Sharpless2-224 is an odd shaped and very faint supernova remnant in the constellation Auriga. It is estimated to be ca 14700 light years away with an age between 13000 and 24000 years.

The surface brightness is extremely low, making this one of the most difficult objects I have ever imaged. Even thou I have over 14 hours of Ha with a fast OTA (@f3.6) the amount of noise is still massive. I re-processed this image over and over again using different noise reduction strategies. The image above is the best result so far and is not too bad, but to me it still look "over processed" up close. Looking at other images out there I can find some comfort in not being alone with my struggles :-). I believe the image above competes well compared to what is on the net today and chose to publish it anyway, and besides I just can not take another round of processing. I may have another try when my noise reduction knowledge is up to the task.

Mouseover of the small thumbnails to the left will show the six different combinations of three narrowband images (Ha, OIII and SII).

Hubble Standard: R = SII, G = Ha, B = OIII. The top-left thumbnail is the familiar Hubble composition where the  Ha, OIII and SII images has been adjusted so that they all have similar histograms. In this way the OIII and SII contribution will show up. If this histogram equalization is not done, the image will be dominated by the Ha channel, i.e. it will be green.

Hubble colour shift: The top-middle thumbnail is based on the top-left, to which a colour shift has been applied Photoshop's Selective Color tool (see Bob Frankes tutorial). One has to be very careful when using the Selective Color tool in PS, since it is easy to shift to much resulting in sharp borders between colours.

Hubble modified: The top-right thumbnail is a modification to the Hubble composition (Orange = Ha (hue=35), Red = SII (hue=0), Cyan = OIII (hue=210). Instead of assigning Ha to green I assigned Ha to orange (Hue = 35 in PS notation), SII to red (as in Hubble) and OIII to a slightly more turquoise hue (Hue = 210 in PS notation). This palette has the advantage of separating the different narrowband images (as do the Hubble colour shifting composition), but has the advantage of not being dependent on careful usage of the Selective Color tool.

Artificial RGB: The bottom-left thumbnail is an attempt at mapping the narrowband images to colours (R = 80%Ha + 20%SII, G = OIII, B = 85%OIII + 15%Ha) so that the image would mimic the result of a broadband RGB image. It has been constructed by using the fact that H-beta emission (blue) accompanies the H-alpha line (red) with H-beta roughly 15% of the strength of the H-alpha. To the best of my knowledge this was first suggested by Richard Crisp. It also uses the fact that OIII is positioned in-between the blue and green part of the spectra. A litte bit of SII is added to the red channel.

ARGB redder: the bottom-middle thumbnail is based on the Artificial RGB image but the OIII contribution has been dialled down. Sometimes the ARGB blend above gives white areas where OIII is the strongest. In order to get a more "realistic" RGB representation the OIII contribution to both the blue and the green channel can be turned down to your subjective liking.

ARGB bluer: the bottom-right thumbnail is based on the Artificial RGB image but the OIII contribution to the blue channels has been dialled up. One of the ideas behind the Hubble palette representation is that it visualizes the different contributions Ha, OIII, SII. This can also be achieved in a RGB like representation by dialling up the OIII contribution to the blue channel. In this way the OIII areas will be bluish, the Ha areas will be reddish and the SII areas will be deep red.

In order to make most use of the weak OIII and SII data I have also used JP Metsavainio's Tone Mapping technique. A pdf that outlines tone mapping can be found here, and in order to bring out as much details as possible I have followed some of the tutorials on Ken Crawford's site.

Below you see images for each emission line respectively. The images has been stacked and stretched so that the histogram peak is equal for each of them. Note that the binning of the OIII and SII data was  3x3. This mean that each pixel receives 9 times as many photons compared to 1x1 binning. Thus a 4hr exposure would correspond to a 36hr exposure binned 1x1. Binning 3x3 of course looses detail, which in this image has been compensated by constructing a luminance using the H-alpha image with a touch of the OIII and SII data in it.

           

Ha                                                                    OIII                                                                  SII 

The following software has been used. MaximDL (image acquisition and guiding), CCDStack (calibration and RGB scaling, PixInsight (cropping, background correction, colour corrections) and Photoshop CS5 (all the rest, incl Noel Carbonis Astronomy Tools), and finally JP Metsavainio's Tone Mapping technique.

This image was processed in May 2013