new icn messageflickr-free-ic3d pan white
Orion and the Way | by -stille-
Back to photostream

Orion and the Way

Didn't quite do the processing right - still some strange background gradients I haven't quite managed to remove

 

27 lights, Canon 800D at ISO 800, Samyang 16mm at f2.8, 1 minute exposures, Omegon Lx2 tracking mount. 30 darks, 120 biases. Processed in PixInsight as below

 

*****Linear processing

*** Integration:

lightvortexastronomy tutorial (www.lightvortexastronomy.com/tutorial-pre-processing-cali...),

 

(15*(1-(FWHM-FWHMMin)/(FWHMMax-FWHMMin)) + 15*(1-(Eccentricity-EccentricityMin)/(EccentricityMax-EccentricityMin)) + 20*(SNRWeight-SNRWeightMin)/(SNRWeightMax-SNRWeightMin))+50 - use #775 as ref

 

*** Crop

 

*** Background extraction:

DBE tolerance 3, no points placed on the Milky Way

 

***Deconvolution

Created star mask for larger stars - large scale structure 2, small scale 1, noise threshold 0.1, scale 6,

Extracted luminance, STF autostretched, then histod shadows 0.2 midtones 0.28 highs 1 to get a range mask

Deconvolve with range mask on, 80 interations, custom PSF, dark 0.01 bright 0.004, local deringing with star mask, wavelet regularization

 

 

*** Color calibration

SNCR applied with range mask on (inverted) to protect nebulas

Background neutralization

Color calibration

 

*** Star reduction (for small and mid stars)

Small star mask - noise 0.15, scale 4, small scale 3 comp 1, smoothness 8, binarize, midtones = 0.02

Range mask from that, 0.05-1

Apply, erosion operator 4 iterations 0.15

 

*** Linear noise reduction

jonrista.com/the-astrophotographers-guide/pixinsights/eff...

 

*TGV - small noise

Created TGV masks - extracted luminosity, standard stretch (tgv_luma_mask), curved it with black point at ~0.2 and white at ~0.5, moved histogram point to middle (tgv_mask)

apply tgv mask inverted to the image, give luma mask as local support

TGV chroma str 7 edge protection 2E-4 smoothness 2 iterations 500

TGV luma str 5 edge protection 1E-5 smoothness 2 iterations 500

 

*MMT - larger noise and TGV artifacts

Created MMT mask - extract luminosity, standard stretch, move histogram point to 75%, apply low range -0.5. Apply inverted

MMT mask - 8 layers, threshold 10 10 7 5 5 2.5 2 2 on rgb

 

 

*****Nonlinear

 

***Initial stretch

*Autostretch, apply to hist

*Create full star mask, max(star_mask_large, star_mask_small)

* HDR transform, 8 layers, B3 spline, star mask applied inverted, preserve hue, lightness mask

 

***MLT stretch

 

**Initial

* created a new multiscale linear transform, kept 4 layers using linear interpolation

* diffed from original image to create a "blurred" version of original image

* extracted luminance from original, used as mask on blurred version

* used curves to create s shape in luminance and saturation, inflection 3/4 up

* pixelmath sum the 3, rescaled, back to original image

 

**Second

* new multiscale linear transform, keep 5 layers

* diff from original

* extract luminance from blurred image, to use as a mask

* masked blurred image with its own luminance, gave it s-shaped RGB curve, slight boost in luminosity, big boost in saturation

* pixelmath sum the 3, rescaled, back to original image

 

**Third

* new multiscale linear transform, keep 8 layers

* diff from original

* extract luminance from blurred image, to use as a mask, hist stretch it (multi_8_substracted_L)

* luminosity increase (1 curve), saturation (even more)

* pixelmath sum the 3, rescaled, back to original image

 

 

*** Darken

* DarkStructureEnhancer, 8 layers, 0.7, 3x3

* DarkStructureEnhancer, 8 layers, 0.7, 5x5

 

 

*** Color saturation

* bumped reds strongly, green-blues less strong

 

*** Sharpen

* Sharpen with multiscale linear transform, bias layers 2-6 (0.05, 0.05, 0.025, 0.012, 0.006)

 

*** Final crop and resize

* rotate 90* clockwise

* crop bottom (slightly weird corner)

 

 

* Rescale back to normal

 

 

**** Not used

 

**Create star and bright nebulas mask

* substract star_mask from luma to get a nebula mask

* exagerrate hugely with curve to get high contrast - RGB line going from 25% of horizontal to 50%

* apply said exagerration to stars too

* sum them up in pixelmath, save as star_nebula_mask

 

 

* new multiscale linear transform, keep 5 layers

* diff from original

* apply inverted stars_nebula_mask

* Local histogram equalization, kernel 200, contrast 1.5

* Local histogram equalization, kernel 400, contrast 1.5

* Saturate with curves (slight s-shape but mostly nuke

 

 

15,686 views
146 faves
9 comments
Uploaded on January 22, 2020