Hemisphere:  Northern  ·  Constellation: Cassiopeia (Cas)  ·  Contains:  12 Cas  ·  LBN 592  ·  LBN 593  ·  LDN 1282  ·  LDN 1283  ·  Sh2-173  ·  The star 12Cas
Getting plate-solving status, please wait...
Phantom of the Opera (HSO), 


Powered byPixInsight

Phantom of the Opera (HSO)

Getting plate-solving status, please wait...
Phantom of the Opera (HSO), 


Powered byPixInsight

Phantom of the Opera (HSO)

Acquisition details

Nov. 28, 2021 ·  Nov. 29, 2021 ·  Nov. 30, 2021 ·  Dec. 1, 2021 ·  Dec. 2, 2021 ·  Dec. 3, 2021 ·  Dec. 4, 2021 ·  Dec. 5, 2021 ·  Dec. 6, 2021 ·  Dec. 7, 2021 ·  Dec. 8, 2021 ·  Dec. 10, 2021 ·  Dec. 11, 2021 ·  Dec. 12, 2021 ·  Dec. 13, 2021
Astrodon H-alpha 5nm: 101×600(16h 50′) -25°C bin 1×1
Astrodon OIII 5nm: 124×600(20h 40′) -25°C bin 1×1
Astrodon SII 5nm: 120×600(20h) -25°C bin 1×1
57h 30′
Avg. Moon age:
13.35 days
Avg. Moon phase:

RA center: 00h22m31s.779

DEC center: +61°4806.94

Pixel scale: 0.834 arcsec/pixel

Orientation: -90.286 degrees

Field radius: 0.558 degrees

WCS transformation: thin plate spline

More info:Open 

Resolution: 3492x3316

File size: 592.6 KB

Locations: Dark Sky Observatory, Fort Davis, TX, United States

Data source: Own remote observatory

Remote source: Non-commercial independent facility


There was a face on Mars but now we have a Face in Space! This is Sh2-173, also known as the Phantom of the Opera because of its somewhat resembles the mask used in the play poster.

This is probably the faintest target I've imaged and I wanted to devote a lot of time to it to see how much it would help. This make it the largest single pane integration I've done. The H-alpha was faint. The SII was fainter. The OIII was exceedingly faint and essentially nonexistent. There was a trace amount of "background haze" in OIII but it was so faint that any stretch that Brough it out of the background was too noisy to use.  It was so faint that I had to zoom into the histogram by a factor of 100 adjusting the midpoint until the nebula even began to show up. Even with a lot of data that limited how much I could push the data without it starting to degrade the appearance. My first attempt was a more aggressive stretch that really pushed the OIII which showed up as this "haze" on the right hand side but it didn't look good (there just wasn't enough signal and so the noise floor was amplified too much). I went back and stretched less aggressively and while it limited how much contrast I could achieve I liked the result better.

I chose HSO as the presentation as I liked the color palette the best. Perhaps I let the nicknames for these influence me too much but this palette worked the best with the "phantom" idea in my brain. Though the greenish SHO also worked, I preferred the muted orange that HSO produced. I even considered an HSS combination but though it was very similar to HSO I preferred the HSO slightly better.

Despite all that, the processing was relatively straightforward:

each, H, S and O:
dynamic crop
deconvolution. (made copies of each image here for use with stars later)
histogram transformation
tgv denoise

channel combination
histogram transformation (to align peaks)
channel combination in H as lum
MLT (very mild sharpening)
pixel math in stars (addition)
histogram transformation (raise black point)
dark structure enhance script
masked off the bottom right star and color matched the halo to the background

channel combination the copies of H, S and O from earlier
histogram transformation
SCNR to remove green

Normally I use the max() function in pixel math to combine stars but the stretch here for the nebula was so large that starXterminator artifacts were seen on the brightest few stars. Adding the stars in the "+" instead of max made the stars bigger but also blended them into the starXterminator blobs better making them less objectionable.

I ended up avoiding LHE which is a favorite of mine for contrast adjustment because it was just too hard on the data even in its mildest forms. Overall this was an interesting image to work on. It was instructive on how to deal with faint data even when I had as much per filter as I have had for entire images in some earlier efforts.