A complex interplay of bright and dark nebulae with several open clusters. Modified Canon 350D, 30x4' exposures, 135 mm f2.8. See here for an old 40 min exposure on the never forgotten Kodak 2415 (300 mm f4 plus deep red filter). The processing I describe here is aimed at optimizing the range of tonal values occupied by the nebula while keeping the star images unclipped. This is accomplished by blending the image (left) with a low pass mask that forces the low frequency structures in the image to climb the histogram toward the bright end (right). The processing enhance the nebula without increasing brightness of stars. Furthermore, star colours are better preserved. This processing is especially useful for wide field images in which nebular structures tend to remain confused in very cluttered star fields.
1) Preprocessing, levels and curves: starting point
Pre-processing is performed as usual (under Iris in my case) with offset, dark, flat and hot pixel treatment. The workflow proceeded with hyperbolic asinh stretching. The image was exported to a 16 bit Photoshop document. Further treatment included level regulation and curve treatment to further reduce the difference between the luminance of the nebulae and the brightest stars.
2) Preparation of the low-pass mask (1)
The RGB image is filtered through a gaussian filter. In this case the width is 20 pixel. This works well with wide field images where mid brightness stars are only a few pixel across. A wider filter can be used if stars where larger.
3) Preparation of the low-pass mask (2)
A wider filter can be used if star images are larger than the above example. In this case-a close up of the reflection nebula around g Cassiopeia-a 50 pixel wide filter was used. Furthermore the luminosity of the patch corresponding to the bright star was reduced by using the burn tool in Photoshop.
4) Regulation of the mask brightness and gamma
Filtering causes a drastic change in the histogram of the image. The mask has to be lifted to bring its brightest pixel near the white point. This is done by adding the same offset to all channels with the brightness control. A certain degree of experimentation is required. As a general criteria keep in mind that the blending will darken the background and increase the brightness of the low frequency parts of the image. If the mask is too dark you might get too much compression at the bottom of the histogram. At this stage the gamma of the mask can also be modified: a good way of proceeding is combining this step with the following.
5) Blending of the mask
The mask is copied and pasted over the original image with the overlay fusion option. This is the time to adjust the brightness and gamma of the mask. Click over the mask layer and go to the adjust level or adjust curve tools. If both layers are visualised you'll be able to perform fine tuning of the effect while looking at the end result.
I have measured RGB values in 5 different locations on the background and on the nebula and the luminance of 5 different stars on the original images and after the masking. Results are shown as a signal ratio: values larger than 1 mean that filtration increased the intensity of the specific signal. The graph demonstrates that the contrast of the red component of the nebula increased of almost a factor 1.6. Furthermore the signal from non saturating stars decreased of almost 20%.
Roll over with the mouse to see a blinking comparison of the images before and after filtration.
Send anything that comes to mind to Gimmi Ratto firstname.lastname@example.org
Copyright © 2005 by Gimmi Ratto. (October 12, 2005)