Perigee Full Moon, August 10, 2014

Contains: Earth's Moon

Technical card

Resolution: 3200x2250

Date: Aug. 10, 2014

Time: 22:00

Frames: 197

Locations: Backyard, Hampton, VA, Hampton, VA, United States

Description

I took advantage of the full moon’s proximity earlier this month and captured this color image using Ha and OIII narrowband filters, inspired by Mike Oates’ excellent tutorial on the PixInsight forum (reference below). The total number of exposures is 197, broken down as follows:

76 Ha (46 x 1/200 s, 30 x 1/250 s @ ISO400)
121 OIII (61 x 1/100 s @ ISO400, 60 x 1/100 s @ ISO800)

Working with DSLR images was tricky since I had the Bayer matrix to deal with, but after lots of experimentation I got most of the images to line up nicely. The red Ha channel was hardest to deal with because it was not exposed very well and the noise fooled the StarAlignment process quite often until I increased the noise detection scales. After integration, the Ha image did not have the same SNR compared to the OIII image which had the benefit of a blue channel plus two green channels to average together. However, after applying some deconvolution to the combined Ha + OIII image, things really tightened up.

The rest of post-processing closely followed Mike's tutorial, with some parameter tweaks along the way. One snafu arose towards the end where I noticed the rim of the moon was bright blue, suggesting that differential pixel rejection between the Ha and OIII integrations had left some blue pixels without much red to compliment them. Since they were saturated blue pixels, I made a mask to select them by extracting the HSI saturation and the blue channels, and multiplied the two together. Interchannel curves provided a convenient way to drop the saturation to a more natural hue and remove the blue rim.

Thank you, Mike for sharing your tutorial!

-------------------------------------------------------------------------------------------
Processing Workflow (PixInsight)
-------------------------------------------------------------------------------------------

Pre-processing
1. Calibrate light frames using bias, flats (only dark flats used).
2. Extract red channel from Ha (BatchChannelExtraction).
3. Extract two green and blue pixels (SplitCFA).
4. Align OIII green and blue images with a green OIII image reference (StarAlignment, distortion correction on -> residual 0.2, iterations 20, star detection -> detection scales 8, peak response 0.95, maximum distortion 0.1, star matching -> descriptors per star 80).
5. Align Ha red images with same green OIII image reference (StarAlignment, distortion correction on -> residual 0.05, iterations100, star detection -> detection scales 8,noise scales 1, peak response 0.95, maximum distortion 0.1, star matching -> descriptors per star 80).
6. Manually inspect alignment and reject as necessary (Blink).
7. Integrate OIII green and blue images, weighting by average signal strength and using Windsorized sigma clipping method (ImageIntegration).
8. Integrate Ha red images, weighting by average signal strength and using Windsorized sigma clipping method (ImageIntegration).
9. Drizzle integration of OIII.
10. Drizzle integration of Ha.

Post-processing
1. Match average brightness between Ha and OIII (LinearFit; OIII as reference)
2. Combine Ha and OIII images (PixelMath: R = Ha, G = 0.5*Ha + 0.5*OIII, B = OIII).
3. Set luminance coefficients to 1 for RGB channels and set gamma = 1 (RGBWorkingSpace).
4. BackgroundNeutralization.
5. Set white balance (ColorCalibration, moon as white reference).
6. Deconvolution (StdDev = 2.4, iterations 100, deringing; mask selecting moon).
7. Deconvolution (StdDev = 1.4, iterations 100, deringing; mask selecting moon).
8. Non-linear stretches (HistogramTransformation).
9. Contrast curve (CurvesTransformation).
10. Increase color saturation (CurvesTransformation, saturation channel).
11. Sharpen luminance and increase contrast
11a. Set gamma = 1 (RGBWorkingSpace).
11b. Extract luminance (ChannelExtraction).
11c. UnsharpMask (mask selecting moon).
11d. Compress dynamic range (HDRMultiscaleTransform, number of layers 6).
11e. Contrast curve (CurvesTransformation).
11f. Combine L back to RGB image (LRGBCombination, L channel weight 0.7)
12. Increase local contrast (LocalHistogramEqualization, kernal radius 128, contrast limit 1.2, amount 0.2; mask selecting moon).
13. Reduce bright blue along moon rim where red Ha data did not match well with blue OIII data.
13a. Extract HSI saturation and blue channels (ChannelExtraction)
13b. Create mask of saturated blue values (PixelMath, HSIsaturation*Blue).
13c. Clone out the few areas unrelated to the blue rim (CloneStame).
13d. Reduce saturation of blue values along rim (InterchannelCurves, target channel CIE c, reference channel L; mask selecting saturated blue values on rim) .
14. Final curve to brighten overall, but especially white areas (CurvesTransformation).
15. Crop and rotate so north is up (DynamicCrop).
16. ICCProfileTransformation to sRGB

References:
Mike Oates: Processing Example: The Moon in NarrowBand using StarAlign & Drizzle
http://pixinsight.com/forum/index.php?topic=7214.0

Comments

Author

Jason Tackett
Like

Histogram

Perigee Full Moon, August 10, 2014, Jason Tackett

Made possible by

O'Telescope BackyardEOS