Username:

Password:

UKAI is moving to Facebook

Author Topic: Help with PI please  (Read 4075 times)

0 Members and 1 Guest are viewing this topic.

Offline Phillyo

  • Poster God
  • *****
  • Posts: 1492
Help with PI please
« on: 19:25:47, 16 December, 2014 »
Calling all PixInsight GURU's...I need a little help. I'm trying to process some more data I have for IC434 from last night, a total of 6hours and 20min (4x20min and 10 x 30min) but I'm struggling with a couple of different processing techniques. I'll post the two pictures I've come out with so far as I don't know which I prefer, which is best, which has gone too far etc etc and I'll also post a link to the stacked data if anyone else wants to have a play and see what they can do?

First image is using my normal routine which, depending on the image, goes something like:

Deconvolution, DBE, stretch, HDR, curves, ACDNR then star mask for a slight star reduction.

Second image is a slightly different one with the following:

Deconvolution, DBE, ATrousWaveletTransform, MultiscaleMedianTransform, stretch, HDR, ACDNR then star mask for slight star reduction.

However with the second one the background, I think, is starting to look a little soft...too much noise reduction before stretching perhaps? (I followed an online tutorial for it but not sure?) However it is slightly more contrasty...?

I'm sure there's more things you can do in PI to make this images "even more nicerer" ;) but I'm such a noob I don't know them!

Anyway, thanks for any help and opinions. All are greatly received!!

Phil

https://www.dropbox.com/s/cz9sr3nm8kdeco7/light-FILTER_Ha-BINNING_1.xisf?dl=0   ------Stacked images.
First


Second


Astrotrac | Avalon Linear Fast Reverse with EQmod | WO Star 71 | Astrodon 3nm Ha, 5nm SII and 5nm OIII | Astro-modded Canon 450D | QSI 583wsg | Lodestar X2 Guidecam | Sequence Generator Pro | Pixinsight

Online chris.bailey

  • Administrator
  • Galactic Poster
  • *****
  • Posts: 10012
  • If at last you do succeed, don't try again
    • Rosemount Observatory
Re: Help with PI please
« Reply #1 on: 19:36:49, 16 December, 2014 »
Phil

You may well find DBE to be the problem here. There is precious little true background in that image so you end up sampling signal. I think you are also doing a few things in the wrong order. ATWT and MMT should be after stretch.

I'll download and have a play.

Chris
LX200|ZS70|FSQ85|FLT110|Altair DF250RC|EQ6 Pro(Rowan Belt Mod)|ParamountMX
ATIK383L+/EFW2/OAG|Lodestar|Baader 36mm LRGBHaSIIOIII
Starlight Express SXVRH16/ONAG/FW|Lodestar X2|Baader 2" Filters
Starlight Express SXVRH814/ONAG/FW|LodestarX2|Baader 2" Filters
Lunt LS60PTBF1200|DMK41|Quark Chromo
Samyang 135mm f1.8

Offline Phillyo

  • Poster God
  • *****
  • Posts: 1492
Re: Help with PI please
« Reply #2 on: 19:59:03, 16 December, 2014 »
Hi Chris,

I'll give it a try without DBE and it's interesting what you say about the ATWT and MMT being done after the stretch because this tutorial said to do it before the stretch?

Phil
Astrotrac | Avalon Linear Fast Reverse with EQmod | WO Star 71 | Astrodon 3nm Ha, 5nm SII and 5nm OIII | Astro-modded Canon 450D | QSI 583wsg | Lodestar X2 Guidecam | Sequence Generator Pro | Pixinsight

Online chris.bailey

  • Administrator
  • Galactic Poster
  • *****
  • Posts: 10012
  • If at last you do succeed, don't try again
    • Rosemount Observatory
Re: Help with PI please
« Reply #3 on: 22:31:09, 16 December, 2014 »
Phil

I never use Atrous Wavelet Transform these days but you are right it should be linear and only MLT for noise control on linear data.

This is my take on it, trying to maintain the overall contrast balance.

DBE using only a handful of very carefully chosen samples.
TGV Denoise (could probably refine it further but default turned down by a factor of 10 and a bit and with local support enabled)
Masked Stretch (125 iterations and with the clipping fraction reduced by a factor of 10, 0.035 background level)
HDRMT (6 layers with L mask enabled)
LHE (radius about 150) with L mask applied
Dark Structure Enhance Script (2 iteration of 0.2)
Unsharp Mask (turned down to about 1.6) with L Mask applied

I tried deconvolution but it made next to no difference other than mess up the star shape and getting rid of ringing artefacts on bright areas makes it a low value proposition in my view. A gentle unsharp mask at the end probably does it better.

Chris

ps that 3nm Ha filter does a wonderful job with Alnitak's glare!
LX200|ZS70|FSQ85|FLT110|Altair DF250RC|EQ6 Pro(Rowan Belt Mod)|ParamountMX
ATIK383L+/EFW2/OAG|Lodestar|Baader 36mm LRGBHaSIIOIII
Starlight Express SXVRH16/ONAG/FW|Lodestar X2|Baader 2" Filters
Starlight Express SXVRH814/ONAG/FW|LodestarX2|Baader 2" Filters
Lunt LS60PTBF1200|DMK41|Quark Chromo
Samyang 135mm f1.8

Online Roger B

  • Administrator
  • Galactic Poster
  • *****
  • Posts: 9010
    • Capricorn1 observatory
Re: Help with PI please
« Reply #4 on: 01:45:58, 17 December, 2014 »
That's a really nice Image,   even without your PI advise. 

I'm still struggling with PI with the blotchy background stuff.    gonna call it a day soon.

Rog 

Scopes - TS65q, W/O 110 Triplet
Mount - EQ6 PRO via  Eqmod,  Rowan belt modded
Cameras - SX694, QSI583WSG,Lodestar Guidecam, GP-Cam, QHY5II.
TS 9mm OAG
Quantum 7x1.25 EFW, Baader LRGB,Ha,OIII,SII Filters 
Quark chromo

SGPro, Pixinsight.

************AND NO TIME TO USE IT************

Offline Nomis Elfactem

  • Administrator
  • Galactic Poster
  • *****
  • Posts: 13011
    • The View From A Room
Re: Help with PI please
« Reply #5 on: 08:10:54, 17 December, 2014 »
Rog,  have you tried to process an image in PS to compare? I'm still on the fence where PI is concerned and a "blotchy" background is my main concern looking at images from you guys (plus its obvious complexity / time hungry to learn) . To see the same image processed in both would be great to see.

S.
Simon

Scopes: Astro-Tech AT-111EDT Triplet, TS65ED Quad, Orion ST80, Modded PST-90 Solar Scope, PST Cak (on loan)
Cameras: SXVF H694, Atik 16ic, Canon EOS 600d, DMK41, DMK21, QHY 5L-II (mono & colour)
Accessories: SX USB Filter Wheel, SX OAG, Baader LRGB Ha SI OII Filters, SharpSky Focuser
Mount: EQ6 (EQMOD), SW Star Adventurer, plus a lot (and I mean a lot) of other bits and pieces

Offline Phillyo

  • Poster God
  • *****
  • Posts: 1492
Re: Help with PI please
« Reply #6 on: 09:01:57, 17 December, 2014 »
Simon,

Feel free to download the stacked image and try processing in photoshop? I think it should open it as it's an XSIF file which supports the TIFF format (I think). If it doesn't then let me know and I'll upload a version to play with on photoshop. As you say it'll be interesting to compare the two :)

Phil
Astrotrac | Avalon Linear Fast Reverse with EQmod | WO Star 71 | Astrodon 3nm Ha, 5nm SII and 5nm OIII | Astro-modded Canon 450D | QSI 583wsg | Lodestar X2 Guidecam | Sequence Generator Pro | Pixinsight

Online chris.bailey

  • Administrator
  • Galactic Poster
  • *****
  • Posts: 10012
  • If at last you do succeed, don't try again
    • Rosemount Observatory
Re: Help with PI please
« Reply #7 on: 09:41:21, 17 December, 2014 »
Phil

On the basis that PI have only just invented the XSIF format there is no chance of PS opening it.

Chris
LX200|ZS70|FSQ85|FLT110|Altair DF250RC|EQ6 Pro(Rowan Belt Mod)|ParamountMX
ATIK383L+/EFW2/OAG|Lodestar|Baader 36mm LRGBHaSIIOIII
Starlight Express SXVRH16/ONAG/FW|Lodestar X2|Baader 2" Filters
Starlight Express SXVRH814/ONAG/FW|LodestarX2|Baader 2" Filters
Lunt LS60PTBF1200|DMK41|Quark Chromo
Samyang 135mm f1.8

Offline Phillyo

  • Poster God
  • *****
  • Posts: 1492
Re: Help with PI please
« Reply #8 on: 09:49:31, 17 December, 2014 »
I'm not going to pretend I understand half of what you did to the image Chris but it looks a lot sharper than my effort. I'll have a look tonight at some of the things you mentioned and see what I can do. Nothing else to do but play with data when it's cloudy outside eh?

And yes, I love this Astrodon filter!! Not cheap, not by any means, but I think it's totally worth it.

I'll save the stacked image as a TIFF file and put it on dropbox for anyone wanting to have a go with it.

Phil
Astrotrac | Avalon Linear Fast Reverse with EQmod | WO Star 71 | Astrodon 3nm Ha, 5nm SII and 5nm OIII | Astro-modded Canon 450D | QSI 583wsg | Lodestar X2 Guidecam | Sequence Generator Pro | Pixinsight

Offline Nomis Elfactem

  • Administrator
  • Galactic Poster
  • *****
  • Posts: 13011
    • The View From A Room
Re: Help with PI please
« Reply #9 on: 09:55:03, 17 December, 2014 »
Yeap, Chris was (as ever) right ... it won't open in CS4.... just tried it  :(
Simon

Scopes: Astro-Tech AT-111EDT Triplet, TS65ED Quad, Orion ST80, Modded PST-90 Solar Scope, PST Cak (on loan)
Cameras: SXVF H694, Atik 16ic, Canon EOS 600d, DMK41, DMK21, QHY 5L-II (mono & colour)
Accessories: SX USB Filter Wheel, SX OAG, Baader LRGB Ha SI OII Filters, SharpSky Focuser
Mount: EQ6 (EQMOD), SW Star Adventurer, plus a lot (and I mean a lot) of other bits and pieces

Online chris.bailey

  • Administrator
  • Galactic Poster
  • *****
  • Posts: 10012
  • If at last you do succeed, don't try again
    • Rosemount Observatory
Re: Help with PI please
« Reply #10 on: 10:38:42, 17 December, 2014 »
Phil

As there are a few people using PI now, I'll post up screenshots of each stage of what I did later. Anyone with current versions of PI then can download you image and follow along, hopefully improving on things too.

So called blotchy backgrounds is an interesting one and I don't think you can lay the blame on PI (at least not in total). TGV and ACDNR both do excellent jobs of noise reduction. TGV in particular is very settings sensitive and this is something PI are working on for the new release coming just after Xmas (so March/April then). They inadvertently released the new version of TGV with the last update and it has been totally re-worked.

On a couple of images I have got very frustrated in not being able to get noise reduction working well BUT adding more data, I found a lot of what I had seen as "blotch" was actually real data or with more data was resolved to not be data at all. Most of the sky is not totally empty, especially in Ha and with lots and lots of data, noise reduction becomes less and less necessary and the tools work a lot better.

I have said before that you can probably get the same or similar results in PS or Startools that you can in PI so its a matter of personal choice rather than one being better than the other. When I look at some of the lengthy PS tutorials with 30 layers and hundreds of processing steps though, PI seems the simpler route to me.

Chris
LX200|ZS70|FSQ85|FLT110|Altair DF250RC|EQ6 Pro(Rowan Belt Mod)|ParamountMX
ATIK383L+/EFW2/OAG|Lodestar|Baader 36mm LRGBHaSIIOIII
Starlight Express SXVRH16/ONAG/FW|Lodestar X2|Baader 2" Filters
Starlight Express SXVRH814/ONAG/FW|LodestarX2|Baader 2" Filters
Lunt LS60PTBF1200|DMK41|Quark Chromo
Samyang 135mm f1.8

Offline Phillyo

  • Poster God
  • *****
  • Posts: 1492
Re: Help with PI please
« Reply #11 on: 10:57:48, 17 December, 2014 »
That would be great Chris, thank you for that. Really appreciate it.

And yes! People are more than welcome to use the data to play with. It's only just over 6 hours and ideally I'd like twice that really but with the terrible weather we've been having I'm happy at even 6!

It's definitely a lot easier to work with more data. Things are just smoother, less pushed etc to get the desired result.

I suppose the greatest thing about this hobby is that in a lot of respects there's no real "right way" of doing things, per se. If you achieve a result that you're happy with and it's what you want to show, then that's how you should do it. It's art. It's photography. :)
Astrotrac | Avalon Linear Fast Reverse with EQmod | WO Star 71 | Astrodon 3nm Ha, 5nm SII and 5nm OIII | Astro-modded Canon 450D | QSI 583wsg | Lodestar X2 Guidecam | Sequence Generator Pro | Pixinsight

Online chris.bailey

  • Administrator
  • Galactic Poster
  • *****
  • Posts: 10012
  • If at last you do succeed, don't try again
    • Rosemount Observatory
Re: Help with PI please
« Reply #12 on: 11:37:42, 17 December, 2014 »
Step 1 - Dynamic Background Extraction

One of the most useful tools in PI for removal of gradients, it is tricky to use on images like this where there is next to no background on which to place samples. The background model is very smooth so only really needs a handful of samples to generate a global background but these need to be carefully placed.
LX200|ZS70|FSQ85|FLT110|Altair DF250RC|EQ6 Pro(Rowan Belt Mod)|ParamountMX
ATIK383L+/EFW2/OAG|Lodestar|Baader 36mm LRGBHaSIIOIII
Starlight Express SXVRH16/ONAG/FW|Lodestar X2|Baader 2" Filters
Starlight Express SXVRH814/ONAG/FW|LodestarX2|Baader 2" Filters
Lunt LS60PTBF1200|DMK41|Quark Chromo
Samyang 135mm f1.8

Online chris.bailey

  • Administrator
  • Galactic Poster
  • *****
  • Posts: 10012
  • If at last you do succeed, don't try again
    • Rosemount Observatory
Re: Help with PI please
« Reply #13 on: 11:44:54, 17 December, 2014 »
Step 2 - Deconvolution - Local Support

Deconvolution can be thought of as of unblurring. Its very powerful but quite complex to use and can cause horrid ringing artefacts around bright stars. The first step in the process is to generate a local support image. this is not a mask per se but acts like one to help deconvolution avoid ringing.

Start with a copy of the image and using Multiscale Linear Transform with 5 layers, deactivate layers 1 and the residual. This helps Star Mask to detect true stars.

Now use star mask with a low setting of noise threshold and with Aggregate and Binarize checked.

Use the Histogram Tool to apply a slight stretch and then convolution to smooth the star profile.

Rename the image Local_Support
LX200|ZS70|FSQ85|FLT110|Altair DF250RC|EQ6 Pro(Rowan Belt Mod)|ParamountMX
ATIK383L+/EFW2/OAG|Lodestar|Baader 36mm LRGBHaSIIOIII
Starlight Express SXVRH16/ONAG/FW|Lodestar X2|Baader 2" Filters
Starlight Express SXVRH814/ONAG/FW|LodestarX2|Baader 2" Filters
Lunt LS60PTBF1200|DMK41|Quark Chromo
Samyang 135mm f1.8

Online chris.bailey

  • Administrator
  • Galactic Poster
  • *****
  • Posts: 10012
  • If at last you do succeed, don't try again
    • Rosemount Observatory
Re: Help with PI please
« Reply #14 on: 11:48:50, 17 December, 2014 »
Step 3 - Deconvolution - Point Spread Function (PSF)

Its very important to characterise the amount of blurring by building a PSF based on the star profiles in the image. 100 Stars is a good target and avoid those in the corner, where the shape may be less than ideal. Export the PSF based on the 100 stars sampled.

It is a good idea at this stage to save out the local_support and psf images for re-use.
LX200|ZS70|FSQ85|FLT110|Altair DF250RC|EQ6 Pro(Rowan Belt Mod)|ParamountMX
ATIK383L+/EFW2/OAG|Lodestar|Baader 36mm LRGBHaSIIOIII
Starlight Express SXVRH16/ONAG/FW|Lodestar X2|Baader 2" Filters
Starlight Express SXVRH814/ONAG/FW|LodestarX2|Baader 2" Filters
Lunt LS60PTBF1200|DMK41|Quark Chromo
Samyang 135mm f1.8

 

ukbuysellRemote Imaging from AustraliaSharpSkyblank APTUKAI on Facebook
Powered by SMF 2.0.15 | SMF © 2006, Simple Machines LLC
DarkBreak by DzinerStudio. Theme modified by The UKAI Team

Page created in 0.221 seconds with 38 queries.
TinyPortal © 2005-2012