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

What are the downsides of using a UBVRI filter set for imaging?

  • Please log in to reply
26 replies to this topic

#1 Tonk

Tonk

    Cosmos

  • *****
  • topic starter
  • Posts: 9,278
  • Joined: 19 Aug 2004
  • Loc: Leeds, UK, 54N

Posted 17 January 2020 - 10:09 AM

What are the downsides of using a UBVRI filter set for imaging? Any good for pretty pictures as well as doing science (on the same rig)?

TIA



#2 Allaboutastro

Allaboutastro

    Mercury-Atlas

  • -----
  • Posts: 2,695
  • Joined: 10 Jan 2010
  • Loc: Grapevine, TX

Posted 17 January 2020 - 12:07 PM

A set of RGB filters like AstroDons will typically square off the bandpasses a little more, permitting more transmission across the broad-bands.   Likewise, they may cut off these spectral bands slightly to block some LP sources.  Really, just google the spectral curves of the imaging filters vs. the photometry filters in question and that should provide you the information you need.  That said, the BVR filters seem to look more like the Bayer matrix filter curves (more overlap), except that BVR doesn't include the extra "toe" in the green and blue filters to give the red some extra information (to better render "violet").   

 

In short, I see no reason why you couldn't take pretty pictures with UBVRI filters and I've often wanted to play with them myself, especially to capture UV and near-IR frequencies just to see what they deliver on various objects. 



#3 555aaa

555aaa

    Vendor (Xerxes Scientific)

  • *****
  • Vendors
  • Posts: 1,818
  • Joined: 09 Aug 2016
  • Loc: Lynnwood, WA, USA

Posted 17 January 2020 - 12:31 PM

The V is (to me anyway) more like a yellow than a green, but you can still get a pleasing result in my opinion. I have some on my astrobin page under brucev but I don't have pixinsight or any decent processing SW.

 

One thing you can do that most people aren't aware of is use the color mixer in Photoshop. The mixer function can act as a colorspace transform. What is really needed is to mathematically define the BVR as a colorspace, and then come up with the mathematical matrixes that transform those into RGB or HSL.  In the channel mixer, what you are doing "artistically" is taking some of one color and putting it into another color. That is basically what happens in a color space transform. I've been meaning to do some testing with color standards for this but not had any time.

 

Once you start putting narrowband in your images you've gone off into the la la land of color accuracy anyway.


  • psandelle likes this

#4 Tonk

Tonk

    Cosmos

  • *****
  • topic starter
  • Posts: 9,278
  • Joined: 19 Aug 2004
  • Loc: Leeds, UK, 54N

Posted 17 January 2020 - 03:07 PM

Cheers, the back story is I have the AstroDon E-Series filters on my remote rig but I'm joining up with 3 other astronomers on that rig (to reduce individual costs). However those guys are university astronomy graduates/researchers who are very interested in having the set changed to UBVRI so they can do proper science projects.

If I go with that suggestion I was just wondering what the compromises are for "pretty pictures". However as you will see in these traces of my filter set - there are specialist filters for comets (bar the H2O+ which is not present). The LRGB are AstroDon E-series

R2VnEL-087hj_620x0_kWXURFLk.jpg

One "issue" I can see from sample UBVRI filter scans is that a lot of flux would be lost due to the slower roll off of these filters compared to the squarer AstroDon traces. Is this a problem?? - presumably it will need slightly longer exposures and as said above - some fiddling with colour balance.



#5 freestar8n

freestar8n

    Vendor - MetaGuide

  • *****
  • Vendors
  • Posts: 9,764
  • Joined: 12 Oct 2007

Posted 17 January 2020 - 04:55 PM

I use Sloan filters I r g and they have advantages of wide spectral width, no gaps, full transmission, and direct match to Apass magnitudes of reference stars. The bands are shifted so that Ha ends up in green so it is somewhat false color when Ha is present but otherwise it isn’t too different. A nice feature is that extended ir allows faint cool stars to show as deep red.

I’m not sure what filters you would use but U and I probably wouldn’t get much signal. So I would use B V R. Or just use Sloan filters as I do.

Many examples on my astrobin Freestar8n.

Frank
  • Tonk and psandelle like this

#6 Tonk

Tonk

    Cosmos

  • *****
  • topic starter
  • Posts: 9,278
  • Joined: 19 Aug 2004
  • Loc: Leeds, UK, 54N

Posted 17 January 2020 - 04:58 PM

Here is the scans of my current filter set  with the AstroDon UBVRcIc set overlaid

BGXzSGhl68_C_620x0_kWXURFLk.jpg



#7 Allaboutastro

Allaboutastro

    Mercury-Atlas

  • -----
  • Posts: 2,695
  • Joined: 10 Jan 2010
  • Loc: Grapevine, TX

Posted 17 January 2020 - 04:58 PM

The main Ha line is typically well covered in the R filter with photometry filters. OIII is typically not (it will be near the B and V crossover).  So, for planetaries, supernovae, WR objects...you'll probably need to go a little longer.   Are they optimal?   No...but they are completely serviceable.   But I always fiddle with color balance, even with my AstroDons and Custom Scientifics...so take it with a grain of salt.  :)



#8 Tonk

Tonk

    Cosmos

  • *****
  • topic starter
  • Posts: 9,278
  • Joined: 19 Aug 2004
  • Loc: Leeds, UK, 54N

Posted 17 January 2020 - 06:26 PM

This is my filter set with Sloan g' r' and i' overlaid.

Here is a thought - as we are thinking of moving up to a 12 slot 31mm filter wheel - we will have two spare slots if I keep all existing filters including the E-Series LRGB. If I add Sloan r' and i' as the two extra then the E series (B + G)/2 should be a good approximation of Sloan g' (there is a 12nm overlap in B and G to consider). Thus I can accommodate LRGBr'i' and have the means to create approx g' data.  Plus B, G, r', i' all butt up together as as series with no gaps from 400nm to 860nm.

vkWSiA3cz9My_620x0_kWXURFLk.jpg


  • psandelle likes this

#9 freestar8n

freestar8n

    Vendor - MetaGuide

  • *****
  • Vendors
  • Posts: 9,764
  • Joined: 12 Oct 2007

Posted 18 January 2020 - 12:24 AM

As I understand it some applications prefer Johnson cousins to Sloan so it might be good to ask potential researchers if it matters. And if you only use two photometric filters they might have a preference for blue vs red to go with green.

Frank

#10 jdupton

jdupton

    Mercury-Atlas

  • *****
  • Posts: 2,730
  • Joined: 21 Nov 2010
  • Loc: Central Texas, USA

Posted 18 January 2020 - 07:43 AM

Tonk,

 

   As Frank indicate just above, the real difference may lie in what the researchers are wanting to do with the data. I have helped work on several real science research projects with one of the professors at the local university. The projects I have gathered data for dictated the use of Johnson-Cousins filters. In the cases I helped with, the researchers mostly only needed V and B filter data as the data was being paired with data from HST, Spitzer, and Chandra gathered during the same data campaign. In one project, they also wanted Johnson-Cousins R data. This ground based data from up to a dozen observatories was used to help "fill in the gaps" in the high cadence (every one to three nights) monitoring of the target when space telescope time was not available.

 

   My impression from the professor is that many researches are now starting to switch to Sloan filters as the preferred set. (The change seems to be happening at an almost glacial pace.) To some extent, it just depends on what the lead researchers want or accustomed to and which other observatories / space telescopes they are using for the rest of the data, if any.

 

   Using that Johnson-Cousins data for regular images is pretty easy. I have have done a few BVR combinations of the research data to illustrate the target being studied. It is really no different than using RGB data from an OSC camera. David Ault and I both have used some of the data that was only taken with BV filters to build a target image by synthesizing R data using G2 type stars in the field to establish approximate color channel ratios. (David's versions were always far better than mine.)

 

 

John


Edited by jdupton, 18 January 2020 - 07:48 AM.

  • psandelle and Jon Rista like this

#11 Tonk

Tonk

    Cosmos

  • *****
  • topic starter
  • Posts: 9,278
  • Joined: 19 Aug 2004
  • Loc: Leeds, UK, 54N

Posted 18 January 2020 - 08:49 AM

Silly question time.

How possible/hard/easy is it to convert photometry measurements from one filter system to another? I.e. is there a route to/from Johnson from/to Sloan?
 

Has anyone considered creating a photometric system and generating standard data for an astrophotographers filter set with a means to convert to Johnson/Sloan - or is that too impractical/worthless a project?

 

Problem with a remote rig is it aint easy/desirable to change filters (hence loading up a 12 slot wheel and trying to find a compromise for all our intended projects :) ). I can see us now going for Johnson  only to find in 5 years time everyone wants Sloan. Ha!

 

 


Edited by Tonk, 18 January 2020 - 08:50 AM.


#12 JuergenB

JuergenB

    Vostok 1

  • -----
  • Posts: 180
  • Joined: 15 Nov 2015
  • Loc: Lilienthal, Germany

Posted 18 January 2020 - 11:19 AM

Tony,

 

Conversions between the different photometric systems are possible. You will find formulae on the sdss3.org websites. Just google for "Johnson to Sloan conversion", which will lead you to the corresponding sites.

 

Cheers

Juergen


  • Tonk and psandelle like this

#13 freestar8n

freestar8n

    Vendor - MetaGuide

  • *****
  • Vendors
  • Posts: 9,764
  • Joined: 12 Oct 2007

Posted 18 January 2020 - 06:59 PM

There are conversions between filters but they often depend on what the object is that's being measured.  Not only star vs. galaxy, but what type of star.

 

So - you don't even need photometric filters to find magnitudes for a given photometric filter set.  Normal imaging filters will do it as long as you calibrate with field stars - but you will lose accuracy because ultimately you need to know how the different object spectra overlap with the filter bandwidth.  If you use the same filters that were used for the reference measurements then you don't have that problem.

 

As I understand it, the Sloan filters were optimized specifically for measuring galaxies - and as a result they aren't as ideal for measuring stars.  That's why the two filter types are still in use.  You can go back and forth between Sloan and J-C - but the transformation will depend on the type of star and its spectrum.

 

Again - if you know what potential collaborators are interested in you can go by their preference.

 

But two filters is enough to capture the magnitude in the V or r' band - along with color based on B-V or g'-r'.

 

Frank


  • psandelle and JuergenB like this

#14 Tonk

Tonk

    Cosmos

  • *****
  • topic starter
  • Posts: 9,278
  • Joined: 19 Aug 2004
  • Loc: Leeds, UK, 54N

Posted 18 January 2020 - 07:33 PM

Another thought on compromises to get our 12 slot filter wheel loaded and the team happy:

Take out the Astrodon E-Series R filter and replace it with Astrodon Sloan r'. All that does is fill in the 'yellow' light pollution gap that is characteristic of the E-Series set (its now similar the Astrodon I-Series set but with the G/R crossover earlier at 560nm instead of 580nm).

Then with the 2 spare slots that I started out with we add Sloan g' and i' to get probably the most useful 3 filter subset of the Sloan filters. Then for pretty pictures we can use Lr'GB and just work out what the exposure ratio for r'GB should be (wont be 1:1:1 anymore) and for the photometry orientated part of our team they have g', r', i' to play with. Would this be useful enough?

Where the scope is going later in the year there will not be any sodium based light pollution to worry about so the E-Series R -> Sloan r' swap isn't an issue other than the exposure ratios.


Edited by Tonk, 18 January 2020 - 07:53 PM.

  • psandelle likes this

#15 Tonk

Tonk

    Cosmos

  • *****
  • topic starter
  • Posts: 9,278
  • Joined: 19 Aug 2004
  • Loc: Leeds, UK, 54N

Posted 18 January 2020 - 07:39 PM

Not only star vs. galaxy, but what type of star.


Yes I read that on the sdss3.org web site and understand why. I'm not familiar with the astronomy approach to all this but being a PhD chemist I do appreciate the issue of mapping spectra types to specific photometric filter system conversion equations

#16 Narrowbandpaul

Narrowbandpaul

    Mariner 2

  • -----
  • Posts: 236
  • Joined: 24 Dec 2013
  • Loc: Constrained by the Heisenberg uncertainty principle

Posted 19 January 2020 - 05:17 AM

Hi All

I’m one of the ‘collaborators’ if you can call me that 😄. I’m by far and away a ‘pretty picture’ person but I am trying to get more involved with my former university, where I did an undergrad in Astronomy and Physics. Another person is a researcher there.

We have a remote system purely intended for the usual pretty picture imaging but I realised of course that our system, whilst not optimised for it, could do some basic photometry and we can calibrate against standard stars. There is no real project in mind except to have a system that can take reasonably accurate photometric observations. We would like to encourage honours and masters students to come up with their own ideas for investigations, be it from exoplanet detection via the eclipsing method or variable star observations, Cepheid Variables, eclipsing binaries, galactic redshift...etc. Basically a system that can produce reasonable results for students and ourselves whilst not getting in the way of pretty picture imaging. The filters need to be unobtrusive and not be an endless source of pain to make look pretty.

Where we are going is dark and has plenty clear nights so we are probably happy to forego the sodium LP gap that imaging filters have.

The combo of Astrodon B and G then Sloan R’ I’ as Tony says provides complete coverage from 400-860nm and has the Ha in the Red channel, which is more conventional, although Franks M8 with the green hue was rather nice.

The nuclear option is to try to create a new system. Realising that the UBVRI is not ideal with the sloped response of the filters and that GRI in Sloan has Ha in green we could use the AD E-Series B and G with the Sloan R and I to define a new standard more accessible for amateurs. The dilemma comes when someone with imaging filters wants to try photometry and has to buy a new set only to realise that there is a serious compromise for pretty pictures. With these 4 you should get the best of both worlds

Plus having 2 proper photometric filters consecutively means we can form the colour R’-I’ and can use colour transforms to calibrate those measurements

I appreciate that no system is perfect and for stars I guess you probably want more intermediate band filters to highlight different absorption lines, but hopefully this proposal is still capable of useful photometry of most objects.

Thanks for all the advice guys

Paul
  • psandelle likes this

#17 leviathan

leviathan

    Viking 1

  • *****
  • Posts: 797
  • Joined: 29 Nov 2011
  • Loc: Azerbaijan

Posted 19 January 2020 - 08:12 AM

Tonk, just get a bigger filter wheel and use both types of filters there.



#18 Narrowbandpaul

Narrowbandpaul

    Mariner 2

  • -----
  • Posts: 236
  • Joined: 24 Dec 2013
  • Loc: Constrained by the Heisenberg uncertainty principle

Posted 19 January 2020 - 08:43 AM

Bigger than 12 slots?

#19 555aaa

555aaa

    Vendor (Xerxes Scientific)

  • *****
  • Vendors
  • Posts: 1,818
  • Joined: 09 Aug 2016
  • Loc: Lynnwood, WA, USA

Posted 19 January 2020 - 11:32 AM

I would make sure that you have a pair of filters that's in the most common color index for either the Sloan or Johnson cousins. For the latter it is the B and V filters with B-V being the color index. I would probably stick with B V R, a clear, G for pretty pictures, and then fill the other slots with narrowband. I have never had very good photometry trying to make a synthetic V from say DSLR RGB colors. What I mean is that when I shoot a standard calibration field from the Landolt catalog, there is a lot more scatter in the results than using the proper filters. And remember that to get an accurate V mag you have to make a B measurement. You measure an instrumental mag and color and you use both to calculate a true mag. I can send you some BVR galaxy fits files to play with if you want. Maybe I will just post them.
  • psandelle likes this

#20 Tonk

Tonk

    Cosmos

  • *****
  • topic starter
  • Posts: 9,278
  • Joined: 19 Aug 2004
  • Loc: Leeds, UK, 54N

Posted 19 January 2020 - 11:39 AM

Tonk, just get a bigger filter wheel and use both types of filters there.

 

Do you know of a wheel with more than 12 slots? Seriously interested!

 

Our current filter set is

Astrodon E-Series L, R, G, B
Astrodon Narrowband: Ha, OIII, SII  
Semrock Narrowband (comets): C2, CO, CN

(the OIII and SII are currently unused  as we use a QSI 8 slot filter wheel)

Thats 10 filters total. Moravian make a 12 slot wheel which we are going to look at. That would give use 2 extra slots over our current filter set. We would like a NIR filter at least. The only photometric one that passes H2O+ (comet ion tails) is the Sloan i'. The JC Ic filter puts H2O+ outside its passband. So to continue loading photometric filters we can take out the E-Series R and replace it with Sloan r'. This leaves one slot of the 12 and the sensible addition is then Sloan g' to give us at least the most useful 3 from the Sloan set.

This would then morph the filter set into:

Astrodon E-Series L, G, B
Astrodon Sloan: g', r', i'

Astrodon Narrowband: Ha, OIII, SII 

Semrock Narrowband (comets): C2, CO, CN

Where r' doubles up as a regular R filter. That's all 12 slots used.

 

+++

 

Another choice would be to ditch the L filter and shoot E-Series R,G,B always in 1x1 binning mode and create a synthetic L data via pixel math (R + G + B).

Then the full set would be:

Astrodon E-Series R, G, B
Astrodon Sloan: g', r', i'

Astrodon Narrowband: Ha, OIII, SII

Semrock Narrowband (comets): C2, CO, CN

However I'm not too much of a fan of Synthetic L


Edited by Tonk, 19 January 2020 - 02:29 PM.


#21 555aaa

555aaa

    Vendor (Xerxes Scientific)

  • *****
  • Vendors
  • Posts: 1,818
  • Joined: 09 Aug 2016
  • Loc: Lynnwood, WA, USA

Posted 19 January 2020 - 12:51 PM

These were taken with my old(er) mechanically geared mount at my windy site so the guiding is less than stellar. It's with the KAI-11002 interline CCD, astrodon photometric filters.

 

http://www.xerxessci...rlick_index.htm

 

Sample fits files in Johnson-Cousins B, V, R + clear + 1 narrowband. The blue is pretty weak, needed a lot more blue but it's what I have handy.

 

The Sloan g and r will be your color index pair in that system with the filters you list above which should be fine. I would not ditch the clear filter because there are so many scientifically interesting transient objects (or moving objects) that you really need every photon.


Edited by 555aaa, 19 January 2020 - 01:03 PM.

  • Tonk and psandelle like this

#22 leviathan

leviathan

    Viking 1

  • *****
  • Posts: 797
  • Joined: 29 Nov 2011
  • Loc: Azerbaijan

Posted 19 January 2020 - 01:50 PM

Tonk, FLI makes 20 (!) position filter wheel, but it's very expensive.

If your setup's back distance allows you can use 2 filter wheels with clear filter in each one and change filters depending on what you need ! tongue2.gif


  • psandelle likes this

#23 Tonk

Tonk

    Cosmos

  • *****
  • topic starter
  • Posts: 9,278
  • Joined: 19 Aug 2004
  • Loc: Leeds, UK, 54N

Posted 19 January 2020 - 02:44 PM

If your setup's back distance allows you can use 2 filter wheels with clear filter in each one and change filters depending on what you need !


I wish! Cant be done in the space I have - and how do you get SGPro to work this? (answer - write your own double FW ASCOM driver)
 

 

FLI makes 20 (!) position filter wheel, but it's very expensive.


I can see pier collisions in my nightmares! .. and as it supports just 28mm filters I'll have to file down the glass of all mine frown.gif.... and that price is unjustified!


Edited by Tonk, 19 January 2020 - 02:47 PM.


#24 555aaa

555aaa

    Vendor (Xerxes Scientific)

  • *****
  • Vendors
  • Posts: 1,818
  • Joined: 09 Aug 2016
  • Loc: Lynnwood, WA, USA

Posted 19 January 2020 - 03:38 PM

You should definitely have a no flip pier. My understanding is that most of these outfits can have one custom fabricated in country. Just my opinion.

#25 freestar8n

freestar8n

    Vendor - MetaGuide

  • *****
  • Vendors
  • Posts: 9,764
  • Joined: 12 Oct 2007

Posted 19 January 2020 - 03:58 PM

I think the option to replace R with r' looks good.  I wouldn't drop L because L or C is important to have.

 

I'm not sure you really need all three photometric filters, so you could probably drop one.

 

Here is a comet image in Sloan i' r' g' from a while ago:

 

https://www.cloudyni...-sloan-filters/

 

get.jpg?insecure

 

I didn't calibrate the colors with foreground stars so the colors are somewhat arbitrary.  But they do map to the Sloan filters.

 

Frank


  • Tonk and psandelle like 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