Jump to content


Photo

Learning PixInsight

  • Please log in to reply
295 replies to this topic

#251 pfile

pfile

    Gemini

  • -----
  • Posts: 3168
  • Joined: 14 Jun 2009

Posted 17 December 2013 - 07:30 PM

that's good because i was going to use my DSLR for comet images... that is until i discovered i could see neither ison nor lovejoy from my observing location :(

there is certainly more than one way to skin a cat; masking the comet could work just as well.

#252 mmalik

mmalik

    DSLR camera modifications

  • *****
  • Vendors
  • Posts: 5411
  • Joined: 13 Jan 2012
  • Loc: USA

Posted 18 December 2013 - 04:54 AM

5) create a star mask; with star mask applied to comet-aligned image, copy stars to comet aligned image with pixel math.


You lost me on #5; do I create star mask from the star integrated image? Which I did. Then I tried applying that mask to comet integrated image which made the image all red, though I could see the star mask getting applied. And to use your words, "copy stars to comet aligned image with pixel math", how do I do that exactly. As you an see, I am totally lost on your #5; your help with actual clicks, may be screen-shots, will be appreciated.

Also, what algorithm did you mean by aggressive pixel rejection in #3 and #4; is 'sigma clipping' ok for such rejection?

On a side note, I am NOT getting any color in the integrated images using PixInsight; I WAS able to get color with ImagesPlus integration of the same data (I mean after proper calibration in each of the software packages). Regards

#253 CharlesW

CharlesW

    Apollo

  • -----
  • Posts: 1331
  • Joined: 01 Nov 2012
  • Loc: Chula Vista & Indio, CA

Posted 18 December 2013 - 09:54 AM

I'm sure you all know this but PI just released an update that has to be accessed from the Software Distribution link on their home page. It requires deleting your current version and installing the new one. Pretty seamless install will no authentication required.

#254 pfile

pfile

    Gemini

  • -----
  • Posts: 3168
  • Joined: 14 Jun 2009

Posted 18 December 2013 - 11:59 AM

well the most common "no color" problem is forgetting to debayer the subs. even though the subs are all 'checker boarded' StarAlignment can sometimes still align them and so your integrated stack is truly a monochrome image.

the 2nd most common "no color" problem is simply not knowing that you have to increase the saturation, but you know that.

on step 5, create a star mask from the star-aligned image as you have done. you may need to tweak the parameters to really get all the stars exposed and perhaps tighten up the mask (by decreasing the growth and compensation parameters). then when the mask is applied what you described is correct; the red areas are masked and the non-red areas are exposed.

then if you put the name of the comet-aligned image into pixel math (the RGB/K field), turn off rescaling, and then apply the pixel math expression to the masked image, the stars should get copied over. if you did not use the mask the entire star-aligned image would be replaced with the comet-aligned image, but the mask prevents that and only the pixels that are revealed by the mask are going to get copied.

agressive pixel rejection means sigma clip, windsor sigma clip or linear fit with the "sigma low" and "sigma high" sliders set to low values. pixels greater than mean+sigma_high and pixels less than mean-sigma_low are rejected, so smaller values reject more pixels. you are cutting off more of the histogram by moving the sigmas closer to the mean.

rob

#255 mmalik

mmalik

    DSLR camera modifications

  • *****
  • Vendors
  • Posts: 5411
  • Joined: 13 Jan 2012
  • Loc: USA

Posted 19 December 2013 - 04:43 AM

then if you put the name of the comet-aligned image into pixel math (the RGB/K field), turn off rescaling, and then apply the pixel math expression to the masked image, the stars should get copied over.


I am not getting the pixel math part at all. This is what I have done so far:

1. Opened star aligned file (called integrationStar.tif)
2. Opened comet aligned file (called integrationComet.tif)
3. Opened StarMask and dragged little triangle of StarMask to star aligned file (integrationStar.tif); it created a mask of star aligned file (called star_mask)
4. Dragged tab of the mask (star-mask) on to integrationComet.tif; this applied mask to comet aligned file which now turns red (integrationComet.tif)
5. Now I open PixelMath, and put integrationComet in the RGB/K field and click apply (little square at bottom) on PixelMath; it seems to run successfully but integrationComet.tif remains red. What next, I am totally lost!

In short, I am not getting your "the stars should get copied over" part.

#256 pfile

pfile

    Gemini

  • -----
  • Posts: 3168
  • Joined: 14 Jun 2009

Posted 19 December 2013 - 12:58 PM

whoops! sorry, i think i said it all backwards.

mask the comet aligned image (the one with no stars) with the starmask. then put the name of the star-aligned image into pixelmath and apply it to the masked comet-aligned image.

also you can turn the mask display on and off even while the mask is active just so you can see what's happening to your image.

rob

#257 mmalik

mmalik

    DSLR camera modifications

  • *****
  • Vendors
  • Posts: 5411
  • Joined: 13 Jan 2012
  • Loc: USA

Posted 20 December 2013 - 04:27 AM

Rob, sorry, but you are not making any sense, at least not to me. My intention was to document this if I were able to execute all steps correctly and in proper sequence. What I would like, if you are willing, you describing in click-level, drag-what-where format, etc., from start to finish, kind of like I have tried to describe in my previous post.

If yes, please start with registering images and go all way to the final result.


On a side note, I am not so sure about what you said earlier, "use the CometAlign process to re-align the star registered images to the comet"; I have done that and I don't see any difference between ImageIntegration result of StarAlignment and CometAlignment processes [i.e., before doing any masking or pixle math-ing, etc.] I am not questioning the logic of what your are saying, but I am questioning the results I am getting which look no different whether I star align or comet align. This is the reason I say we trace our steps all the way back and start from scratch. Regards

FYI: I am using Kevin's Lovejoy data that he shared in this... thread.

#258 harry page 1

harry page 1

    Mariner 2

  • -----
  • Posts: 209
  • Joined: 25 Jul 2009

Posted 20 December 2013 - 12:05 PM

Hi
I will see If someone will lend me some data and I will do a vid as its a common question

Harry

#259 mmalik

mmalik

    DSLR camera modifications

  • *****
  • Vendors
  • Posts: 5411
  • Joined: 13 Jan 2012
  • Loc: USA

Posted 20 December 2013 - 12:48 PM

That will be great Harry; you may use this... data on Lovejoy from Kevin in this... thread. Look forward to your video tutorial on this as it has been quite a roadblock for me, from learning perspective I mean. Regards and thanks in advance!


Note: Harry, Kevin's data is quite good and I have verified the star align vs. comet align in ImagesPlus and can clearly see the difference. I am just not able to do that in PixInsight yet 1) at the align/integration stage (where I could see discernible difference between star aligned and comet aligned version) and 2) definitely need help with high order masking and pixel mathing routines to produce the final image.

#260 dmilligan

dmilligan

    Explorer 1

  • -----
  • Posts: 60
  • Joined: 21 Nov 2013

Posted 20 December 2013 - 03:25 PM

On a side note, I am not so sure about what you said earlier, "use the CometAlign process to re-align the star registered images to the comet"; I have done that and I don't see any difference between ImageIntegration result of StarAlignment and CometAlignment processes [i.e., before doing any masking or pixle math-ing, etc.] I am not questioning the logic of what your are saying, but I am questioning the results I am getting which look no different whether I star align or comet align. This is the reason I say we trace our steps all the way back and start from scratch. Regards


Comet Alignment is not an automatic process. You have to click on the location of the comet in the first and last images, or enter the X,Y coordinates manually (or enter the dX,dY values manually). After you have added the files, under the parameters section, click the first 'show' button. The first image will appear. Click on the location of the comet (there will be a green circle around where you clicked). Then do this again for the other 'show' button (for the last image).

#261 mmalik

mmalik

    DSLR camera modifications

  • *****
  • Vendors
  • Posts: 5411
  • Joined: 13 Jan 2012
  • Loc: USA

Posted 21 December 2013 - 04:50 AM

Comet Alignment is not an automatic process. You have to click on the location of the comet in the first and last images, or enter the X,Y coordinates manually (or enter the dX,dY values manually). After you have added the files, under the parameters section, click the first 'show' button. The first image will appear. Click on the location of the comet (there will be a green circle around where you clicked). Then do this again for the other 'show' button (for the last image).


dmilligan, big thanks! You had alluded to this in the other thread as well but this somehow got lost in translation. This was the disconnect and now most of the steps are starting to make sense and are seeming to be working.

Rob, I think I got it now; thanks for all the help. As I said above, problem boiled down to I not specifying the location of the comet when running CometAlign (presumption on my part that process will auto detect comet in the image); rest of the steps you laid out, do seem to work and make sense now!

I have posted a very preliminary version of my go at Kevin's data here... (I still need to refine my masking/pixel math-ing skills in PixInsight; work in progress...)

#262 mmalik

mmalik

    DSLR camera modifications

  • *****
  • Vendors
  • Posts: 5411
  • Joined: 13 Jan 2012
  • Loc: USA

Posted 21 December 2013 - 05:40 AM

Everyone, here is an overview of the comet processing thus far; comet data courtesy of Kevin...!

Comet Processing:
1. Batch Preprocessing (Calibration)
2. Batch Debayer
3. StarAlignment (Registration)
4. ImageIntegration (Combine/Stack) of StarAlignment data using Sigma Clipping [to remove comet]
5. CometAlignment using StarAlignment (Registration) data [specify comet in first & last image; see pic...]
6. ImageIntegration (Combine/Stack) of CometAlignment data using Sigma Clipping [to remove stars]
7. Create StarMask of ImageIntegration (StarAlignment data)
8. Apply mask to ImageIntegration (CometAlignment data)
9. PixelMath of ImageIntegration (StarAlignment data) [RGB/K: field], and apply to masked ImageIntegration (CometAlignment data) [to transfer stars from Star to Comet integration]
10. Remove mask from ImageIntegration (CometAlignment data) [Need confirmation of this step??]
11. Process image to finish per your routine...

Attached Files



#263 Starhawk

Starhawk

    Space Ranger

  • *****
  • Posts: 5602
  • Joined: 16 Sep 2008
  • Loc: Tucson, Arizona

Posted 21 December 2013 - 08:19 AM

Ah, so this must be what transformed the final presentation I saw at SWAP this year from being a discussion about pixinsight into a full featured rant about how no one should trust the internet or anything they read on cloudy nights (I was shocked at his turn, since I hadn't run across anything about it in CN at that time). That was combined with a full discussion of Photoshop vs. pixinsight, which was equally unilluminating for anything I hoped to learn by attending the lecture. As the time expired, I found I was still wondering what and all pixinsight can do. Apparently one of the hazards of attending a paid astrophotography conference is you're just a captive audience.

Luckily, there is information one can find on CN. And since the audience doesn't feel like they somehow have to stay to either get their money's worth of teach themselves a lesson, rants are poorly attended.

Thank you for posting this.

-Rich

#264 pfile

pfile

    Gemini

  • -----
  • Posts: 3168
  • Joined: 14 Jun 2009

Posted 21 December 2013 - 04:22 PM

warhen got his rant on?

i guess sometimes that happens. likely he's sensitive to this because in truth there is a lot of misinformation around about pixinsight. i have heard some people say some crazy stuff, even in person.

glad to hear you got it sorted out, mike.

also i noticed while working on this that the pentax bayer format is apparently BGGR... unless for some reason my fits reader is set up backwards. that can happen, in which case the bayer pattern gets reversed.

also one thing that can help the comet-aligned integration is to integrate every other or maybe every 3rd image and then integrate those stacks. if the comet did not move enough between consecutive subframes it can difficult to reject the stars since they overlap too much.

rob

#265 harry page 1

harry page 1

    Mariner 2

  • -----
  • Posts: 209
  • Joined: 25 Jul 2009

Posted 23 December 2013 - 02:14 PM

Hi
Here is a little vid using kevins Data ( I have asked but he has not got back to me , If He is unhappy I will pull it)
http://www.harrysast....com/comet.html
Happy Christmas all
Harry Page

#266 mmalik

mmalik

    DSLR camera modifications

  • *****
  • Vendors
  • Posts: 5411
  • Joined: 13 Jan 2012
  • Loc: USA

Posted 23 December 2013 - 08:19 PM

Thanks Harry for the video; much appreciated. I noticed you used star aligned files for ImageIntegrarion step (around 3:35) in the video; I think you meant to use comet aligned (ca) files for that step, correct me if I am wrong? Regards

#267 harry page 1

harry page 1

    Mariner 2

  • -----
  • Posts: 209
  • Joined: 25 Jul 2009

Posted 23 December 2013 - 08:58 PM

Hi
Nope made a mistake there I will add a comment on video to make sure people know what to do correctly
Cheers
Harry

#268 mmalik

mmalik

    DSLR camera modifications

  • *****
  • Vendors
  • Posts: 5411
  • Joined: 13 Jan 2012
  • Loc: USA

Posted 26 December 2013 - 04:05 AM

Thanks Harry for fixing the video; looks good now!

Everyone, I have also added instructions for the same to 'Appendix R' of this... document. Regards

#269 wboeck

wboeck

    Sputnik

  • *****
  • Posts: 32
  • Joined: 01 Dec 2004
  • Loc: Nebraska

Posted 29 December 2013 - 11:33 AM

Harry,
I have been following your videos on Pixinsight and they have made my images look so much better. I really appreciate you providing these for us to use. I would be completely lost without them.

I am currently trying to process my ISON DSLR raw images but am having problems with the batch preprocessing script. It is telling me that it "Cannot integrate less than three frames" I am using eight raw images with flats, darks and bias. It keeps giving me the same error. I attached a screen image with the error. Any help would be greatly appreciated.

Wayne

Attached Files



#270 harry page 1

harry page 1

    Mariner 2

  • -----
  • Posts: 209
  • Joined: 25 Jul 2009

Posted 29 December 2013 - 11:51 AM

Hi
Pi can not find any stars in 7 of your images and therefore ignores them ! this is usually to poor signal or bad focus etc. I can not tell without seeing them , can you send them to me
Harry

#271 wboeck

wboeck

    Sputnik

  • *****
  • Posts: 32
  • Joined: 01 Dec 2004
  • Loc: Nebraska

Posted 29 December 2013 - 11:56 AM

Harry,
yes, I can send them to you. What is the best method? They are approximately 12.4 mb each.

thanks,
Wayne

#272 harry page 1

harry page 1

    Mariner 2

  • -----
  • Posts: 209
  • Joined: 25 Jul 2009

Posted 29 December 2013 - 12:03 PM

Hi send me your email address and I will you details on how to upload to my FTP server

Harry

#273 wboeck

wboeck

    Sputnik

  • *****
  • Posts: 32
  • Joined: 01 Dec 2004
  • Loc: Nebraska

Posted 29 December 2013 - 12:44 PM

Harry
just pm'd you my email address.

thanks for your help

Wayne

#274 mmalik

mmalik

    DSLR camera modifications

  • *****
  • Vendors
  • Posts: 5411
  • Joined: 13 Jan 2012
  • Loc: USA

Posted 02 January 2014 - 01:47 PM

I have been curious and a fellow CNer prompted this question as well; how would one go about processing planetary or lunar images in PixInsight? And most of those come in video format to begin with. Regards

#275 pfile

pfile

    Gemini

  • -----
  • Posts: 3168
  • Joined: 14 Jun 2009

Posted 02 January 2014 - 03:51 PM

i think the support for stacking planetary images is not there. the closest that exists is the script called FFTAlignment. i have used that to align low-resolution solar images for creating animations, but it of course lacks the "decomposition" that registax does to align small tiles very accurately.

of course you can post-process a registax-stacked planetary image in pixinsight using the Deconvolution process and other sharpening methods. there is a tutorial on the pixinsight website which deals with deconvolution of a high-resolution lunar image.

rob






Cloudy Nights LLC
Cloudy Nights Sponsor: Astronomics