Jump to content


Photo

Food for thought... Speckle and AS2!

  • Please log in to reply
12 replies to this topic

#1 tjensen

tjensen

    Mercury-Atlas

  • -----
  • Posts: 2609
  • Joined: 16 Feb 2005
  • Loc: Chapel Hill, NC

Posted 28 February 2013 - 07:32 PM

Hey all,

Emil's interesting post on speckle interferometry and his amazing solar image got me thinking about an old program called specklem1. So I dusted it off and ran my last good Jupiter's L channel thru it and AS2! to see if I could see any differences.

Mert is the real expert on this software and has written a very useful guide. I basically just used the defaults. It isn't an intuitive program, and doesn't handle large AVIs, so you are forced to do some pre-processing. It's also very slow. So it's hard to decide about benefits at this stage.

Here is a comparison between Specklem1 ( left image, first 2500 frames of the capture) and AS2!(25% of 6400 frames used). No further processing was done on the specklem1 image. The AS version had some mild wavelets and I tried to balance the contrast a bit, but I didn't want to overdo the processing too much and ruin the comparison. No drizzle on this either.

To my eye, the specklem1 version seems to be a tad more contrasty so the details stand out more, and they are a little more defined out to the rim. I need to do a proper run thru with different settings. There are a myriad and I haven't a clue what they all do or mean, so it will all be trial and error. When I get that figured out, I'll put together a proper LRGB for your inspection. Probably won't be fore a few days though.

So I want anyone's opinions and suggestions on this. Especially if you know the program (Mert, are you watching ;))

Cheers
PS, sorry they are so small... I wanted to get them side by side and had to stay within the image guidelines.

Attached Files



#2 Sunspot

Sunspot

    Fly Me to the Moon

  • *****
  • Posts: 5471
  • Joined: 15 Mar 2005
  • Loc: Surprise, AZ

Posted 28 February 2013 - 07:39 PM

Interesting. I prefer the right image. To me the detail pops versus the left where it is just a bit soft.

Paul

#3 Freddy WILLEMS

Freddy WILLEMS

    Fly Me to the Moon

  • *****
  • Posts: 7228
  • Joined: 13 Nov 2005
  • Loc: Hawaii, Honolulu

Posted 28 February 2013 - 09:47 PM

Interesting, I guess Mert also uses this for processing... Interesting to give it a closer look !
Maybe you can 'add or merge' the two images together with Images Plus ..To get the best of both worlds..


#4 yock1960

yock1960

    Gemini

  • *****
  • Posts: 3081
  • Joined: 22 Jun 2008
  • Loc: (Mostly Cloudy) Ohio, USA

Posted 28 February 2013 - 10:23 PM

I'll give you more contrast on the left side, but I like the sharpness on the right. Both nice though, and as you say Specklem1 is slow....it wouldn't be much of a choice. We'll have to wait for (how long? :)) for Emil to implement it!

Steve

#5 JoseBorrero

JoseBorrero

    Gemini

  • *****
  • Posts: 3137
  • Joined: 04 Sep 2009
  • Loc: Puerto Rico Island

Posted 28 February 2013 - 11:39 PM

Great Image! :waytogo: where I can find an Splecklem1 download link?

#6 johnnyha

johnnyha

    Fly Me to the Moon

  • *****
  • Posts: 6500
  • Joined: 12 Nov 2006
  • Loc: Sherman Oaks, CA

Posted 01 March 2013 - 12:36 AM

Interesting. I prefer the right image. To me the detail pops versus the left where it is just a bit soft.

Clearly.

#7 Tassydevil

Tassydevil

    Viking 1

  • -----
  • Posts: 636
  • Joined: 17 Aug 2011
  • Loc: Tasmania, Australia

Posted 01 March 2013 - 12:45 AM

Well for my eyes the right image wins.

#8 MvZ

MvZ

    Surveyor 1

  • *****
  • Posts: 1577
  • Joined: 03 Apr 2007
  • Loc: The Netherlands

Posted 01 March 2013 - 04:56 AM

> We'll have to wait for (how long? ) for Emil to implement it!

Very long ;)

I never really looked at Splecklem1, but I will try to have a look. About the AS!2 image, how many APs did you use, and where were they placed? It appears the APs weren't optimally placed, because there is some edge-blurring.

#9 tjensen

tjensen

    Mercury-Atlas

  • -----
  • Posts: 2609
  • Joined: 16 Feb 2005
  • Loc: Chapel Hill, NC

Posted 01 March 2013 - 08:33 AM

I agree... the AS!2 image has a better finished look. My point here was to show that I think speckle has some potential. I didn't do any wavelets to the speckled image. That might improve it somewhat. And as I said, there are a lot of settings, and two different speckle applications, that could be tweaked to improve things.

Emil: I used 92x50px alignment points automatically placed.

I'm not sure where to get a download anymore. Maybe Mert does.

#10 MvZ

MvZ

    Surveyor 1

  • *****
  • Posts: 1577
  • Joined: 03 Apr 2007
  • Loc: The Netherlands

Posted 01 March 2013 - 09:27 AM

92 APs is WAY too much for Jupiter. 50 size is probably also way too small. Place +-20 of them manually and you should get much better results. Perhaps then you can add a couple of smaller ones around the GRS, but this is overkill and will produce non-optimal results.

A placement like this should be used:
Posted Image

I know it is difficult to make comparisons in general, but if done incorrectly it leads to false conclusions and basically more uncertainty instead of making things clearer.

The same goes of course for the speckle result. I have no idea what algorithms are underneath that software. Is there really speckle interferometry at the heart of that software??

#11 tjensen

tjensen

    Mercury-Atlas

  • -----
  • Posts: 2609
  • Joined: 16 Feb 2005
  • Loc: Chapel Hill, NC

Posted 01 March 2013 - 10:21 AM

Hey Emil,

Thanks for the tips.

Yes, I agree... there are lots of variables. I do think that the AS result is superior and much faster. I really wanted to start more of a discussion on speckle interferometry.

As to the underlying algorithms for specklem1, I have no idea. The original webpage was in Russian and I've had it lying around for several years. I never really got it to give better results than RegiStax. And that was way before your software hit the scene.

#12 Mert

Mert

    Skylab

  • *****
  • Posts: 4216
  • Joined: 31 Aug 2005
  • Loc: Spain, Pamplona

Posted 02 March 2013 - 01:59 PM

Hey Tim, just came back and saw your thread, very interesting!
I think that the writer of the program Christoph Stelzer wouldn't have problems me sharing the program with others wanting to try things out.
His download link has disapeared a long time ago!

I will upload the program to my web-page in the download
section so you al can download if needed.
For me Specklm runs fairly fast since I use a I7 with 8Gb
of ram and a SSD.
Nowadays with the power of AS!2 I've become lazy and don't
process much with Speklm lately but have been busy with
some other parameters that seem to have a very interesting
effect (Sinlge Photon Brightness ) this seems to take care
of "fringes" in tiff iages I had.

#13 Mert

Mert

    Skylab

  • *****
  • Posts: 4216
  • Joined: 31 Aug 2005
  • Loc: Spain, Pamplona

Posted 02 March 2013 - 02:37 PM

Ok, for those of you who would like to download the
installable file, it is aprox. 9Mb size so not soo big.
This is the latest version Mr. Stelzer updated with some
fixes for TIF files.
You can download from my webpage in the download
section, use the local download link, the other is not
active anymore ( but still there for historical reasons )
If there is interest I can share some more information
on Speckl interferometry ( some pdf's )
Just let me know.
Regards,






Cloudy Nights LLC
Cloudy Nights Sponsor: Astronomics