Jump to content

sgl_imaging_challenge_2021_4.gif.6a323659519d12fc7cafc409440c9dbf.gif

IKO - M33 LRGBHa - Processing Competition


Grant
 Share

Recommended Posts

Well here is my first attempt at one of these IKO challenges. I see many suberb versions already, which was a little daugnting, but once I started looking at this excellent data, I was encouraged to keep trying. Apart froma final HLVG step in PS the data was combined and all processing done in ImagesPlus (IP)

M33(SGL)_LHaRGB_IP_PS(HLVG).thumb.jpg.8dbc1da815a1ed8cbdea6f33937a7c23.jpg

Main steps were:

  • Crop edges to remove stacking artefacts
  • Combine HaRGB using IP's mix LRGB tool, mapping Ha to R at 50%.
  • ArcSinH stretch
  • Colour balance to neutralise background
  • Addtional Hybrid ArcSinH stretch
  • Noise reduction using Frequency Filter Low Pass with inverted luminance mask
  • Curves to brighten galaxy without lifting the background
  • Open and crop the luminance data using same crop settings as for the HaRGB combine image
  • AcrSinH stretch
  • Feature Mask tool to create separate stars and galaxy layers
  • Mutiscale decomposition tool to sharpen galaxy only layer
  • Noise reduction to galaxy layer using Frequency Filter Low Pass with inverted luminance mask
  • Recombine luminance galaxy layer and stars layer
  • Add Luminance to HaRGB layer
  • Save LHaRGB combination as TIFF
  • Open TIFF in PS(CS2)
  • Apply HLVG (strong) to remove green cast in corners (NB I probably should have done this on the HaRGB layer before adding Lum, but I didn't want to go back and forth between PS and IP)
  • Save as TIFF and JPEG

I had another look at the data and decided to use the Ha mapped to red at 100%. I also paid a bit more attention to the overal colour balance to try to preseve some of the blue in the galaxy arms. The luminance data was then applied as Lab Lum at 50% rather than straight Lum as I felt that this preserved the colours a tab better, giving a softer blend in the star forming Ha regions. Hopefully the differences with the additional Ha are subtle, but noticable.

Anyway, here is my V2....

M33(SGL)_L(Lab)HaRGB(V2)_IP_PS(HLVG-S).thumb.jpg.f9978a626f7413a5599aefe00fa4d3b6.jpg

Thanks for looking.

Edited by geoflewis
  • Like 3
Link to comment
Share on other sites

Something I've noticed after looking at these on another screen, is I need a new monitor at home, or at least try to calibrate it again! Mine's so pink 😐

Link to comment
Share on other sites

Hi,

Excellent data. So pleased to be given a chance to process it. Anyway, this is my go at processing. Mostly done in PixInsight, with final tweaks in Photoshop:

Slight Crop on each channel

DBE on each channel

Slight NR (MLT) on Ha, R, G and B

Combined R, G and B into RGB colour

BN and CC of the RGB image

RGB stretched with Masked Stretch to preserve the colour

Ha stretched with HT to match mean intensity of the RGB

Ha added to the RGB in blend mode Lighten (MAX function in PixelMath) to give HaRGB

Lum given a very light Deconvolution, using a mask to protect the stars and background

Slight NR (MLT) on the Lum

Lum stretched with an initial application of ArcSinh, then HT to match mean intensity of the HaRGB

Slight LHE on the galaxy arms (using the same mask used for Deconvolution)

Ha added to the Lum, again in blend mode Lighten, to give HaRGB

HaRGB intensity matched to Lum with LinearFit

Combined the Lum with the HaRGB using Channel Combination to give LumHaRGB

Transfer to Photoshop as a 16bit TIFF

Curves, Levels and Selective Colour tweaks

Slight reduction in size of the smaller stars

Flatten and save as JPEG

Cheers

Bill

M33.jpg

Edited by A320Flyer
  • Like 5
Link to comment
Share on other sites

My effort with a combination of Pixinsight and Photoshop.

 

Mild crop of all frames to take black edges off.

Skipped DBE with this as I felt it gave too hard an edge and not enough fall off from edge of galaxy to background

Pixelmath to combine Ha and Red (ha*0.8)+(red*0.4)

Combine (Ha/R)GB

Colour calibration / background neutralization

Repair HSV Seperation

Mild RGB ArcSin stretch

Further histogram stretch

Histogram stretch to Lum

Combine LRGB

 

Move over to Photoshop.

Mild levels, curves

Carbonis Astronomy tools increase star colour, star reduction, contrast enhancement

Mild colour saturation

 

Back to PI

Another round of mild star reduction with Morphological Transformation (with star mask)

Couple of mild iterations of dark structure enhance to bring out some of the central structures and spirals

Range mask over the centre and some tweaks to hue and saturation.

 

Thanks for the data. I normally struggle with LRGB but really enjoyed working with this!

RGB_HSV_ArcSin_L_Ps_PI.thumb.jpg.1bfccc2a42927d5fff8234465e8c5219.jpg

Link to comment
Share on other sites

always been StarTools for me Grant.   Hawian Pallett Ha to Red, G to OII, B  to SII  - something a little different.
hawian_01_final_PNG.thumb.png.41e945e3f1bd11308eb0964338f7a3fe.png  (ver 1.7).

Edited by bonzodog
Link to comment
Share on other sites

This was interesting to process, I have always found M33 such a messy galaxy, really needs to tidy itself up a bit 😀

Processing info, below image.

m33-HaRGB-v5.thumb.png.d8b397ac8ed489dc39a2e40e2b8e2c2f.png

 

Pixinsight

Combine RGB via pixelmath
DBE
Balance channels with HistogramTransformation
Starnet
Multiple small stretches with HistogramTransformation on both stars and galaxy images

Export to Photoshop
Galaxy layer
multiple rounds of copy layer, curves, mask, copy merged to new layer
Combine Galaxy & stars

Back to PI

Combine Ha+RGB via pixelmath - I wanted to keep Ha subtle (that's the way I like it)

Export to Photoshop again
Adjust contrast, colour and saturation to taste
Flatten and export

  • Like 2
Link to comment
Share on other sites

Super data, thank you for sharing.

Processed in StarTools 1.7.458

This is the warts version as I need to redo the RGB process.

First step was load the hav2 as linear and bin | crop | wipe | autodev with ROI | denoise | saved as ha.tiff

Using the Compose module I loaded the R G B and L, though this is where I went wrong, I looked at the time time for each channel and when adding the luminance weighted it three times more which rather washed my colours away.

image.png.e826e51d0428a796b01aac6f2e2c29bf.png

RGB L process  loaded as linear and bin | crop | wipe | autodev with ROI | HDR |  mask and wavelet sharpening |  mask tweak for centre wavelet sharpening | decon | colour | saved as RGB.tiff

Load lum as linear and bin | crop | wipe | autodev with ROI | denoise | saved as lum.tiff

Compose module and opened saved ha tiff on red and as RGB legacy and no to colour interpolation.

Layer module and opened saved lum and set to lighten and 80% and saved as halum.tiff

Opened halum as none linear then layer module and opened RGB and set to lighten this created the final composition then turn tracking on and off and denoise and then a conservative star mask avoiding the galaxy stars and shrink stars | lift background colour and save | saved tiff to png for upload.

8537530_NewComposite20210319halumrgbv1.1.thumb.png.2576730f198605cab2794487f7805d57.png

 

Edited by happy-kat
  • Like 2
Link to comment
Share on other sites

Super data, thank you for sharing, second pass correcting prior overwhelming of the luminance data

Processed in StarTools 1.7.458

First step was load the hav2 as linear and bin | crop | wipe | autodev with ROI | denoise | saved as ha.tiff

Using the Compose module I loaded the R G B (I haven't yet worked on the luminance in a satisfactory way)

RGB L process  loaded as linear and bin | crop | wipe | autodev with ROI | HDR |  mask and wavelet sharpening |  mask tweak for centre wavelet sharpening | decon | colour | saved as RGB.tiff

Compose module and opened saved ha tiff on red and as RGB legacy and no to colour interpolation.

Layer module and opened saved RGB and set to lighten 90% this created the final composition then turn tracking on and off and denoise | lift background level and save | saved tiff to png for upload.

1511066563_NewComposite21032021RGBHAv4.1.thumb.png.ae34a1fab018e8b93b8c89b0ea5a3ec0.png

  • Like 1
Link to comment
Share on other sites

First time when I process something in PixInsight. Hard to learn but very useful tool.

Had big issues integrating Ha in RGB (tried for a week to process and combine Ha but without any satisfying results) so the image is without Ha.

Main steps (these steps were not performed in this order, they are written as I remember):

-Cropping with DynamicCrop
-DynamicBackgroundExtraction
-MultiscaleLinearTransform
-Matching the Images with LinearFit
-Stretching using HistogramTransformation
-Colour-Combining with ChannelCombination
-Performing the Colour Calibration with PhotometricColorCalibration
-Removing the Green Colour Cast with SCNR
-Applying Luminance Image with LRGBCombination
-BackgroundNeutralization
-ACDNR
-Deconvolution with DynamicPSF
-HDRMultiscaleTransform
-CurvesTransformation
-Reducing Star Sizes with MorphologicalTransformation
-Saturation Enhancements and Hue Shifts with CurvesTransformation

 

M33.thumb.png.58dca35a448d6ddfaf048ffd4d9ace03.png

Link to comment
Share on other sites

Well I found this very hard.  I rarely do LRGB imaging because of LP so my processing is a bit sketchy.  It has been great to look through all the images, especially @Goran combination of his top ten.  This was my benchmark but I couldn't get close to replicating it.  I have had a good few runs using all manner of PI techniques but in the end this was nearly all done in PS.   I found the hardest part of the process was controlling those blue arms.  The popped so readily with any sharpening/saturation processes and easily ended up dominating the image.  I also found that it was very easy to lose the outer halo when increasing contrast.

This has been my favourite processing challenge to date!

 

 

 

 

IKI_LRGB.png

  • Like 5
Link to comment
Share on other sites

Thanks for the opportunity to have a go with this data guys.

Processing started in PixInsight, some noise reduction using MLT with a starmask on each layer, and then layers each stretched to a similar extent using the histogram tool.

Channels combined using channel combination.

This RGB image, Ha and luminance files (all stretched) saved as TIFF files for use in PS.

Main RGB image colour balanced in PS, then Ha added a an RGB layer in screen mode.

Luminance then added, a final tiny bit of star reduction, tiny bit of selective sharpening on the core.

 

Thanks for looking..

 

M33LumDFPost.tif

Link to comment
Share on other sites

Just now, Derbyshire Dave said:

Thanks for the opportunity to have a go with this data guys.

Processing started in PixInsight, some noise reduction using MLT with a starmask on each layer, and then layers each stretched to a similar extent using the histogram tool.

Channels combined using channel combination.

This RGB image, Ha and luminance files (all stretched) saved as TIFF files for use in PS.

Main RGB image colour balanced in PS, then Ha added a an RGB layer in screen mode.

Luminance then added, a final tiny bit of star reduction, tiny bit of selective sharpening on the core.

 

Thanks for looking..

 

M33LumDFPost.tif 34.64 MB · 0 downloads

Whoops, meant to add the jpg, not tif.

 

M33LumDFPost.jpg

  • Like 1
Link to comment
Share on other sites

Here is my last minute entry. Perhaps controversially, I have done most of the heavy lifting in post by utilising machine learning courtesy of Luminar AI and Topaz DeNoiseAI. Surprisingly effective... Channels first combined in AstroPixelProcessor with only around 30% of the Ha blended in. Continuing with APP, I performed background calibration, star colour calibration with the adaptive blackbody + extinction model and a fairly heavy crop to remove edge artefacts and frame the galaxy centrally. Then into Luminar AI where I used the Enhance and Structure AI sliders just a little bit to bring out the spiral arms. I followed this up with a small adjustment using the small and medium details sliders to bring out the inherent grainy texture and detail in the dust. I then dialled down the highlights to fix the core. Lastly I ran the image through Topaz DeNoiseAI using the "AI Clear" algorithm with noise reduction and sharpening both set low. Following this I dialled up the sliders to recover original detail and reduce colour noise, which cleaned up artefacts quite well.

This whole procedure is certainly lazy, but I'm happy with the results!

 

M33-LHaRGB-cbg-csc-mod-St2-post-DeNoiseAI-clear.jpg

Link to comment
Share on other sites

Super data, thank you for sharing, third pass correcting prior mistakes with the luminance. mainly to bring out the extents of the galaxy and colour.

Processed in StarTools 1.7.458

First step was load the hav2 as linear and bin | crop | wipe | autodev with ROI | denoise | saved as ha.tiff

Using the Compose module I loaded the R G B  and luminance, though the latter was weighted to 20m (1/3 of the time as it was 3 x more and overpowering)

RGB L process  loaded as linear and bin | crop | wipe | autodev with very large ROI just 30 pixels in all round | HDR |  mask and wavelet sharpening |  mask smaller than lst and medium wavelet sharpening | colour | shrink | denoise saved as RGBL.tiff

Compose module and opened saved ha tiff on red and as RGBL legacy and no to colour interpolation.

Layer module and opened saved RGBL and set to lighten 90% this created the final composition then turn tracking on and off and denoise | saved tiff to png for upload.

1590671223_NewComposite29032021v1.6.thumb.png.038ab10fa4bb16503af41bd0f89477b6.png

  • Like 1
Link to comment
Share on other sites

We're having a week off from StarGaZine this week so we will announce the winners on Tuesday instead 🙂

Happy long bank holiday weekend everybody!

  • Like 1
  • Thanks 1
Link to comment
Share on other sites

Too late for competition, but I thought I'd post anyway - great set of data - thanks!

Processed in Pixinsight:
Demure DeNoise to all masters then...
Luminance:- DBE -> Deconvolution -> Noise Reduction with MLT -> Mild STF/HT; then galaxy core masked & further STF/HT; galaxy core mask inverted & CT to brighten core. 2 clones made & HDRMT applied to enhance large scale structure + small scale structure, then Pixel Math to blend results with original.
Red+Green+Blue:- Channel Combination -> PCC -> DBE; 2 stage STF/HT as per Luminance ->LRGB Combination -> ACDNR -> Curves Transformation to tweak saturation.

Cheers
Ivor

M33v2.jpg

  • Like 2
Link to comment
Share on other sites

  • 4 weeks later...

Only stumbled on this this morning - brilliant idea! It's so fascinating to see what a range of results come out of the same (excellent) data!

Although the party is over, I couldn't resist playing with it myself and figured I might upload it, just to add to the variety :)

I only used Photoshop and tried to keep it kind of natural and retain a sense of scale. 

441235021_M33Contest2.thumb.jpg.063cfb41511ce3881b1818fb24081d85.jpg

 

Seems like the contrast to the blacks comes out a bit harder in the JPG here than in my PS file ... 

 

Edited by Daniel Karl
  • Like 3
Link to comment
Share on other sites

  • 1 month later...

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
 Share

  • Recently Browsing   0 members

    No registered users viewing this page.

×
×
  • Create New...

Important Information

We have placed cookies on your device to help make this website better. You can adjust your cookie settings, otherwise we'll assume you're okay to continue. By using this site, you agree to our Terms of Use.