Notes on Revision B
------------------------------------------------------------
Well it turns out that the monochromatic nature of the original processing for this image was due to the strong red color bias in my flat frames which saturated the blues and suppressed other colors. I discovered this error in my calibration just recently and found a way to balance the RGB channels of my flat frames using pixel math. This procedure is documented in my M92 image description. Using these newly calibrated light frames, I was able to get much more sensible colors out of this galaxy. I definitely want to revisit this one from a darker location since it has such low contrast which is exacerbated by the red zone I imaged it from.
Notes from the original version (all hypotheses here have been discounted with Revision B)
-------------------------------------------------------------
Oh light pollution how you’ve vexed me again! Here is an image of our third-closest galactic neighbor, the Triangulum Galaxy, captured at the Grundland Astronomy Park in Hampton, VA a few weeks ago. For most of the time I was acquiring data, M33 occupied the darkest part of the sky which faces out over the Chesapeake Bay. However residual light pollution from Hampton Roads conspired with the low surface brightness of this object to make it a bear to pull the galaxy out of the noise once I subtracted the light-polluted background. It is clear that I would need double or triple the integration time to really do this object justice.
Like so many other M33 images, mine is strikingly monochromatic which makes me wonder where all the yellows are that exist in most of the better images of this galaxy. In the last few steps I manipulated the Lab *b channel to deepen the blue and brighten the yellow. I also boosted red in the galaxy slightly since red and green mix to make yellow in the RGB color model. These steps combined with selective color saturation of the HII regions (red) added the tiniest bit of diversity to the color in the image. In reality, I need more integration time and perhaps some H-alpha data to really do a nice job. Looks like I will be revisiting this one again. Still, not too bad for an evenings’ work.
-------------------------------------------------------------------------------------------
Processing Workflow for Revision B (PixInsight)
-------------------------------------------------------------------------------------------
Pre-processing
1. For master flat frame, scale CFA channels 1-3 to match median of CFA0 channel (SplitCFA,PixelMath expression $T*(median(CFA0)/median($T)), MergeCFA)
2. Calibrate light frames using darks, bias, new flats
3. Integrate using sigma clipping rejection.
Post-processing
1. Initial crop (Dynamic crop) .
2. Reduce background gradient and neutralize background (DynamicBackgroundExtraction, subtraction).
3. Neutralize background (BackgroundNeutralization).
4. Set white balance (ColorCalibration; use entire image including galaxy).
5. Non-linear stretch (HistogramTransformation, lower midtones slider aggressively).
6. Non-linear stretch (HistogramTransformation, raise blackpoint slider).
7. Increase contrast (CurvesTransformation “s” curve).
8. Set luminance coefficients to 0.333333 for RGB channels (RGBWorkingSpace).
9. Restore colors and smooth star profiles to the heavily stretched stars using the method described by Vincent Peris (see Peris reference below).
10. Compress dynamic range. (HDRMultiscaleTranform w/ luminance mask).
11. Increase local contrast slightly (LocalHistogramEqualization; kernal radius: 128, contrast limit:1.2, amount: 0.58).
12. Reduce luminance noise (ACDNR to luminance with luminance mask).
13. Reduce chrominance noise (ACDNR to chrominance with strong luminance mask).
14. Small contrast curve (CurvesTransformation).
15. Increase galaxy color saturation (CurvesTransformation w/mask selecting galaxy only).
16. Increase star color saturation (CurvesTransformation w/mask selecting stars).
17. Reduce green. (SCNR to green).
18. Increase galaxy brightness relative to background (ExponentialTransformation w/lightness mask & luminance mask).
19. Lower background brightness slightly (CurvesTransformation to RGB/K)
20.Reduce star sizes (MorphologicalTransformation; morphological selection: iterations, amount, selection = 2, 0.70, 0.32 w/star mask).
21. Remove dark pixels (PixelMath [Schwarz script]; w/inverted luminance mask)
22. Reduce background noise (MultiscaleLinearTransform; k-sigma noise thresholding, amount 0.6 w/ luminance mask).
23. Increase local contrast slightly (LocalHistogramEqualization; kernal radius: 128, contrast limit:1.2, amount: 0.58 w/mask selecting galaxy).
24. Reduce red histogram slightly to balance median RGB values of background (HistogramTransformation to red).
25. Reduce background to a median background value of about 0.1 HistogramTransformation).
26 Final Crop (DynamicCrop).
27. Set ICC profile to sRGB for web publishing (ICCProfileTransformation).
-------------------------------------------------------------------------------------------
Processing Workflow for Original image (PixInsight)
-------------------------------------------------------------------------------------------
Pre-processing
1. Calibrate light frames using darks, bias, flats
2. Integrate using Windsorized sigma clipping rejection.
Post-processing
1. Initial crop (Dynamic crop) .
2. Reduce background gradient and neutralize background (DynamicBackgroundExtraction, subtraction).
3. Reduce background gradient and neutralize background (DynamicBackgroundExtraction, subtraction).
4. Neutralize background (BackgroundNeutralization).
5. Set white balance (ColorCalibration; use entire image including galaxy).
6. Set luminance coefficients to 0.333333 for RGB channels (RGBWorkingSpace).
7. Reduce background noise (MultiscaleLinearTransform w/ luminance mask).
8. Non-linear stretch (HistogramTransformation, lower midtones slider aggressively).
9. Non-linear stretch (HistogramTransformation, raise blackpoint slider).
10. Increase contrast (CurvesTransformation “s” curve). In previous images I would use a star mask to prevent the stars from bloating further, but I found that dark rings formed around the stars in later steps due to inadequate protection around the stars. For this image, I didn’t bother with the star mask and just replaced the stars with MaskedStretch versions in the next step which helped restore some star color. The effort was mostly in vain. The bright stars we very saturated due to my exposure time and ISO choices.
11. Restore colors and smooth star profiles to the heavily stretched stars using the method described by Vincent Peris (see Peris reference below).
12. Compress dynamic range. I attenuated the effect significantly by using a basic luminance mask that was not stretched in any way to specifically target the galaxy. I needed to do this because the image does not have sufficient SNR to push the dynamic range hard (HDRMultiscaleTranform w/ luminance mask).
13. Increase local contrast slightly (LocalHistogramEqualization; kernal radius: 128, contrast limit:1.2, amount: 0.58).
14. Reduce luminance noise (ACDNR to luminance with luminance mask).
15. Reduce chrominance noise (ACDNR to chrominance with strong luminance mask).
16. Small contrast curve (CurvesTransformation).
17. Increase galaxy color saturation (CurvesTransformation w/mask selecting galaxy only).
18. Increase star color saturation (CurvesTransformation w/mask selecting stars).
19. Reduce green. Here I had to be careful because removing green too aggressively compromised my ability to add “diversity” to the galaxy color later on. I only applied 63% as a consequence (SCNR to green; amount 0.63).
20. Increase red color saturation of HII regions (ColorSaturation w/mask selecting HII regions).
21. Increase diversity of yellows and blues in galaxy by applying a subtle curve to the Lab *b channel having an inflection point at the origin while deepening the blues and brightening the yellows. (CurvesTransformation w/strong galaxy mask; applied twice).
22. Reduce background to a median background value of about 0.1 (CurveTransformation).
23. Remove red color noise in background along the edges of the image likely caused by weak signal and imperfect flattening (SCNR to red w/range mask protecting galaxy and stars).
24. Match median values of G and B channels using a small preview and lowering the blackpoint slightly. This helps remove any small color casts. (HistogramTransformation).
25. Enhance yellow in galaxy by lower midtones slider of red channel to add some more red since red and green combine to produce yellow. This worked decent to add a bit of variety to the color (HistogramTransformation w/mask selecting galaxy).
26 Final Crop (DynamicCrop).
27. Set ICC profile to sRGB for web publishing (ICCProfileTransformation).
References
“Dynamic Range and Local Contrast” software tutorial by Vincent Peris:
http://pixinsight.com/tutorials/NGC7023-HDR/index.htmlManfred Schwarz’ galaxy tutorial:
http://www.astrophoto.at/PixInsight/
Comments