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

LP filter vs gradient removal software for galaxies

Filters Imaging
  • Please log in to reply
15 replies to this topic

#1 jhart

jhart

    Sputnik

  • *****
  • topic starter
  • Posts: 34
  • Joined: 16 May 2020
  • Loc: Austin, Texas

Posted 13 May 2021 - 09:25 PM

With the advancement in software processing to remove light pollution and other gradients from stacked images, are light pollution or other broad band filters necessary or even useful?  Specifically I am asking in the context of imaging of galaxies and generally with the use of OSC CMOS cameras.



#2 bobzeq25

bobzeq25

    ISS

  • *****
  • Posts: 24,641
  • Joined: 27 Oct 2014

Posted 13 May 2021 - 10:24 PM

People differ.  For galaxies, I personally don't use any LP filter stronger than a Baader neodymium, which passes the vast majority of light, dropping out mostly where sodium vapor lights emit.  

 

https://39.cdn.ekm.n...59-973F2D1EDEB8

 

Others use none.  Still others use more powerful filters.

 

But gradient reduction in processing is fundamental, most all even semi serious imagers use it.  The question is, will adding an LP filter provide additional improvement?  I think not (the effect of the Baader neodymium is subtle, but at least it does little harm, also).  Given the variation among imagers there's little way to be sure unless you try it.    I have.  A CLS gathers dust on my shelf.  <smile>


Edited by bobzeq25, 13 May 2021 - 10:26 PM.


#3 fewayne

fewayne

    Apollo

  • *****
  • Posts: 1,338
  • Joined: 10 Sep 2017
  • Loc: Madison, WI, USA

Posted 13 May 2021 - 10:51 PM

Certainly the light pollution/gradient removal in APP is easy to use and amazingly effective, and I've no doubt that PixInsight does at least that good a job. I'm kinda with Bob, collect all the photons and let software sort it out.


  • bobzeq25 likes this

#4 sbharrat

sbharrat

    Viking 1

  • -----
  • Posts: 622
  • Joined: 28 Nov 2020
  • Loc: NJ, USA

Posted 14 May 2021 - 06:44 AM

Certainly the light pollution/gradient removal in APP is easy to use and amazingly effective, and I've no doubt that PixInsight does at least that good a job. I'm kinda with Bob, collect all the photons and let software sort it out.

I guess I just don't know how to use the tools properly. I struggle with completely removing the LP gradient with no filter at all. I am not a total neophyte: I have gone through all the Adam Block videos on DBE and ABE and I regularly use those processes in my workflow. Not sure what it is. I am in Bortle 8 and limited visibility limits me to about 2h / night for a target so my images are over many nights. Maybe the LP is different enough over that many nights that makes this harder??

 

I used to use an L-Pro but when I stopped doing that (based on galaxy imaging advice here), I just can't get the background clean. I added back the Baader Neodymium and now that gets me a pretty even background after DBE (hopefully without losing too much signal as the neo looks to be pretty low touch). 


  • bobzeq25 likes this

#5 bobzeq25

bobzeq25

    ISS

  • *****
  • Posts: 24,641
  • Joined: 27 Oct 2014

Posted 14 May 2021 - 08:50 AM

I guess I just don't know how to use the tools properly. I struggle with completely removing the LP gradient with no filter at all. I am not a total neophyte: I have gone through all the Adam Block videos on DBE and ABE and I regularly use those processes in my workflow. Not sure what it is. I am in Bortle 8 and limited visibility limits me to about 2h / night for a target so my images are over many nights. Maybe the LP is different enough over that many nights that makes this harder??

 

I used to use an L-Pro but when I stopped doing that (based on galaxy imaging advice here), I just can't get the background clean. I added back the Baader Neodymium and now that gets me a pretty even background after DBE (hopefully without losing too much signal as the neo looks to be pretty low touch). 

You can never remove light pollution entirely.  At the very best (theoretically), 90%.  I settle for removing most of it.

 

Dark skies are always best.  Hands down.

 

Imaging emission nebulae with a proper camera and narrowband or pseudo narrowband filters can come amazingly close.

 

Galaxies in light polluted skies are hard.  You can get pretty decent images of them, though.  A variety of things helps.  Good software.  Good processing skills.  Mono camera with LRGB filters.   Fast optics, long total imaging times.  Bigger brighter ones, or groups.


Edited by bobzeq25, 14 May 2021 - 08:51 AM.


#6 jonnybravo0311

jonnybravo0311

    Surveyor 1

  • *****
  • Posts: 1,728
  • Joined: 05 Nov 2020
  • Loc: NJ, US

Posted 14 May 2021 - 10:19 AM

I guess I just don't know how to use the tools properly. I struggle with completely removing the LP gradient with no filter at all. I am not a total neophyte: I have gone through all the Adam Block videos on DBE and ABE and I regularly use those processes in my workflow. Not sure what it is. I am in Bortle 8 and limited visibility limits me to about 2h / night for a target so my images are over many nights. Maybe the LP is different enough over that many nights that makes this harder??

 

I used to use an L-Pro but when I stopped doing that (based on galaxy imaging advice here), I just can't get the background clean. I added back the Baader Neodymium and now that gets me a pretty even background after DBE (hopefully without losing too much signal as the neo looks to be pretty low touch). 

I wonder if using the LocalNormalization process would help flatten out your multi-night subs with differing LP gradients, and thus allow DBE to better manage them... might be worth checking out?


  • sbharrat likes this

#7 larryjh

larryjh

    Vostok 1

  • *****
  • Posts: 182
  • Joined: 06 Apr 2016
  • Loc: Marriottsville, MD, USA

Posted 14 May 2021 - 10:29 AM

Siril has a gradient removal tool (they call it background extraction) that can work on a per-sub basis, i.e., compute and remove a different gradient on each sub.  I don't have any experience with PI, so maybe their process does that too, but, if not, maybe worth a try in Siril if you think your subs have different gradients over multiple nights.

 

Edit:  I should have included this link to discussion regarding Siril background extraction: https://www.cloudyni...sense/?hl=siril


Edited by larryjh, 14 May 2021 - 10:47 AM.

  • jonnybravo0311 likes this

#8 jonnybravo0311

jonnybravo0311

    Surveyor 1

  • *****
  • Posts: 1,728
  • Joined: 05 Nov 2020
  • Loc: NJ, US

Posted 14 May 2021 - 10:41 AM

Siril has a gradient removal tool (they call it background extraction) that can work on a per-sub basis, i.e., compute and remove a different gradient on each sub.  I don't have any experience with PI, so maybe their process does that too, but, if not, maybe worth a try in Siril if you think your subs have different gradients over multiple nights.

I actually wrote a SiriL script that incorporates the per-sub background extraction. It's the standard OSC_Preprocessing script with the addition. You can find it on my Google Drive if you're interested: https://drive.google...iew?usp=sharing

 

Only reason I talked about PI and LocalNormalization is because that's the tool Shaun uses to do his image processing as far as I know. I'm not 100% sure if LN will work (or, if the per-sub extraction in SiriL will either), but it might be something worth exploring to help deal with the gradients.

 

Of course, if the Neo filter is helping and providing satisfactory results... then as my dear old daddy used to say, "Son, if it ain't broke, don't fix it."


  • larryjh, limeyx and sbharrat like this

#9 klaussius

klaussius

    Apollo

  • *****
  • Posts: 1,193
  • Joined: 02 Apr 2019
  • Loc: Buenos Aires

Posted 14 May 2021 - 12:15 PM

I guess I just don't know how to use the tools properly. I struggle with completely removing the LP gradient with no filter at all. I am not a total neophyte: I have gone through all the Adam Block videos on DBE and ABE and I regularly use those processes in my workflow. Not sure what it is. I am in Bortle 8 and limited visibility limits me to about 2h / night for a target so my images are over many nights. Maybe the LP is different enough over that many nights that makes this harder??

 

I used to use an L-Pro but when I stopped doing that (based on galaxy imaging advice here), I just can't get the background clean. I added back the Baader Neodymium and now that gets me a pretty even background after DBE (hopefully without losing too much signal as the neo looks to be pretty low touch). 

 

I'm in a similar situation. I'm in Bortle 9, BTW, with leds all around.

 

I've tried no filter, and now I'm trying with an Orion 5561 (it's like an IDAS LPS D2). It feels like the filter helped a lot. Before using the filter, I had horrible green and blue data, requiring a ton of RGB time to get decent color data. With the Orion 5561, however, I get a better SNR match between the filters, and that helps a lot during processing.

 

I think the experience of those in B8/9 is considerably different from those in B6. In B6 skies, it may be possible to get decent results with no filter. In B8/9, however, the skyglow makes everything during processing harder. A non-exhaustive list:

 

  • Substantially more noise, obviously. You'll need far more integration time to get a decent image of dim galaxy.
  • Limited exposure time forces you to stack hundreds of subs. In my skies, for example, I'm limited to 30s luminance subs if I don't want to saturate stars, maybe 60s if I don't care about stars. That results in easily 300+ subs during stacking, taking conisderable effort.
  • Flats have to be perfect, no room for mistakes. The intense skyglow will make any flat miscorrection appear as a very difficult to deal gradient. That doesn't happen in darker skies, nor with narrowband. Vignetting or dust motes under dark skies are far less troublesome, but with a bright sky, they're a pain to deal with.
  • B8/9 means a city, so that usually comes with limited sky time, which makes racking up integration time quite laborious. For example, I'm usually limited to 3h per night, depending on the target's path (most of my east side is blocked).

So, for a B6 imager, maybe going no filter and racking up integration time is a viable strategy. But I can't say I feel like I'm able to under B9... and using filters helps a bit, and I'll take it, drawbacks included.

 

It's good to be aware of the drawbacks, though, and to not go for the extreme filters that are simply not a good fit for galaxies and broadband targets in general.


  • jhart and sbharrat like this

#10 jhart

jhart

    Sputnik

  • *****
  • topic starter
  • Posts: 34
  • Joined: 16 May 2020
  • Loc: Austin, Texas

Posted 14 May 2021 - 01:48 PM

Some very good and detailed observations.  Thanks.

 

Based on some comments, particularly from klassius, it appears that Bortle level and perhaps focal ratio are important factors to consider.  From my home where I mostly image I have ~9 B skies.  When targeting galaxies I use a slow ~10 focal ratio with my SCT.  Is the loss of photons in certain ranges of the spectrum with a filter in those conditions worth it?

 

It appears that the Orion SkyGlow 5561 and Baader Neodymium Skyglow are the mildest (i.e. lesser blocking) with slightly different transmissivity (other than the Baader looks like it blocks more IR).  The Optolong L-Pro and IDAS LPS-P2 seemed to be very similar and more aggressive.  What would be the significant pluses and minuses among these filters, again for galaxies, in B9 skies and f10?

 

I have not gotten into PixInsight and rely on the APP light pollution tool or StarTools wipe mainly.  My skills are not great but I can usually get fairly dark and gradient free background -- let me re-phrase and say sometimes rather than usually -- without a filter.  I cannot say what degradation that causes to the desirable parts of the image as I have no apples-to-apples comparison.  Does filter vs no filter create other post-processing headaches (again in the context of my type of set up for galaxies)?

 

Thanks again,

Jeff



#11 ngc1535

ngc1535

    Vendor - Caelum Observatory

  • -----
  • Vendors
  • Posts: 123
  • Joined: 18 Sep 2008

Posted 14 May 2021 - 04:01 PM

Shaun,

 

I wonder if you really have gone through *all* of my material. :)

In particular, you might be interested in one of the recent workflow examples (the Tau Canis Majoris region) in which I highlight the use of the NormalizeScaleGradient script by John Murphy.

Have you had a chance to review this?

 

It is a new thing that I think is better than Local Normalization- at least in my limited experimentation with it. I was really impressed.

I will create a vid on the details of its usage eventually- probably after the release of the PI update.

-adam


  • jhart, jonnybravo0311 and sbharrat like this

#12 jhart

jhart

    Sputnik

  • *****
  • topic starter
  • Posts: 34
  • Joined: 16 May 2020
  • Loc: Austin, Texas

Posted 14 May 2021 - 04:24 PM

Hi adam,

 

Thanks for the suggestion.  To answer your question I have not looked at that script.  Actually I looked at PI a while ago, maybe I downloaded the trial version I can't remember, and kind of got freaked out, thinking it was way above my pay grade.  I have learned a few things since then about imaging and maybe I should try again.  Especially if the LP/gradient removal abilities, including that new script, might obviate the need to buy more filters, it could make the cost of PI an economic investment simply in this context.  The time investment to learn it is another issue.

 

Do you successfully use PI to remove LP and gradients without the need for any sort of LP filter?  Same question for Bortle 9 skies?

 

Jeff



#13 bobzeq25

bobzeq25

    ISS

  • *****
  • Posts: 24,641
  • Joined: 27 Oct 2014

Posted 14 May 2021 - 05:40 PM

Hi adam,

 

Thanks for the suggestion.  To answer your question I have not looked at that script.  Actually I looked at PI a while ago, maybe I downloaded the trial version I can't remember, and kind of got freaked out, thinking it was way above my pay grade.  I have learned a few things since then about imaging and maybe I should try again.  Especially if the LP/gradient removal abilities, including that new script, might obviate the need to buy more filters, it could make the cost of PI an economic investment simply in this context.  The time investment to learn it is another issue.

 

Do you successfully use PI to remove LP and gradients without the need for any sort of LP filter?  Same question for Bortle 9 skies?

 

Jeff

I'm Bortle 7.  Look at my astrobin.  Everything except the emission nebulae was done with little or no filtration. 

 

The emission nebulae mostly used filters.  Some narrowband, some duoband with a one shot color camera.

 

Astro Pixel Processor gives you excellent gradient reduction in easy to use software.  It also has a workflow that teaches you how to process.  Definitely "worth it", not something I often say in that global context.


Edited by bobzeq25, 15 May 2021 - 10:13 AM.

  • jhart likes this

#14 Ivo Jager

Ivo Jager

    Vendor ( Star Tools )

  • *****
  • Vendors
  • Posts: 468
  • Joined: 19 Mar 2011
  • Loc: Melbourne, Australia

Posted 15 May 2021 - 09:54 PM

With the advancement in software processing to remove light pollution and other gradients from stacked images, are light pollution or other broad band filters necessary or even useful?  Specifically I am asking in the context of imaging of galaxies and generally with the use of OSC CMOS cameras.

Hi,

 

Software-based LP modelling is not a substitute for avoiding the light pollution signal.

The problem is that software can model and subtract the unwanted signal just fine, but not its noise component (as it is obviously random and you cannot "model" randomness). So even after subtraction, the LP's noise component will remain imprinted onto your dataset.

 

Use a filter if you can, or go narrowband. Or shoot luminance with an LP filter and color without an LP filter. Or just put up with the color skew/removal of a LP filter.

 

Hope this helps,


  • Jinux and jhart like this

#15 jhart

jhart

    Sputnik

  • *****
  • topic starter
  • Posts: 34
  • Joined: 16 May 2020
  • Loc: Austin, Texas

Posted 16 May 2021 - 10:46 AM

Hi Ivo,

 

I am a Star Tools user and will follow your advice.

 

Jeff



#16 sbharrat

sbharrat

    Viking 1

  • -----
  • Posts: 622
  • Joined: 28 Nov 2020
  • Loc: NJ, USA

Posted 16 May 2021 - 11:37 PM

Shaun,

 

I wonder if you really have gone through *all* of my material. smile.gif

In particular, you might be interested in one of the recent workflow examples (the Tau Canis Majoris region) in which I highlight the use of the NormalizeScaleGradient script by John Murphy.

Have you had a chance to review this?

 

It is a new thing that I think is better than Local Normalization- at least in my limited experimentation with it. I was really impressed.

I will create a vid on the details of its usage eventually- probably after the release of the PI update.

-adam

Haha. I should be more careful with my wording... Of course I could not have gone through *all* your material!! But I thought I had gone through your DBE videos. Even on this though, no, I did not see what you are referencing (yet). Will definitely look for it. 

 

(Frankly, I was so happy with the results I have got with neodymium on galaxies... less noisy with similar imaging time than L-pro but still able to clean background to my [loose] standards... that I stopped trying to further optimze this part.)

 

Thanks and cheers!


Edited by sbharrat, 16 May 2021 - 11:42 PM.



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





Also tagged with one or more of these keywords: Filters, Imaging



Cloudy Nights LLC
Cloudy Nights Sponsor: Astronomics