New kit and Pixinsight workflow: critique my image of M81 and M82, Bode's Galaxy and the Cigar Galaxy

[SiD the Turtle]

Hi all,

I'd like some constructive criticism on my workflow and thoughts on the my final image of M81 and M82. I've technically been an astrophotographer for a few years, but as a parent in the cloudy UK, this is the first target where I feel I have enough data at least, though from a less than ideal focal length and image scale.

I put together a new travel rig (as in take on a plane travel rig) which has compromises on pixel scale, but hopefully that drizzle can resolve. Giving it a test run, seeing as it was galaxy season taking a picture of M81 and M82 seemed sensible, though on this rig it'll be quite small.

My own critique of the final image is that it's a little cold, and the detail is lost compared to photos I've seen online, I think because this rig isn't really the best fit for such small objects. Any steps I've missed?

Final Image

Kit

• William Optics Redcat 51 v3
• ZWO AM3 + carbon fibre tripod
• ZWO ASI2600MC-DUO (with the integrated guide cam for zero flexure + ease of travelling and setup)
• Optolong L-Pro light pollution filter
• ZWO ASIAIR for image acquisition
• 129 x 180s exposures at -20c, or ~6 and a half hours of exposure
• Darks, biases and flats taken
• Bortle 5 suburbia, lots of moon but hopefully the filter helps there

Process

Everything was accomplished in PixInsight, I am trying a workflow that is automated as possible. I am not an artist, and I struggle with the processing part, so hoping I can be as 'mathematical' as possible. Images before stretch have been saved here as pngs with STF autostretch so they are visible.

1. Weighted Batch Pre-processing default settings + drizzle x2 (image resized to 25% of original here as it's huge)

2. Dynamic crop

3. SpectrophotometricColourCalibration - didn't do much but that could be STF?

4. Auto Dynamic Background Extraction (ADBE) - a new script that doesn't use AI (https://www.youtube.com/watch?v=XqXfzr1ZLdk), seems to work well and way faster than using DBE

5. NoiseXTerminator (trial, I want to use GraXpert but it's so slooooooooow on my laptop)

6. Statistical Stretch (auto stretch tool from the same guy as Auto DBE above) - seeing some faux 'nebulosity' here?

7. Manual histogram transform to make the contrast a bit deeper

8. Star Reduction using this: https://pixinsight.com/forum/index.php?threads/new-script-starreduction.21109/ to put the focus on the galaxies

Edited June 20 by SiD the Turtle

[Elp]

Looks good.

I don't use PI but colour calibration should have aligned the r, g, b histogram peaks but the image looks to have more green in it, that's one thing to address.

You normally have to boost colour saturation and contrast too to make faint detail pop. Also with galaxies they can benefit from shadow/highlight control to boost the shadows for the dark dust but dial back the highlights slightly to not to blow out the bright cores. I'd guess the highlights level is too high ATM because the stars are all the same colour, most starfields have blue and red stars present, again something colour calibration should have sorted but also requires a saturation boost to get deep defined colours in the stars, light pollution can also wash out the colour.

[Clarkey]

I would leave your noise reduction to the end (or nearly the end of processing). I would also stretch manually so you can determine how far to go with the data. For galaxies you may want to consider a masked stretch and then follow this with something like HDRMT. This will give you better definition and contrast - also the core will not be blown out. SCNR will remove the green cast although DBE / ABE should do this quite well. FWIW I normally do my background, then the SPCC. A bit more colour would help too.

Ultimately, there is no right and wrong - just practice. I have been imaging for 4 years now and I still consider myself a competent beginner. Strangely I have just been working on a Bodes galaxy this afternoon. I did have the advantage of an f4 12" Newtonian though😁 (Courtesy Roboscopes). However, I have been having major flat issues which is causing real difficulties....

[Oddsocks]

Being critical as you asked, I think I would have expected a little more "crispness" in the image from a refractor with six hours of data, the final image appears a little soft, but perhaps the sky conditions were against you for this object, the focus a little off, or the settings you used in Weighted Batch Pre-Processing were not optimised.

I still avoid WBPP in PI and do my own calibration steps manually, I like to watch what is happening at each step of the calibration and integration process so that I can go back and tweak the appropriate settings if necessary, I feel that using WBPP takes you away from that immediacy and when presented with a poor final image output from WBPP you have no idea which step during WBPP needs adjusting to improve the output.

I think it's interesting that the lead software designer of PI still maintains that the output of WBPP, and the predecessor BPP, were only provided to give an early indication of what detail is in the data and to achieve the best from the data the data reduction settings used in WBPP might require fine tuning, or use manual data reduction steps, but without the experience of calibrating and integrating subs manually in PI how would a user ever know what to adjust in WBPP settings to improve the output?

Anyhow, a little SNCR in PI to remove the green cast, a boost of selective colour saturation in PI, just applied to the red and blue parts of the spectrum, and to bring out more detail in the galaxy use HDRWaveletTransform (applied to 5 layers for this screen-grab with Lightness Mask enabled) will bring little more life into the galaxy cores.

Applying the above suggestions to a screen grab of your final image from the forum produces the result below, although of low quality and with rendering artefacts due to working with a screen-grab it shows that there is more to reveal in your data and maybe a careful review of the settings used in WBPP will give you a better starting point before post-processing begins, and which would allow you more freedom in the later stages.

 

Edited June 20 by Oddsocks
Rewrite of 3rd paragraph to improve clarity

[SiD the Turtle]

Thanks all for the input so far, given me lots of ideas to try and processes I hadn't heard of, keep them coming!

2 hours ago, Elp said:

I'd guess the highlights level is too high ATM because the stars are all the same colour, most starfields have blue and red stars present, again something colour calibration should have sorted but also requires a saturation boost to get deep defined colours in the stars, light pollution can also wash out the colour.

Yes the raw subs have colour in the stars, as done the raw stacked image, the levels must be pulling out the colour. Maybe the cause of the blown out core too.

1 hour ago, Clarkey said:

Strangely I have just been working on a Bodes galaxy this afternoon. I did have the advantage of an f4 12" Newtonian though😁

Beautiful, is there HA data in there for the reds?

53 minutes ago, Oddsocks said:

...

Thanks for the detailed breakdown, lots to unpack and consider, cheers!

Edit: even a little SCNR makes the image so much better!

Edited June 20 by SiD the Turtle

[Clarkey]

3 minutes ago, SiD the Turtle said:

Beautiful, is there HA data in there for the reds?

Yes it is an HaLRGB. It was really a test image for a new camera, but still worth processing.

[SiD the Turtle]

Thanks for all the input everyone. I took another stab using StarXterminator to pull out the stars, use HDRMT on the starless image, denoise it, recombine the stars with Pixel Math then run SCNR (colour calibration makes everything too green for my liking).

I've also posted the plain widefield, I think it has a nice look. Also flabbergasted that when I annotated it just for fun, it found 386 galaxies!

 

[Oddsocks]

IMO you can still increase the colour saturation a little, either wavelength selectively just in the blue and red parts of the spectrum, or across the whole spectrum, using the ColorSaturation tool (under IntensityTransformations in P.I.).

The output after applying SpectroPhotometricColorCalibration in P.I. is colour balanced but always low on saturation and usually needs a degree of colour saturation boost, just how much is down to the users preference and the level of chrominance noise present in the image.

You should be able to enhance the red filaments that surround the core of M82 and also differentiate better the colour temperature of the stars in the cropped view, just by applying a little saturation boost.

But otherwise a better result than the first version I think and the longer/deeper you can go in this part of the sky reveals more and more of the dust and gas clouds framing m81/m82, which are just starting to appear in the un-cropped view.

 

[SiD the Turtle]

Thanks @Oddsocks I'll give it a try tonight and see what I can do!

And yes indeed, the more I stare at the widefield, the more I spot those dust and gas clouds... beautiful. Would be great to get another couple of full nights' data here when I have clear skies and it's not the summer solstice so it actually gets dark!