color image, all greyscale layers must be colorized in red, green and blue according to the electromagnetic spectrum in which they where observed. Screen mode combines them into a color image which can be further adjusted.
If we use additional filters which are located outside the visible spectrum, like near infrared (NI) or ultraviolet (UV), things become more tricky. We can not colorize red ‘redder’. In this case we have to keep the chromatic order by colorizing NI red and UV blue. The result will be indeed a bit more ‘artisic’ and requires some experience.
Enough said, I promised to keep it simple. Lets move on to the cleaning part.
Isolating the gaps
To isolate all the bad parts such as gaps, lines and areas with little information, I was checking if there is another picture of this region available.
I found an image from The Digital Sky Survey (DSS). It had a much lower resolution which is a problem, because after scaling and aligning it with the VST image, the stars where much bigger and of course blurry. I had to shrink the stars by using Surface blur and some other methods in Photoshop first.
But after this was done, I was able to generate a subtraction mask that isolated just the useful parts of the DSS which I then applied to the VST.
Generating a Mask
After subtracting the mask, the defects on the VST where reduced significantly. I used this method on all filters, and with a lot more of traditional cleaning, I was finally able to create a fairly clean image. Overall it took almost ten days to finish it.
Of course I’m just scratching the surface here. Many more steps where necessary to complete this task. But in order to keep it short and as little confusing as possible, I think it is better to segment the different aspects into several tutorials, which will be available in the near future.
It was quite a struggle to free this picture from all the artifacts and to reveal it’s entire beauty. But it was worth the effort. It became Astronomical Picture of the Day (APOD) and attracted wide interest in public.
Who said hopeless case?
A little sneak preview about the upcoming topics:
- Best JPEG compression
- Nikon vs Canon
- Lightroom & Photoshop
- The Power of Luminance Masks
– Luminance sharpening
– Luminance highlights
– Luminance colors
- Lets talk about taste
- HDR Obsolete – but if, do it right
- Astro Stacking
- Vixen Polarie
- Common Mistakes