by Roger N. Clark
Astrophotography image processing example is shown with globular star cluster M22 in a region of strong interstellar dust. In a region filled with nebulosity, stretching while maintaining correct color balance poses challenges. The rnc-color-stretch algorithm can produce a near-final result.
The Night Photography Series:
The globular star cluster M22 lies near the center of the Milky Way galaxy in an area filled with orange interstellar dust (Figure 1). Images of this region are a tough test for image processing to maintain a natural color balance. We commonly see globular star clusters with white and blue stars. But globular star clusters contain cooler stars that mostly appear yellow to orange to red with only a few slightly bluish-white stars. This can be verified by doing an internet search. For example, Color-Magnitude Diagram for the Globular Cluster M22 show B-V magnitudes in the 0.5 to 1.5 range with a few stars near 0 and up to about 2.5. Examining Table 1 from Color of Stars in this series shows B-V ~0 is blue-white, 0.5 is yellowish-white, 1.0 is yellow-orange, 1.5 is orange-red, and 2.5 would be redder.
This article discusses strategies to stretch the image data to bring up faint stars and dust, but keeping a consistent color balance throughout the intensity range. This is quite difficult. In earlier articles, the blackpoint was easily found from very dark areas in the image. But in the M22 region, there are no really dark areas, so there is no black point. Automated tools as well as stretching by hand to make dark things black result in incorrect color balance with scene intensity. This is just the opposite of removing light pollution, which is commonly orange. Suppressing light pollution in astrophotos is commonly done incorrectly with color balance, as discussed earlier in this series, resulting in bluing of scene intensity as intensity drops. A similar effect happens when subtracting too much signal in regions filled with orange interstellar dust--it produces a bluing with decreasing scene intensity.
M22 was imaged with a Canon 7D Mark II 20-megapixel digital camera and 300 mm f/2.8 L IS II lens plus a 1.4x teleconverter giving 420 mm at f/4 and ISO 1600. A total of 9 minutes of exposure was made (nine 1-minute exposures). The raw data were converted in photoshop with daylight white balance and stacked according to methods discussed earlier in this series.
The result of applying rnc-color-stretch on the stacked image, with default sky determination, which assumes a dark part of the image is a blackpoint, is shown in Figure 2. The result is a bluing of faint stars and dust with decreasing scene intensity and is not natural.
Another strategy is to use one iteration and a higher power value to rnc-color-stretch. Figure 3 shows an example, again with default blackpoint. The result still shows bluing with decreasing scene intensity.
As we saw in Figure 1, there is no actual blackpoint in the scene, so we must set the level and color of the darkest point in the image. Figure 4 sets the sky zero levels at red = 8000, green = 6000, and blue = 5200 on a 16-bit 0 to 65535 scale. This sky level point improved the result, but it is not quite enough. A sky level of RGB = 12000, 8000, 5000 shows a better result (Figure 5). The result in Figure 5 could be considered final.
While the result in Figure 5 could be final, I wanted to emphasize the stars, not the strong interstellar dust. So I brought the image into photoshop and used the curves tool to decrease the dust intensity, to produce my final image in Figure 6. Cropping a little shows the image in Figure 7. Here one can see the different colors in the stars, indicating their differences in temperature.
The rnc-color-stretch software is free and open source. You can download it from
8a) Software for nightscape and astrophotographers in this series.
The rnc-color-stretch algorithm can produce a final or close to final stretched astrophoto in a variety of environments. Optimal results depend on input parameters. The region imaged should be understood in case there is no blackpoint in the scene. If there is no blackpoint, the user must specify the level and color of the darkest parts of the image. Determine if the sky level and color is reasonable by examining faint stars and checking if the color trend is consistent. For highest accuracy, check the B-V color index on individual stars (e.g. using Stellarium), and use the index as an indicator of star color (See Table 1 from Color of Stars in this series. If faint stars are too blue, try increasing red in the -rgbskyzero value (and usually increase green a smaller amount).
References and Further Reading
Clarkvision.com Astrophoto Gallery.
Clarkvision.com Nightscapes Gallery.
The Night Photography Series:
First Published November 28, 2016
Last updated November 28, 2016