Jump to content

  •  

CNers have asked about a donation box for Cloudy Nights over the years, so here you go. Donation is not required by any means, so please enjoy your stay.

Photo

Finally Have Good Data To Learn the Processing Side

  • Please log in to reply
10 replies to this topic

#1 AstroVagabond

AstroVagabond

    Vostok 1

  • -----
  • topic starter
  • Posts: 184
  • Joined: 28 Nov 2020
  • Loc: San Mateo, CA

Posted 18 October 2021 - 09:45 PM

When I became a beginner in December 2020 I separated my learning challenge into two categories. 

 

1.) Data Acquisition : Assemble gear, learn how to use, and discover and implement the best practices and processes to acquire good data. 

2.) Data Processing : Learn to process the data using Astro Pixel Processor and PixInsight.

 

I think I'm at a good place on how to collect good data. Now the hard part starts. smile.gif First using Astro Pixel Processor, then PixInsight. Here is my first attempt at integrating and combining LHaRGB data in APP. I have much to learn being a beginner image processor as you can see from the colors, or lack of colors, in the image below. Here is 17 hours of a 30 hour data acquisition project for M 31. 

 

M31 LHaRGB Project Iteration 2
 

 


  • Benschop, Sky King, Dren and 2 others like this

#2 Benschop

Benschop

    Viking 1

  • *****
  • Posts: 509
  • Joined: 12 Jan 2015
  • Loc: Ontario

Posted 18 October 2021 - 09:59 PM

Indeed - you have good data.

I am not an image processing expert... I'm on the same learning curve.

What I have found helpful, is to document workflows.

The ones that didn't work out, I ditch.

The good workflows, I keep, and I reference these steps for the next image processing.


  • AstroVagabond likes this

#3 AstroVagabond

AstroVagabond

    Vostok 1

  • -----
  • topic starter
  • Posts: 184
  • Joined: 28 Nov 2020
  • Loc: San Mateo, CA

Posted 18 October 2021 - 10:01 PM

Indeed - you have good data.

I am not an image processing expert... I'm on the same learning curve.

What I have found helpful, is to document workflows.

The ones that didn't work out, I ditch.

The good workflows, I keep, and I reference these steps for the next image processing.

Great feed back and I agree. 


  • Benschop likes this

#4 Oort Cloud

Oort Cloud

    Apollo

  • -----
  • Posts: 1,025
  • Joined: 19 Nov 2020
  • Loc: New Jersey, USA

Posted 18 October 2021 - 10:12 PM

I agree, the colors are a bit off, but good data indeed. Lots of nice deep detail in the dust lanes. Can't help with APP or PI specifically, as I'm a Siril/GIMP guy myself, but a Photometric Color Calibration is where I'd start. I'm fairly certain both of those have something along those lines. FWIW, I always end up tweaking colors afterward, but PCC usually gives me a very nice starting point.
  • nebulasaurus and AstroVagabond like this

#5 jonnybravo0311

jonnybravo0311

    Mercury-Atlas

  • *****
  • Posts: 2,951
  • Joined: 05 Nov 2020
  • Loc: NJ, US

Posted 18 October 2021 - 10:33 PM

Ooooh... purple! Where's Mike? :D

 

OK, let's start with the basics. You've got what appears to be very decent data. Here's a "get a quick image" process. I'm going to ignore the Ha for now, and just work with the RGB. Let's get that data processed... then we can play with luminance and Ha enhancements. Note, the process I'm about to lay out is _NOT_ how I would approach and edit this data for a final presentation. This will be a quick and dirty method to give you an idea of how your data looks when built into an image.

 

Step 1: combine your R, G and B channels using the ChannelCombination process. To everyone who is furiously smashing their keyboards typing responses to the effect of "that's absurd, why are you combining the channels as step 1???" I ask you to read the previous paragraph again. I'm just laying out the process to turn Bill's data into a less purple version.

 

Step 2: Crop. With dithering (you _were_ dithering, right?) and setting up / tearing down and plate solving tolerances and more, you're going to have stacking artifacts that need to be removed. I can't tell you how much you'll need to crop, but I typically start at around 98% and work down from there.

 

Step 3: DynamicBackgroundExtraction. Lay out your samples in areas outside of M31, M110, etc. Ensure you've got good coverage and a decent tolerance. Trial and error here until you get a good model. Use subtraction, not division.

 

Step 4: PhotometricColorCalibration. Use the default settings. Put in the coordinates / pixel size / focal length... all should be filled in automatically by clicking the "Acquire from Image" button.

 

Step 5: ScreenTransferFunction to do an unlinked auto-stretch

 

Step 6: HistogramTransformation. Drag the blue triangle from the ScreenTransferFunction to the bottom of the HistogramTransformation process, then drag the blue triangle from the HistogramTransformation process to your image.

 

That's it. At this point you should have an image that is (hopefully) not purple.


  • Benschop, galacticinsomnia, Mike in Rancho and 1 other like this

#6 Mike in Rancho

Mike in Rancho

    Surveyor 1

  • -----
  • Posts: 1,624
  • Joined: 15 Oct 2020
  • Loc: Alta Loma, CA

Posted 18 October 2021 - 11:26 PM

Moi?  Hiding from this one...

 

This is a very special purple, and I think we have to go back to the beginning.  As you in fact surmised with your basic repair workflow.  laugh.gif

 

I figured I could just load up the jpg and push a little green bias, like Eyal noticed in an image of mine recently (and it worked).  But no!  The purple here has serious staying power.  shocked.gif

 

That said, even with the jpg, the core can be tamed, and much extra detail brought forth.  Alas, that detail is all...purple.  elephant.gif

 

Uncertain if things got off track with processing, or pre-processing (the missing step 1.5 above - maybe certain stacking settings in APP or PI?).  confused1.gif


Edited by Mike in Rancho, 18 October 2021 - 11:27 PM.


#7 terry59

terry59

    Voyager 1

  • *****
  • Posts: 10,552
  • Joined: 18 Jul 2011
  • Loc: Colorado, USA

Posted 19 October 2021 - 07:42 AM

Your problem, IMO, is using APP to process your data. I rented it for a year and tried processing with it. I got it to work but my standard processing tools are far easier.

 

Adam Block can teach you PI


  • jonnybravo0311 and AstroVagabond like this

#8 AstroVagabond

AstroVagabond

    Vostok 1

  • -----
  • topic starter
  • Posts: 184
  • Joined: 28 Nov 2020
  • Loc: San Mateo, CA

Posted 19 October 2021 - 12:17 PM

Your problem, IMO, is using APP to process your data. I rented it for a year and tried processing with it. I got it to work but my standard processing tools are far easier.

 

Adam Block can teach you PI

 

I recently (October 8th) downloaded the trial license of PixInsight as my APP renters license expires in January 2022. PixInsight has been on my learning roadmap from the beginning. With the APP license expiring soon it was time to make the move.

 

I had an email conversation with Adam on October 10th to seek clarification about his online training courses. Shortly after I signed up last week for Adam's Fundamentals online course available at Adam Block Studios.

 

Seems we may be in agreement.

 

A key factor in my ability to make progress as a beginner was taking the time upfront to create a roadmap for gear acquisition, education, skill development, and budget expenditures. 


  • terry59 and jonnybravo0311 like this

#9 AstroVagabond

AstroVagabond

    Vostok 1

  • -----
  • topic starter
  • Posts: 184
  • Joined: 28 Nov 2020
  • Loc: San Mateo, CA

Posted 19 October 2021 - 12:33 PM

Ooooh... purple! Where's Mike? laugh.gif

 

OK, let's start with the basics. You've got what appears to be very decent data. Here's a "get a quick image" process. I'm going to ignore the Ha for now, and just work with the RGB. Let's get that data processed... then we can play with luminance and Ha enhancements. Note, the process I'm about to lay out is _NOT_ how I would approach and edit this data for a final presentation. This will be a quick and dirty method to give you an idea of how your data looks when built into an image.

 

Step 1: combine your R, G and B channels using the ChannelCombination process. To everyone who is furiously smashing their keyboards typing responses to the effect of "that's absurd, why are you combining the channels as step 1???" I ask you to read the previous paragraph again. I'm just laying out the process to turn Bill's data into a less purple version.

 

Step 2: Crop. With dithering (you _were_ dithering, right?) and setting up / tearing down and plate solving tolerances and more, you're going to have stacking artifacts that need to be removed. I can't tell you how much you'll need to crop, but I typically start at around 98% and work down from there.

 

Step 3: DynamicBackgroundExtraction. Lay out your samples in areas outside of M31, M110, etc. Ensure you've got good coverage and a decent tolerance. Trial and error here until you get a good model. Use subtraction, not division.

 

Step 4: PhotometricColorCalibration. Use the default settings. Put in the coordinates / pixel size / focal length... all should be filled in automatically by clicking the "Acquire from Image" button.

 

Step 5: ScreenTransferFunction to do an unlinked auto-stretch

 

Step 6: HistogramTransformation. Drag the blue triangle from the ScreenTransferFunction to the bottom of the HistogramTransformation process, then drag the blue triangle from the HistogramTransformation process to your image.

 

That's it. At this point you should have an image that is (hopefully) not purple.

 

Once again I really appreciate you sharing your knowledge. This image is an interim image as I still have ~ 13 hours of data to collect. I published it to get some feedback on the data quality which I'm getting and appreciate. It let's me know I'm on the right track when it comes to collecting data. This signals to me it's time to start to learn image processing.

 

I've started the PixInsight journey which will be my focus going forward as I outline in my comments above. But APP will serve a useful purpose through January 2022 as I start up the PixInsight learning curve. I'm a slow starter but always get to where I need to go. I'll keep your outline handy as I reprocess the M31 data set multiple times over the next couple of months.

 

Thank you to you, and everyone, for sharing feedback and knowledge. As a beginner this is very important to me.


Edited by AstroVagabond, 19 October 2021 - 12:39 PM.

  • Benschop and jonnybravo0311 like this

#10 jonnybravo0311

jonnybravo0311

    Mercury-Atlas

  • *****
  • Posts: 2,951
  • Joined: 05 Nov 2020
  • Loc: NJ, US

Posted 19 October 2021 - 12:42 PM

Once again I really appreciate you sharing your knowledge. This image is an interim image as I still have ~ 13 hours of data to collect. I published it to get some feedback on the data quality which I'm getting and appreciate. It let's me know I'm on the right track when it comes to collecting data. This signals to me it's time to start to learn image processing.

 

I've started the PixInsight journey which will be my focus going forward as I outline in my comments above. But APP will serve a useful purpose through January 2022 as I start up the PixInsight learning curve. I' a slow starter but always get to where I need to go. I'll keep you outline handy as I reprocess the M31 data set multiple times over the next couple of months.

 

Thank you to you, and everyone, for sharing feedback and knowledge. As a beginner this is very important to me.

It's my pleasure to offer up whatever knowledge I have acquired during my journey. I had, and continue to receive, plenty of help along the way. Paying it forward seems appropriate :).

 

As you'll notice in my process, I clearly mentioned it was not how I'd do my final editing, but rather that it was a quick and easy way to get a color-balanced non-linear image you could evaluate to determine data quality. You certainly don't need to do it. I just found the purple to be quite a distraction, and thought I'd provide you with a pretty easy workflow should you care to use it on any interim images you created.


  • Benschop and AstroVagabond like this

#11 ntph

ntph

    Messenger

  • -----
  • Posts: 432
  • Joined: 20 Oct 2012
  • Loc: Sudbury, ON

Posted 19 October 2021 - 04:43 PM

And before Bob jumps in here--wink.gif

 

Just in case you don't already have one or more of these great reference sources:

 

Warren Keller's Inside PixInsight ;   RBA's  Mastering PixInsight ;  Charles Bracken's Deep Sky Imaging Primer

Kayron's  LightVortexAstronomy website PixInsight tutorials

Jon Rista's PixInsight tutorials

and lots of others... plus what you get here from lots of helpful folks like Jonny

 

Adam's tutorials and YouTube videos are very well produced and show how to approach using PI's deep toolbox to not only create good images but to analyze them and solve problems. With the approach you outlined about how you have made it this far and how you intend to go on, I have no doubt that you will quickly start getting the results you're hoping for. And start telling us about some of the pearls you get from Adam...


  • jonnybravo0311 likes this


CNers have asked about a donation box for Cloudy Nights over the years, so here you go. Donation is not required by any means, so please enjoy your stay.


Recent Topics






Cloudy Nights LLC
Cloudy Nights Sponsor: Astronomics