Not mega-happy about this one - it's basically a tryout for when I'll get to add more exposure time to this object - but the basic gist of it, assuming I'm going to keep playing along with vintage glass for it - is aim it at heart & soul rather than heart & double (a more boring framing but one that doesn't end with double nuking my exposures) and find one of those nice big-telescope mosaics in wideband to use as a reference to figure out what's going on with the background. Until now, using star-poor areas as samples for background extraction and neutralization has worked well, since these are usually black dust clouds, but here we're in the middle of the Perseus O-B association and it's looking like very dimly illuminated gas everywhere

1 stack of 66 images, Canon 800D at ISO 800, Pentax 135mm f2.5 lens at f4, 30s exposures, tracked with Omegon minitrack lx2. a couple dozen darks, 120 biases. Processed in PixInsight as below

***** Integration

lightvortexastronomy tutorial (https://www.lightvortexastronomy.com/tutorial-pre-processing-calibrating-and-stacking-images-in-pixinsight.html)

* CC defect list + master dark (sigma = 10)

* weighing:

15*(1-(FWHM-FWHMMin)/(FWHMMax-FWHMMin))

+ 25*(1-(Eccentricity-EccentricityMin)/(EccentricityMax-EccentricityMin))

+ 10*(SNRWeight-SNRWeightMin)/(SNRWeightMax-SNRWeightMin)

+ 30*(1-(Median-MedianMin)/(MedianMax-MedianMin))

+ 20

* img 6951 ref

* generalized student ESD rejection, remove range_low 0.2

* drizzle integration

***** Linear processing

* linearfit R and B channels to G, channel combine

* crop

* DBE tolerance 3

* PCC on Heart Nebula, FL 135 pixel 2

* EZDenoise with TGV -1 exponent from calculated value

* SNCR green

***** Nonlinear processing

*** Initial stretch and halo fixes:

* EZSoftstretch 0.25 0.10

** Halo fixes

* blue_mask = B - (G + R)/2

* stretch blue mask shadows 0.04 mids 0.08

* with masks on, SNCR blue, SNCR green

*** Mask creation

** Star mask

* binarize star mask created by starnet

* create range mask 0.8 - 1, 5 iterations of dilation

* star_mask = maximum of above, blurred with 4 layers of a trous wavelet transform

** Better blue mask:

* blue mask = B - (G + R) / 2

* histo stretch shadows 0.03 mids 0.01 - save as blue_mask_2

*** Multilayer processing

http://www.stelleelettroniche.it/en/2014/09/astrophoto/m42-ngc1975-multilayer-processing-first-part/

** First iteration (detail)

* duplicate, starnet into starless and stars

* extract 7 MMT layers from starless, diff

* extract red channel from starless, stretch shadows 0.1 mids 0.25, then pixelmath 1 - abs($T - 0.5) -0.2 rescaled, apply as mask on left, s-shaped RGB curve, pump up saturation, then remove mask and SNCR

* starless = starless + mlt + left, normalized

* heart = (starless + stars) * (1 - star_mask) + star_mask * heart

Beginning to suspect mis-calibrated background, so doing another background neutralization with RoI left 3200 top 2800 width 800 height 400 and 0.3 upper limit

** Second iteration (heart nebula)

* duplicate, starnet into starless and stars

* extract 7 MMT layers from starless, diff

* mask creation

* heart nebula: extract channels from left, left_R * left_R - (left_B * left_B + left_G * left_G) / 4 = heart_mask, clonestamp out leftover double cluster, stretch shadows 0.02 mids 0.24. Clean up everything but heart and soul, apply stretch twice more

* with heart mask on, pump up lightness slightly and saturation a lot

* starless = starless + mlt + left, normalized

* heart = (starless + stars) * (1 - star_mask) + star_mask * heart

*** LHE

* LHE 512 1.5 on clone

* mask creation: take heart, blur 6 a trous layers, stretch shadows 0.02 mids 0.24, blur 8 layers more, max(heart_mask, star_mask) = lhe_mask

*** Finishers

* MMT sharpen - 6 layers, biases 0.05 0.05 0.025 0.025 0.012 0.005, with star_mask on

* Resample 4