M 42 in HOS using HDR

M42 is a object that cannot be adequately imaged in a single exposure.  Hence the use of High Dynamic Range or HDR.  With this process images of several exposures ( each 16 bit integer) are combined to form a single 64 bit floating point image.  This image is able to encode both the very bright and very dim.  Originally I was going to use just 60 second exposures for the Triangulum and 300 second exposures for the nebula.  The latter did well on M42, but did not capture the surrounding H alpha clouds.  Thus I had to add another round of 600 second exposures.

I really love ACP Expert.  Unfortunately ACP Expert adds a lot of overhead.  On my normal 900 sec images this extra time can be safely ignored.  If you are trying to shoot 60 60 sec images within 4 hours then it cannot.  Even with the 300 and 600 sec images I needed to eliminate any useful, but unnecessary operations if I was going to complete the project before the clouds returned.  Thus I shot the entire project directly with scripts in ACP.  The scripts eliminated plate solving and for the 60 second images eliminating RBI Flood (which is itself 2 minutes).  Doing these allowed me to capture > 15 images at 60 seconds and 300 seconds in H, O, and S.  I was only able to capture Hydrogen at 600 seconds, but that was enough to add depth to the gas clouds surrounding M42.

Annotated Image

The image is presented north up. 

Close up of M42

This is a half size zoom into the heart of M42/M43.  Clicking on this image will bring up a higher resolution version.

Zoomable Image

Processing Details

	Red = Hydrogen
	Green = Oxygen * 1.25
	Blue = Sulfur * 1.25

The resulting image was VERY noisy outside of the bright inner nebula.  The final result is a tribute to TGVDenoise.

Here are some other processing items of interest

• Getting the proper stretch required some tuning.  I used the manual settings of ScreenTransferFunction to find a proper balance of background and stretch.  I then transferred these setting into HT to make the image non linear.
• I used a Median MMT
• The initial image was very noisy. It required multiple passes of TGVDenoise with a variety of different masks to smooth the various sections of the image.
• My first attempt at mid range and background masks each resulted in very noisy masks.  Since the goal was to use these in TGVDeNoise that was not acceptable.  To fix this I ended up using TGVDeNoise with very strong settings to smooth the masks.  I found this worked much better than Convolution.  Naturally after I smoothed the masks I had to remove the stars by placing a starmask over the range mask and then setting all of the locations with stars to 0 (fully masked).
• LocalHistogramEqualization brought out more detail in M42.