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

Loss of color diversity in LRGB photography when the filters do not overlap

  • Please log in to reply
292 replies to this topic

#76 loujost

loujost

    Mariner 2

  • -----
  • topic starter
  • Posts: 243
  • Joined: 02 Oct 2017

Posted 12 June 2023 - 05:45 PM

What about these UBVRI photometric filters from Baader? They overlap quite a bit between red/green, but not as much between green/blue. Hb is the only line that seems suppressed. Need violet? Mix in the ultraviolet filter, although I guess the mix may be up to one's taste. Would you end up mixing those by hand or could you make a calibrated mixing ratio based on transmission losses? I would assume you wouldn't want to use an L filter for this kind of filter set.

Those look really interesting!!!! Thanks, I had never heard of them, The red-green peaks and their overlap are very close to that of human cones. As you say, it looks like you could recover violets using the U filter to make up for the lack of red-blue overlap.

 

If only they were a bit less expensive I'd buy them right this minute. These really do look like the ideal solution.


Edited by loujost, 12 June 2023 - 05:46 PM.


#77 Jon Rista

Jon Rista

    ISS

  • *****
  • Posts: 26,034
  • Joined: 10 Jan 2014
  • Loc: Colorado

Posted 13 June 2023 - 10:14 PM

Those look really interesting!!!! Thanks, I had never heard of them, The red-green peaks and their overlap are very close to that of human cones. As you say, it looks like you could recover violets using the U filter to make up for the lack of red-blue overlap.

 

If only they were a bit less expensive I'd buy them right this minute. These really do look like the ideal solution.

You probably have, though? Its the same as Johnson or Cousins filters, right? There may be some slight differences in each of the three when it comes to exact transmission, but generally speaking Bessel, Cousins and Johnson filters are all the same photometric filters. The "V" filter, for Green (Verde) is THE V filter...the same V used with a lot of photometric calibration routines.

 

The Bessel filters, or Johnson-Cousins as they are termed in the actual survey documents and reports, is the same filter set used for the SDSS full sky survey as well. 

 

You HAVE to have heard of these, Lou! :p



#78 italic

italic

    Apollo

  • *****
  • Posts: 1,108
  • Joined: 05 Apr 2022
  • Loc: San Francisco, CA

Posted 13 June 2023 - 10:30 PM

You probably have, though? Its the same as Johnson or Cousins filters, right? There may be some slight differences in each of the three when it comes to exact transmission, but generally speaking Bessel, Cousins and Johnson filters are all the same photometric filters. The "V" filter, for Green (Verde) is THE V filter...the same V used with a lot of photometric calibration routines.

 

The Bessel filters, or Johnson-Cousins as they are termed in the actual survey documents and reports, is the same filter set used for the SDSS full sky survey as well. 

 

You HAVE to have heard of these, Lou! tongue2.gif

I thought SDSS had its own filters, u'g'r'i'z', while Johnson Cousins are the UBVRI set. SDSS filter passes don't overlap very much and the cutoff is very sharp. UBVRI is much slower falloff and they overlap more, namely V/R. Are these the only two standard photometry filter sets?


  • jdupton and CharLakeAstro like this

#79 freestar8n

freestar8n

    MetaGuide

  • *****
  • Freeware Developers
  • Posts: 13,442
  • Joined: 12 Oct 2007
  • Loc: Melbourne, Australia

Posted 14 June 2023 - 12:07 AM

I thought SDSS had its own filters, u'g'r'i'z', while Johnson Cousins are the UBVRI set. SDSS filter passes don't overlap very much and the cutoff is very sharp. UBVRI is much slower falloff and they overlap more, namely V/R. Are these the only two standard photometry filter sets?

Yes - the SDSS survey uses Sloan u'g'r'i'z' filters while Johnson-Cousins are UBVRI.

 

The V is a green filter but it stands for "Visual" because it is a good match to human visual luminance response and is the basis for visual magnitudes of stars.  For similar reasons the Bayer array has doubled green pixels.

 

I have been using Sloan i' r' g' filters as R G B channels for many years because I wanted a wide coverage of the spectrum with no gaps.  But the r' is so wide that the Ha signal shows up in the G channel.  This has benefits because it means you make fuller use of possible colors when imaging galaxies - but it does look unnatural.  I also like that the i' extends out to 800nm because it captures small, cool stars as deep red.

 

Recently I have switched to Johnson-Cousins filters from Chroma.  These are interferometric rather than using colored glass and as a result have sharp, non-overlapping bands with high transmission and are a better match to human response than Sloan.

 

I think it makes sense to use overlapping filters - and it makes no sense to have a gap for light pollution when most people have continuum light pollution.  It just loses perfectly good signal.

 

I don't think there is much point in seeking "true" color somehow - so I aim for ways to capture good color information that is visually pleasing while also interpretable and informative.

 

Frank


  • Tomvictor and CharLakeAstro like this

#80 loujost

loujost

    Mariner 2

  • -----
  • topic starter
  • Posts: 243
  • Joined: 02 Oct 2017

Posted 14 June 2023 - 04:14 AM

You probably have, though? Its the same as Johnson or Cousins filters, right? There may be some slight differences in each of the three when it comes to exact transmission, but generally speaking Bessel, Cousins and Johnson filters are all the same photometric filters. The "V" filter, for Green (Verde) is THE V filter...the same V used with a lot of photometric calibration routines.

 

The Bessel filters, or Johnson-Cousins as they are termed in the actual survey documents and reports, is the same filter set used for the SDSS full sky survey as well. 

 

You HAVE to have heard of these, Lou! tongue2.gif

Yes, I had known of the photometric calibration filters but did not know that Baader sells these.

 

Yes - the SDSS survey uses Sloan u'g'r'i'z' filters while Johnson-Cousins are UBVRI.

 

The V is a green filter but it stands for "Visual" because it is a good match to human visual luminance response and is the basis for visual magnitudes of stars.  For similar reasons the Bayer array has doubled green pixels.

 

I have been using Sloan i' r' g' filters as R G B channels for many years because I wanted a wide coverage of the spectrum with no gaps.  But the r' is so wide that the Ha signal shows up in the G channel.  This has benefits because it means you make fuller use of possible colors when imaging galaxies - but it does look unnatural.  I also like that the i' extends out to 800nm because it captures small, cool stars as deep red.

 

Recently I have switched to Johnson-Cousins filters from Chroma.  These are interferometric rather than using colored glass and as a result have sharp, non-overlapping bands with high transmission and are a better match to human response than Sloan.

 

I think it makes sense to use overlapping filters - and it makes no sense to have a gap for light pollution when most people have continuum light pollution.  It just loses perfectly good signal.

 

I don't think there is much point in seeking "true" color somehow - so I aim for ways to capture good color information that is visually pleasing while also interpretable and informative.

 

Frank

I am curious about your observation that non-overlapping filters are better matches to human vision. The photo I showed in the opening post of this thread shows that this is not true, at least when spectrally pure colors are important (as they often are in AP). Can you elaborate on how you came to your conclusion?



#81 whwang

whwang

    Aurora

  • *****
  • Posts: 4,949
  • Joined: 20 Mar 2013

Posted 14 June 2023 - 04:27 AM

Recently I have switched to Johnson-Cousins filters from Chroma.  These are interferometric rather than using colored glass and as a result have sharp, non-overlapping bands with high transmission and are a better match to human response than Sloan.

 

Hi Frank,

 

I am curious about your experience with Chroma interferometric Johnson-Cousins filters.  Do they produce halos around stars?  I have used Astrodon and Baader Johnson-Cousins filters.  Their filter curves are the classic bell-shaped curves, and both sets produce intense halos around stars.  I kind of wonder if the Chroma ones improve this.



#82 Eric Benson

Eric Benson

    Vostok 1

  • -----
  • Posts: 112
  • Joined: 02 Sep 2009

Posted 14 June 2023 - 08:43 AM

Hi Frank,

 

I am curious about your experience with Chroma interferometric Johnson-Cousins filters.  Do they produce halos around stars?  I have used Astrodon and Baader Johnson-Cousins filters.  Their filter curves are the classic bell-shaped curves, and both sets produce intense halos around stars.  I kind of wonder if the Chroma ones improve this.

The answer should be a resounding yes!

See math in the attached picture.

Summary: halo intensity is the product of the AR coating reflectivity and the transition bandwidth (aka shoulder steepness).

 

Regards,

EB

Filter_Halo_Intensity_Calculation.jpg


  • sharkmelley likes this

#83 Eric Benson

Eric Benson

    Vostok 1

  • -----
  • Posts: 112
  • Joined: 02 Sep 2009

Posted 14 June 2023 - 08:59 AM

As to the OP question, I believe whwang answered it in the first page of responses.

 

To reiterate what he said:

Excepting emission and absorption lines all the emitted light is from blackbody sources (stars, reflected or direct) with a characteristic weighting in the RGB filter bands owing to the source temperature.

We measure the color of celestial objects with a very coarse spectrometer, the RGB filter array.

We reconstruct the color of said objects from this 'spectral' data. The colors assigned are from a continuous palette generated by our software as instructed by the weights and intensities of the three channels. So we can easily create colors via interpolation of physical continuum sources.

 

EB



#84 whwang

whwang

    Aurora

  • *****
  • Posts: 4,949
  • Joined: 20 Mar 2013

Posted 14 June 2023 - 09:48 AM

The answer should be a resounding yes!

See math in the attached picture.

Summary: halo intensity is the product of the AR coating reflectivity and the transition bandwidth (aka shoulder steepness).

 

Hi Eric,

 

I knew what you wrote.  But that's theory.  I asked Frank because I want to know the real-world results.



#85 Jon Rista

Jon Rista

    ISS

  • *****
  • Posts: 26,034
  • Joined: 10 Jan 2014
  • Loc: Colorado

Posted 14 June 2023 - 06:04 PM

I thought SDSS had its own filters, u'g'r'i'z', while Johnson Cousins are the UBVRI set. SDSS filter passes don't overlap very much and the cutoff is very sharp. UBVRI is much slower falloff and they overlap more, namely V/R. Are these the only two standard photometry filter sets?

Apologies, this is correct.

 

I use World Wide Telescope as one of my planetariums. It has something that is just called "DSS"... I guess that's what I'm thinking of, and it is just B and R filters from the J-C UBVIR filter set, as far as I know... Thing is, I can't seem to find any info on what this is. Searching for `DSS filters` always seems to bring up SDSS :p



#86 Jon Rista

Jon Rista

    ISS

  • *****
  • Posts: 26,034
  • Joined: 10 Jan 2014
  • Loc: Colorado

Posted 14 June 2023 - 06:07 PM

Yes - the SDSS survey uses Sloan u'g'r'i'z' filters while Johnson-Cousins are UBVRI.

 

The V is a green filter but it stands for "Visual" because it is a good match to human visual luminance response and is the basis for visual magnitudes of stars.  For similar reasons the Bayer array has doubled green pixels.

 

I have been using Sloan i' r' g' filters as R G B channels for many years because I wanted a wide coverage of the spectrum with no gaps.  But the r' is so wide that the Ha signal shows up in the G channel.  This has benefits because it means you make fuller use of possible colors when imaging galaxies - but it does look unnatural.  I also like that the i' extends out to 800nm because it captures small, cool stars as deep red.

 

Recently I have switched to Johnson-Cousins filters from Chroma.  These are interferometric rather than using colored glass and as a result have sharp, non-overlapping bands with high transmission and are a better match to human response than Sloan.

 

I think it makes sense to use overlapping filters - and it makes no sense to have a gap for light pollution when most people have continuum light pollution.  It just loses perfectly good signal.

 

I don't think there is much point in seeking "true" color somehow - so I aim for ways to capture good color information that is visually pleasing while also interpretable and informative.

 

Frank

How could they be called Johnson-Cousins if they don't overlap? A Johnson, Cousins, or Bessel filter, at least as far as I knew, was a filter that specifically conformed to their specifications. Not just bandpass shape, but also peak transmission and all that. I was looking at the differences, it sounds like its actually Bessel/Cousins, and Johnson, and the Johnson have some slightly different specs for some of the bands. 

 

Anyway...if Chroma is selling a filter set with hard cutoffs that don't overlap, but they are calling it Johnson-Cousins, then IMO they are polluting the name of a very explicitly specified photometric filter set with something that doesn't properly conform to the spec, and they should rename the filters... shrug.gif

 

EDIT:

 

Is it this?

 

https://www.chroma.c...3-bessell-ubvri

 

If so, I feel Chroma is doing the community a disservice by formulating their own set of filters, naming them after a standard, without actually complying with the standard... That really kind of bugs me. tongue2.gif

 

EDIT 2:

 

They have another set, called "Classic UBVRI"...again, however, these filters don't actually conform to the Johnson-Cousins spec, as they all have peak transmissions at about 100%:

 

https://www.chroma.c...assic-ubvri-set

 

Interestingly, the peak transmissions are better for astrophotography, as are the overlaps...but, they aren't actually standard photometric filters as they don't conform to the spec. These, however, might in fact be some of the best filters for color astrophotography imaging around, given their transmission levels and bandpasses...

 

To be fair, they do say "based on", and are not claiming that they actually are J-C photometric filters.


Edited by Jon Rista, 14 June 2023 - 06:14 PM.

  • jdupton likes this

#87 Jon Rista

Jon Rista

    ISS

  • *****
  • Posts: 26,034
  • Joined: 10 Jan 2014
  • Loc: Colorado

Posted 14 June 2023 - 06:17 PM

Yes, I had known of the photometric calibration filters but did not know that Baader sells these.

Apparently, Chroma sells an astrophotography-tuned version as well, which may be better with their higher transmission rates, for pretty picture imaging:

 

https://www.chroma.c...assic-ubvri-set

 

Its really expensive though...full 50mm set is over three grand. O_o



#88 freestar8n

freestar8n

    MetaGuide

  • *****
  • Freeware Developers
  • Posts: 13,442
  • Joined: 12 Oct 2007
  • Loc: Melbourne, Australia

Posted 14 June 2023 - 06:34 PM

Yes, I had known of the photometric calibration filters but did not know that Baader sells these.

 

I am curious about your observation that non-overlapping filters are better matches to human vision. The photo I showed in the opening post of this thread shows that this is not true, at least when spectrally pure colors are important (as they often are in AP). Can you elaborate on how you came to your conclusion?

I was just saying very specifically that the bandpass of the J-C filters is a better match to visual R, G, B bandpass than Sloan are - particularly since with Sloan the the Ha line maps to green - if you are using i' r' g' filters for R, G, B as I was.  I wasn't trying to make a strong statement about true color or something.

 

Frank



#89 freestar8n

freestar8n

    MetaGuide

  • *****
  • Freeware Developers
  • Posts: 13,442
  • Joined: 12 Oct 2007
  • Loc: Melbourne, Australia

Posted 14 June 2023 - 06:45 PM

Hi Eric,

 

I knew what you wrote.  But that's theory.  I asked Frank because I want to know the real-world results.

Hi Wei-Hao

 

I have only had one imaging session with them so far and I see no sign of halos.  I wouldn't expect to see any because the transmission is 98-99% and the edges of the bandpass are very sharp.  They include transmission spectra for each filter and they all look good.

 

With absorptive glass filters it would be a different story - which is why I specifically went for this set with sharp cutoffs.

 

I haven't processed the images yet but will try to get around to it.  Antennae galaxies and NGC6300.

 

People often report problems with filter halos, but unless I see measurements of the halos that confirm they match with the filter thickness and f/ratio - I have concerns it might be something else in the system.

 

Frank



#90 freestar8n

freestar8n

    MetaGuide

  • *****
  • Freeware Developers
  • Posts: 13,442
  • Joined: 12 Oct 2007
  • Loc: Melbourne, Australia

Posted 14 June 2023 - 06:58 PM

How could they be called Johnson-Cousins if they don't overlap? A Johnson, Cousins, or Bessel filter, at least as far as I knew, was a filter that specifically conformed to their specifications. Not just bandpass shape, but also peak transmission and all that. I was looking at the differences, it sounds like its actually Bessel/Cousins, and Johnson, and the Johnson have some slightly different specs for some of the bands. 
 
Anyway...if Chroma is selling a filter set with hard cutoffs that don't overlap, but they are calling it Johnson-Cousins, then IMO they are polluting the name of a very explicitly specified photometric filter set with something that doesn't properly conform to the spec, and they should rename the filters... shrug.gif


All I can say is - I wanted a set that had sharp cutoffs with no gaps and was a rough match to visual - and would also be useful for photometry and color calibration.  And I was happy to find these.  I also had them mount a separate filter that I will describe later.

 

They call them "Bessel" filters and I think a lot of the terminology used for these various filters is loose - but it doesn't bother me.  With sharp cutoffs they won't be an exact match to the original filter curves but I expect they would calibrate well for photometry.

 

What does bother me is that some years ago a few people decided what the filter bandpass for amateur astro "should" be - including gaps and a bit of overlap - and there has been little exploration of alternatives that might have advantages.  Sloan have a clear advantage in spanning a wider part of the spectrum with no gaps and therefore gathering more total signal.  But I would like to try a system with a wider green that overlapped R and B.  Ironically that brings it closer to OSC imaging - which people usually regard as a waste - with that extra green.

 

Frank


  • jdupton likes this

#91 loujost

loujost

    Mariner 2

  • -----
  • topic starter
  • Posts: 243
  • Joined: 02 Oct 2017

Posted 14 June 2023 - 08:12 PM

As to the OP question, I believe whwang answered it in the first page of responses.

 

To reiterate what he said:

Excepting emission and absorption lines all the emitted light is from blackbody sources (stars, reflected or direct) with a characteristic weighting in the RGB filter bands owing to the source temperature.

We measure the color of celestial objects with a very coarse spectrometer, the RGB filter array.

We reconstruct the color of said objects from this 'spectral' data. The colors assigned are from a continuous palette generated by our software as instructed by the weights and intensities of the three channels. So we can easily create colors via interpolation of physical continuum sources.

 

EB

Eric, you say except for emission lines, all sources are blackbody sources. But your "exception" includes most of the colorful DSOs, which are common subjects for AP.

 

Regarding black body radiators, I agree with you, to a good approximation.



#92 Jon Rista

Jon Rista

    ISS

  • *****
  • Posts: 26,034
  • Joined: 10 Jan 2014
  • Loc: Colorado

Posted 14 June 2023 - 08:23 PM

All I can say is - I wanted a set that had sharp cutoffs with no gaps and was a rough match to visual - and would also be useful for photometry and color calibration.  And I was happy to find these.  I also had them mount a separate filter that I will describe later.

 

They call them "Bessel" filters and I think a lot of the terminology used for these various filters is loose - but it doesn't bother me.  With sharp cutoffs they won't be an exact match to the original filter curves but I expect they would calibrate well for photometry.

 

What does bother me is that some years ago a few people decided what the filter bandpass for amateur astro "should" be - including gaps and a bit of overlap - and there has been little exploration of alternatives that might have advantages.  Sloan have a clear advantage in spanning a wider part of the spectrum with no gaps and therefore gathering more total signal.  But I would like to try a system with a wider green that overlapped R and B.  Ironically that brings it closer to OSC imaging - which people usually regard as a waste - with that extra green.

 

Frank

Hmm, like Lou, I am curious why you prefer the sharp square cutoffs, over the overlapping ones. Sharp cutoffs will cause metamerism and other issues with color accuracy and diversity...



#93 whwang

whwang

    Aurora

  • *****
  • Posts: 4,949
  • Joined: 20 Mar 2013

Posted 14 June 2023 - 08:35 PM

Hi,

 

Frank, Thanks for confirming about the halo situation for the Chroma filters.  I may ask you again in half year or so to see if the conclusion still holds after you use it a few more times.  In my case, I have several different filters in my wheel.  Only when I switch to the classic Johnson filters there are halos.  Since the only thing that changes is the filter, and since the filters that have halos are the classic Johnson filters, I believe it is reasonable to conclude that the halos are caused by the filters but nothing else.  This makes me consider the Chroma ones.

 

Jon, I believe DSS is Digitized Sky Survey, which is the scanned data of the photographic plates from the Palomar Schmidt and UK Schmidt.  I don't think their B and R are Johnson B and R.  They must be taken with different Kodak 103a emulsions, one with better red sensitivity and the other with better blue sensitivity.  One of them ® might be used together with a filter to cut off the blue, but I am less certain about this.

 

Regarding the bell-shaped classic Johnson-Cousins, and square ones with sharp cutoff, I highly believe people should move to the latter if the goal is scientific studies.  The original Johnson-Cousin set was developed for phototube photometers, which typically had only one phototube in it (one pixel) with large "pixel scale" (like several arcsec per tube).  They can achieve highly accurate photometry, and therefore the exact shape of the filter matters.  Also, at that time, the evolution of phototube was quite slow.  A certain type of product could easily get widely used in observatories for years before a new generation of products appeared.  What this means is that astronomers in different places could use photometers of identical QE curves for their works.  And if the filters are also identical, they can easily compare their results to very high accuracy (way lower than 0.01 magnitude).

 

Now that kind of accuracy is gone.  CCDs have different QE curves from the old photometers.  And there are many different types of CCD in the market, and they evolve fast.  Basically the QE curve of the camera in every observatory is different.  So it is already not very meaningful to require using a Johnson-Cousins filter set that's identical to the ones people used half a century ago.  Furthermore, CCD has many pixels, and that leads to a series of calibration issues.  Today's good CCD calibration is like a joke if we use the standard half century ago for single-pixel photometers.  We just can't get the accuracy of phototubes that used to provide us.  But CCD has many pixels.  This advantage trumps the sub-0.01 mag accuracy.

 

Because accurate photometry (at the level half century ago) is no longer possible, and because a fixed QE curve is also not possible, some astronomers have already let go the idea of using filters that are identical the the old Johnson filters.  It just means nothing.  Today, many observatories shift to the Sloan filters and abandon the Johnson-Cousins system.  Among those who still prefer the Johnson-Cousins system, some of them shift to square-shaped transmission curves with sharp cutoff, like the Chroma ones.  They still call such filters Johnson-Cousins ones as long as the effective wavelengths match.  This mimics the photometry done in the "real" Johnson-Cousins system to perhaps 0.05 mag level (hard to do better with CCD anyway).  My favorite example is Subaru's SuprimeCam filters.  There are other examples.

 

Once we let go the sub-0.01 mag accuracy, the advantages of square-shaped interferometric filters for scientific studies are vastly clear: better overall QE, better defined passbands, and no halos.

 

As for Chroma, I have no idea why they call their Johnson-Cousins filters "Bessel."  Even Bessel himself doesn't call them Bessel filters.  See his article here.


  • CharLakeAstro likes this

#94 loujost

loujost

    Mariner 2

  • -----
  • topic starter
  • Posts: 243
  • Joined: 02 Oct 2017

Posted 14 June 2023 - 08:51 PM

Hi,

 

Frank, Thanks for confirming about the halo situation for the Chroma filters.  I may ask you again in half year or so to see if the conclusion still holds after you use it a few more times.  In my case, I have several different filters in my wheel.  Only when I switch to the classic Johnson filters there are halos.  Since the only thing that changes is the filter, and since the filters that have halos are the classic Johnson filters, I believe it is reasonable to conclude that the halos are caused by the filters but nothing else.  This makes me consider the Chroma ones.

 

Jon, I believe DSS is Digitized Sky Survey, which is the scanned data of the photographic plates from the Palomar Schmidt and UK Schmidt.  I don't think their B and R are Johnson B and R.  They must be taken with different Kodak 103a emulsions, one with better red sensitivity and the other with better blue sensitivity.  One of them ® might be used together with a filter to cut off the blue, but I am less certain about this.

 

Regarding the bell-shaped classic Johnson-Cousins, and square ones with sharp cutoff, I highly believe people should move to the latter if the goal is scientific studies.  The original Johnson-Cousin set was developed for phototube photometers, which typically had only one phototube in it (one pixel) with large "pixel scale" (like several arcsec per tube).  They can achieve highly accurate photometry, and therefore the exact shape of the filter matters.  Also, at that time, the evolution of phototube was quite slow.  A certain type of product could easily get widely used in observatories for years before a new generation of products appeared.  What this means is that astronomers in different places could use photometers of identical QE curves for their works.  And if the filters are also identical, they can easily compare their results to very high accuracy (way lower than 0.01 magnitude).

 

Now that kind of accuracy is gone.  CCDs have different QE curves from the old photometers.  And there are many different types of CCD in the market, and they evolve fast.  Basically the QE curve of the camera in every observatory is different.  So it is already not very meaningful to require using a Johnson-Cousins filter set that's identical to the ones people used half a century ago.  Furthermore, CCD has many pixels, and that leads to a series of calibration issues.  Today's good CCD calibration is like a joke if we use the standard half century ago for single-pixel photometers.  We just can't get the accuracy of phototubes that used to provide us.  But CCD has many pixels.  This advantage trumps the sub-0.01 mag accuracy.

 

Because accurate photometry (at the level half century ago) is no longer possible, and because a fixed QE curve is also not possible, some astronomers have already let go the idea of using filters that are identical the the old Johnson filters.  It just means nothing.  Today, many observatories shift to the Sloan filters and abandon the Johnson-Cousins system.  Among those who still prefer the Johnson-Cousins system, some of them shift to square-shaped transmission curves with sharp cutoff, like the Chroma ones.  They still call such filters Johnson-Cousins ones as long as the effective wavelengths match.  This mimics the photometry done in the "real" Johnson-Cousins system to perhaps 0.05 mag level (hard to do better with CCD anyway).  My favorite example is Subaru's SuprimeCam filters.  There are other examples.

 

Once we let go the sub-0.01 mag accuracy, the advantages of square-shaped interferometric filters for scientific studies are vastly clear: better overall QE, better defined passbands, and no halos.

 

As for Chroma, I have no idea why they call their Johnson-Cousins filters "Bessel."  Even Bessel himself doesn't call them Bessel filters.  See his article here.

I think none of us question the value of precise-bandwidth, low-overlap filters for scientific purposes. Also for ease of image processing. But they inevitably cause loss of color diversity in emission bands, and they do not mimic the response of the human eye.


  • Jon Rista likes this

#95 whwang

whwang

    Aurora

  • *****
  • Posts: 4,949
  • Joined: 20 Mar 2013

Posted 14 June 2023 - 09:00 PM

I just address Jon’s point/question about the Johnson-Cousins filters and I mentioned that this is for scientific studies, where aesthetic color is not a concern at all.

Edited by whwang, 14 June 2023 - 09:01 PM.


#96 freestar8n

freestar8n

    MetaGuide

  • *****
  • Freeware Developers
  • Posts: 13,442
  • Joined: 12 Oct 2007
  • Loc: Melbourne, Australia

Posted 14 June 2023 - 10:17 PM

Hmm, like Lou, I am curious why you prefer the sharp square cutoffs, over the overlapping ones. Sharp cutoffs will cause metamerism and other issues with color accuracy and diversity...

Over the years in various OSC/Mono debates in CN the common argument has been that square cutoffs are more "pure" or something - and I always countered that broad and overlapping are a better match to human visual response.  So if that is what you are after I would just use OSC and a Bayer array.  The advantage of Bayer over dyed glass filters in a wheel is that Bayer definitely won't produce halos.  DSLR's take very nice images of landscapes, humans - most anything - so I don't know why OSC/Bayer wouldn't be fine for astro - if that is what you are after.

 

In my case I am after well defined, deterministic and consistent color across different objects - while also having high throughput and minimal halos.  Square cutoffs fit that bill.

 

Frank


Edited by freestar8n, 14 June 2023 - 10:32 PM.


#97 freestar8n

freestar8n

    MetaGuide

  • *****
  • Freeware Developers
  • Posts: 13,442
  • Joined: 12 Oct 2007
  • Loc: Melbourne, Australia

Posted 14 June 2023 - 10:53 PM

Hi,

 

Frank, Thanks for confirming about the halo situation for the Chroma filters.  I may ask you again in half year or so to see if the conclusion still holds after you use it a few more times.  In my case, I have several different filters in my wheel.  Only when I switch to the classic Johnson filters there are halos.  Since the only thing that changes is the filter, and since the filters that have halos are the classic Johnson filters, I believe it is reasonable to conclude that the halos are caused by the filters but nothing else.  This makes me consider the Chroma ones.

 

Jon, I believe DSS is Digitized Sky Survey, which is the scanned data of the photographic plates from the Palomar Schmidt and UK Schmidt.  I don't think their B and R are Johnson B and R.  They must be taken with different Kodak 103a emulsions, one with better red sensitivity and the other with better blue sensitivity.  One of them ® might be used together with a filter to cut off the blue, but I am less certain about this.

 

Regarding the bell-shaped classic Johnson-Cousins, and square ones with sharp cutoff, I highly believe people should move to the latter if the goal is scientific studies.  The original Johnson-Cousin set was developed for phototube photometers, which typically had only one phototube in it (one pixel) with large "pixel scale" (like several arcsec per tube).  They can achieve highly accurate photometry, and therefore the exact shape of the filter matters.  Also, at that time, the evolution of phototube was quite slow.  A certain type of product could easily get widely used in observatories for years before a new generation of products appeared.  What this means is that astronomers in different places could use photometers of identical QE curves for their works.  And if the filters are also identical, they can easily compare their results to very high accuracy (way lower than 0.01 magnitude).

 

Now that kind of accuracy is gone.  CCDs have different QE curves from the old photometers.  And there are many different types of CCD in the market, and they evolve fast.  Basically the QE curve of the camera in every observatory is different.  So it is already not very meaningful to require using a Johnson-Cousins filter set that's identical to the ones people used half a century ago.  Furthermore, CCD has many pixels, and that leads to a series of calibration issues.  Today's good CCD calibration is like a joke if we use the standard half century ago for single-pixel photometers.  We just can't get the accuracy of phototubes that used to provide us.  But CCD has many pixels.  This advantage trumps the sub-0.01 mag accuracy.

 

Because accurate photometry (at the level half century ago) is no longer possible, and because a fixed QE curve is also not possible, some astronomers have already let go the idea of using filters that are identical the the old Johnson filters.  It just means nothing.  Today, many observatories shift to the Sloan filters and abandon the Johnson-Cousins system.  Among those who still prefer the Johnson-Cousins system, some of them shift to square-shaped transmission curves with sharp cutoff, like the Chroma ones.  They still call such filters Johnson-Cousins ones as long as the effective wavelengths match.  This mimics the photometry done in the "real" Johnson-Cousins system to perhaps 0.05 mag level (hard to do better with CCD anyway).  My favorite example is Subaru's SuprimeCam filters.  There are other examples.

 

Once we let go the sub-0.01 mag accuracy, the advantages of square-shaped interferometric filters for scientific studies are vastly clear: better overall QE, better defined passbands, and no halos.

 

As for Chroma, I have no idea why they call their Johnson-Cousins filters "Bessel."  Even Bessel himself doesn't call them Bessel filters.  See his article here.

Sounds good - and I'll aim to take some pics of bright stars as a test of halos.

 

This Bessell/Bessel thing now bugs me and in looking into it - I think the term "Bessel filter" for photometry is a mistake that only exists in amateur astro.  Bessel was a mathematician/astronomer and Bessel functions are named after him - but Bessell is the guy who clarified and helped standardize the transmission curves of different existing filters.  And as you say, he doesn't use the term "Bessel filter" or even "Bessell filter."  I always like learning something like this so I stop making mistakes in the terminology.

 

The AAVSO folks go in depth on this stuff, including the chroma "Bessel" filters here:   https://www.aavso.or...lter-selections They don't seem to like the "faux-Bessel" filters by chroma for photometry work due to transformations involved - but I haven't seen how involved they actually are.  That aspect of it isn't critical for me.

 

When I started using Sloan filters many years ago I thought it's what everyone would be using for photometry.  But I later learned that they were designed specifically for surveys of different galaxies at different redshifts - and that is a totally different application compared to stellar studies in our own galaxy.  For that I think people still prefer UBVRI.  I think the broad and overlapping curves have benefit there - so I don't think they are going away soon.

 

My main view on aesthetic imaging and color is that there are probably much better filter sets that would create pleasing colors in less time with more signal - but it's expensive to make custom filter sets to try things out.

 

Frank


Edited by freestar8n, 14 June 2023 - 10:55 PM.

  • jdupton and CharLakeAstro like this

#98 whwang

whwang

    Aurora

  • *****
  • Posts: 4,949
  • Joined: 20 Mar 2013

Posted 14 June 2023 - 11:45 PM

Ah. I wasn't careful enough about Bessell vs. Bessel.  I didn't even notice there is such a difference.  Thank you for pointing this out.

 

Generally speaking, for stellar works, one wants to use Johnson-Cousins UBVRI.  For galaxies, Sloan ugriz are the new standard.  And because galaxy and cosmology studies dominate the wide-field surveys industry, the stellar astronomers who piggyback on those surveys are forced to use Sloan filters.  Even those who are not part of the surveys may be forced to use Sloan filters,. This is because the surveys cover very wide area on the sky with great depth and photometric accuracies, and therefore it is easier to use the survey data to calibrate their photometry.

 

The transformation between square UBVRI to classic UBVRI can be quite easy, if you don't demand accuracies better than 0.02 mag.  Those who really want to reach 0.01 mag, will need to perform a more complicated transformation.  But I doubt if this is meaningful for CCD.  People may significantly under estimate their error bars.



#99 Jon Rista

Jon Rista

    ISS

  • *****
  • Posts: 26,034
  • Joined: 10 Jan 2014
  • Loc: Colorado

Posted 14 June 2023 - 11:55 PM

Hi,

 

Frank, Thanks for confirming about the halo situation for the Chroma filters.  I may ask you again in half year or so to see if the conclusion still holds after you use it a few more times.  In my case, I have several different filters in my wheel.  Only when I switch to the classic Johnson filters there are halos.  Since the only thing that changes is the filter, and since the filters that have halos are the classic Johnson filters, I believe it is reasonable to conclude that the halos are caused by the filters but nothing else.  This makes me consider the Chroma ones.

 

Jon, I believe DSS is Digitized Sky Survey, which is the scanned data of the photographic plates from the Palomar Schmidt and UK Schmidt.  I don't think their B and R are Johnson B and R.  They must be taken with different Kodak 103a emulsions, one with better red sensitivity and the other with better blue sensitivity.  One of them ® might be used together with a filter to cut off the blue, but I am less certain about this.

 

Regarding the bell-shaped classic Johnson-Cousins, and square ones with sharp cutoff, I highly believe people should move to the latter if the goal is scientific studies.  The original Johnson-Cousin set was developed for phototube photometers, which typically had only one phototube in it (one pixel) with large "pixel scale" (like several arcsec per tube).  They can achieve highly accurate photometry, and therefore the exact shape of the filter matters.  Also, at that time, the evolution of phototube was quite slow.  A certain type of product could easily get widely used in observatories for years before a new generation of products appeared.  What this means is that astronomers in different places could use photometers of identical QE curves for their works.  And if the filters are also identical, they can easily compare their results to very high accuracy (way lower than 0.01 magnitude).

 

Now that kind of accuracy is gone.  CCDs have different QE curves from the old photometers.  And there are many different types of CCD in the market, and they evolve fast.  Basically the QE curve of the camera in every observatory is different.  So it is already not very meaningful to require using a Johnson-Cousins filter set that's identical to the ones people used half a century ago.  Furthermore, CCD has many pixels, and that leads to a series of calibration issues.  Today's good CCD calibration is like a joke if we use the standard half century ago for single-pixel photometers.  We just can't get the accuracy of phototubes that used to provide us.  But CCD has many pixels.  This advantage trumps the sub-0.01 mag accuracy.

 

Because accurate photometry (at the level half century ago) is no longer possible, and because a fixed QE curve is also not possible, some astronomers have already let go the idea of using filters that are identical the the old Johnson filters.  It just means nothing.  Today, many observatories shift to the Sloan filters and abandon the Johnson-Cousins system.  Among those who still prefer the Johnson-Cousins system, some of them shift to square-shaped transmission curves with sharp cutoff, like the Chroma ones.  They still call such filters Johnson-Cousins ones as long as the effective wavelengths match.  This mimics the photometry done in the "real" Johnson-Cousins system to perhaps 0.05 mag level (hard to do better with CCD anyway).  My favorite example is Subaru's SuprimeCam filters.  There are other examples.

 

Once we let go the sub-0.01 mag accuracy, the advantages of square-shaped interferometric filters for scientific studies are vastly clear: better overall QE, better defined passbands, and no halos.

 

As for Chroma, I have no idea why they call their Johnson-Cousins filters "Bessel."  Even Bessel himself doesn't call them Bessel filters.  See his article here.

 

Hmm, Digitized Sky Survey.... I'll have to see if that's what the software uses. I can't find any specific details on it in the app. The app only states that only "the B and R filters are used"...

 

As for scientific use cases and filters, I have no disagreements with anything here. I would choose Sloan if I was doing anything scientific, I think. 

 

I was only interested in the Bessel/Johnson/Cousins for their potential to better replicate the natural response of cone cells in the human eye, where red and green overlap considerably, and blue and red also have a certain amount of overlap as well. Johnson-Cousins filters still don't optimally represent human response, so they wouldn't be perfect, but, I do think, from the standpoint of reproducing color the way a human would see all these objects in space, IF THEY COULD, J-C filters are probably the closest existing option. The Astronomik Type-2c filters are also better than standard LRGB filter sets, IMO...but, their overlaps are somewhat minimal and not really in line with how human tristimulus response is either.

 

I guess, FTR, so everyone reading the thread has some kind of reference point, this is the natural response of the three types of cone cells in the human eye:

 

aO69jsb.png

 

You can see how different it is even from most OSC CFA filter bandpasses, let alone RGB filters used on mono cameras. Actually, I am not even sure this plot is entirely accurate, I thought that blue cone sensitivity was actually a lot higher than red and green cone sensitivity... Anyway...this is only with the reference point of the desired color being what a human would see. This is not to say that classic LRGB filters are not viable at all, or anything like that...just that they won't necessarily allow you to, or make it easy to, reproduce objects how a human would see them. Not everyone has that goal, though, either...so, if you are happy with LRGB filters, that's great. Its more of a personal thing, personal discovery, personal aesthetics...for one, I want to see what these objects would look like if I could see them myself. I think, they would look quite different than commonly depicted. Second, I think making sure you aren't running into issues with metamerism and the like, can help make certain structures appear more varied in color and maybe even structure, than a classic LRGB filter set, especially those with LP gaps. 


Edited by Jon Rista, 15 June 2023 - 12:03 AM.


#100 loujost

loujost

    Mariner 2

  • -----
  • topic starter
  • Posts: 243
  • Joined: 02 Oct 2017

Posted 15 June 2023 - 07:56 AM

Jon, that's pretty close, but most references also show a bump in red sensitivity in deep blue. This is what produces violet




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