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

RVB photometric filters for "regular" imaging?

This topic has been archived. This means that you cannot reply to this topic.
14 replies to this topic

#1 Meeka

Meeka

    Sputnik

  • -----
  • topic starter
  • Posts: 49
  • Joined: 06 Aug 2013

Posted 02 February 2015 - 01:57 AM

I'm going to be upgrading my filters soon (currently have Orion's LRGB and Astrodon's V photometric for variable stars), and I was wondering if it would be possible to use RVB photometric filters instead of RGB ones. I'd like to expand my photometric filter range, especially since I have a 5 position wheel, but I don't want to do it at a significant cost to my other imaging. 

 

I also have quite a bit of light pollution, so I'm trying to find out which would be the best option. I was looking at the wavelengths transmitted by Astrodon's RVB photometric and the E-Series RGB, and it looks like the red filter is significantly different, but I don't know what effect that would have on images.

 



#2 Rick J

Rick J

    Cosmos

  • *****
  • In Memoriam
  • Posts: 8,376
  • Joined: 01 Mar 2008

Posted 02 February 2015 - 10:46 PM

They are certainly different.  RGB imaging filters are near 100% efficient across their band width while photometric BVR are strongly peaked meaning much lower efficiency needing two to three times the exposure for similar S/N.  With good color balance practices, eXcalibrator would to the best job though G2V can sort of work, it can be done.  Many of the pro images you see are done this way including HST images.  I have both however.  Since my photometric camera is an ST-7 it uses 1.25" filters while the imaging camera is 42mm on the diagonal and needs 2" filters.  I have made trial images with the ST-7 that looked OK.  Not the same as the RGB filters give but unless seen side by side most wouldn't notice the difference.  Just that while the ST-7 is much high QE than the STL-11000 I still needed 15 minute subs for it with the photometric filters vs 10 with the low QE STL-11000 and imaging RGB filters.  So if convenience and money are important then it will work.  Just needs a bit more work from you.

 

Rick 



#3 Gianluca67

Gianluca67

    Viking 1

  • -----
  • Posts: 699
  • Joined: 26 Nov 2008

Posted 25 January 2016 - 08:53 AM

Hi Meeka, I just would like to know if you went ahead with the RVB filters instead of the RGBs. As I am doing a lot of photometry and I am in the process of renewing my filters with 2" ones and would like to know if RVB filters work okay for "pretty images" as well. Can you also show us some pics?  

Thank you

 

Gianluca


Edited by Gianluca67, 25 January 2016 - 04:45 PM.


#4 freestar8n

freestar8n

    Vendor - MetaGuide

  • *****
  • Vendors
  • Posts: 11,740
  • Joined: 12 Oct 2007

Posted 25 January 2016 - 07:20 PM

Hi-

I am doing something similar, but I am using Sloan filters, i' r' g' for R, G, B. Also, I don't try to emulate true colors and instead I calibrate each channel with reference stars in the field that have Sloan magnitude data. As a result the colors aren't "true" in terms of matching the eye response - but they are true in terms of being calibrated and reflecting the flux in each part of the spectrum.

As Rick said, the Johnson-Cousins filters have disadvantages because they aren't as transmissive as normal rgb filters. But in addition they aren't tuned to block out light pollution, and they aren't tweaked to capture narrow band emission colors.

The plot below shows Astrodon E-type filters compared to Johnson-Cousins photometric filters, and you can see the difference in shape and transmission - but in addition the E-type has a big gap at around 600nm. That gap is intended to block out dominant light pollution and at the same time "balance" the response in each channel so you can use the same exposure times. But it does mean you are losing a lot of light in that gap.

Frank

Attached Thumbnails

  • AstrodonCousins.png


#5 freestar8n

freestar8n

    Vendor - MetaGuide

  • *****
  • Vendors
  • Posts: 11,740
  • Joined: 12 Oct 2007

Posted 25 January 2016 - 07:29 PM

Another thing you can see in the above plot is that the Oiii line at 501nm is recorded by both B and G filters - so the final color will involve a mixture - while HB will be blue and Ha, Nii, Sii will be red.

In my case I use Sloan photometric filters, which have a much wider passband and no gaps. The lack of gaps is bad for light pollution - but good for recording maximum signal across the channels.

One problem is that HB and Oiii both record as pure blue, while Ha, Nii, Sii record as pure green. This results in a completely different look to Ha regions - but you can still distinguish them because they stand out as green.

My recent images are all done this way, and a galaxy example (m83) is here: http://astrogeeks.co...images/m83.html Globular cluster (M4): http://astrogeeks.co.../images/m4.html and Nebula (m8): http://astrogeeks.co...ages/m8irg.html

Here is a plot of the spectra for the i' r' g' Sloan filters, compared to Johnson-Cousins.

Frank

Attached Thumbnails

  • SloanCousins.png


#6 groz

groz

    Vanguard

  • *****
  • Posts: 2,148
  • Joined: 14 Mar 2007

Posted 29 January 2016 - 05:57 PM

That's really interesting frank, thanks for posting.  Some very interesting food for though comes out of that.



#7 freestar8n

freestar8n

    Vendor - MetaGuide

  • *****
  • Vendors
  • Posts: 11,740
  • Joined: 12 Oct 2007

Posted 29 January 2016 - 08:45 PM

That's really interesting frank, thanks for posting.  Some very interesting food for though comes out of that.


Thanks. You can see from the plots that extending into the IR and not having a gap at 600 greatly increases the signal in the green and red channels. This may leave things "unbalanced" in terms of total signal in each channel - but you can always "balance" things by doing more sub-exposures for the channels with weak signal. To me this makes much more sense than actually blocking signal from the strong channels. It's just being thrown away.

The wide spectrum does require that the optics perform well in each color - but I find the fwhm's well matched using EdgeHD11. Other telescopes like refractors may have a problem - and it may be much worse if using Johnson-Cousins since they extend below 400nm where typical detectors still have some sensitivity - as shown in the plots for the Sony 694 QE.

Having consistent and calibrated colors allows images to be interpreted directly and compared to each other - since there has been no artistic manipulation of the hue. Here is an image of NGC 206 in M31, and you can see the full range of colors corresponding to star types. The IR really picks up the cool stars as deep red - in contrast with the blue ones - but you only see this range of red to blue in the star forming regions. 3/4 up the image and slightly to the left is a distinctive orange smudge - and you can tell from the color alone that it is a distant galaxy seen through the disk of M31. There is a lot of info in the colors themselves - and the colors are due to the physics of the objects themselves and not due to my artistic manipulation.

Frank

Attached Thumbnails

  • ngc206mlevels.jpg


#8 groz

groz

    Vanguard

  • *****
  • Posts: 2,148
  • Joined: 14 Mar 2007

Posted 30 January 2016 - 12:20 PM

 

Thanks. You can see from the plots that extending into the IR and not having a gap at 600 greatly increases the signal in the green and red channels. This may leave things "unbalanced" in terms of total signal in each channel - but you can always "balance" things by doing more sub-exposures for the channels with weak signal. To me this makes much more sense than actually blocking signal from the strong channels. It's just being thrown away.
 

 

I guess it depends on ones objectives, and where you are located.  That gap around 600 is quite desireable in a lot of locations, to the point folks will put a light pollution filter on a one shot camera, to create that kind of gap.

 

We've got the 'retirement telescopes' here now, and the domes will be in place over the next couple of months.  The plan is to 'make do' with the cameras we have for a while, but I am starting to ponder what the final configurations will look like in terms of cameras, and filters.  In particular for the RC, I'm thinking sloan filters will be more useful than lrgb in the long run.  Like you are hinting, I too think that images with properly calibrated colors are more desirable than something which has been endlessly tweaked in photoshop, but that depends too on what you are trying to achieve.  If you want something pretty to hang on the wall, tweaking can make it more so.  If you want something that gives a graphic display of various star types etc, then your method of calibrating the one posted above will have more of that information preserved.  If you are taking frames in an area with minimal light pollution (we will be), then the gap at 600 isn't desireable, but, if you are in an area that's flooded with street lights, the signal in that range may well be flooding the image to the pint you are losing all the 'interesting' stuff in the background.

 

Ofc, if money was no object (not our case), then I'd just put a wheel on each telescope that can hold all 3 variants, as well as a full set of narrow bands.  Maybe if I hit the LotoMax, that's what we will do, but in the interim, if I'm using a typical 7 or 8 position wheel, we will have to make tradeoffs on filter selections.  Jury is still out, but, for the big one I'm hedging toward sloan + narrow band, and I think my wife would prefer LRGB + Narrow band in hers, but, that may yet change too.



#9 David Ault

David Ault

    Gemini

  • -----
  • Posts: 3,424
  • Joined: 25 Sep 2010

Posted 30 January 2016 - 01:11 PM

I gave a presentation on The Astro Imaging Channel some time ago talking about using data collected with scientific filters to make 'pretty pictures'.  My approach was a little different from Frank's (which I really like by the way) as my primary target was to approximate what would have been captured using RGB filters.  Assuming the object you are imaging doesn't have any true green color to it (i.e. a Hanny's Voowerp or some planetary nebulae) you can get very close.  A big part of the presentation was talking about where you can find scientific data that you can practice with or just create nice images from.

 

Regards,
David



#10 freestar8n

freestar8n

    Vendor - MetaGuide

  • *****
  • Vendors
  • Posts: 11,740
  • Joined: 12 Oct 2007

Posted 30 January 2016 - 05:38 PM

I'm actually imaging from an extremely light polluted location with two bright streetlights about 50' away shining directly at the 'scope. I use a large patio umbrella to block the direct light - but they still light up the sky. Even so, I can do pretty well on faint objects with the i'r'g' filters - but the main problem is gradients - and I hate dealing with gradients - mainly because it loses the deterministic techniques I would like to use.

As for filter count - I am helped by having the 9-filter efw2 wheel - and to keep costs down I use 1.25" filters with the atik 383L+ 8300 ccd. There is only tiny vignetting in the corners at f/7.

One thing that motivated me to work this way is my experience with the m51 supernova in 2011, described here: http://astrogeeks.co...ar8nWriteup.pdf

I happened to be imaging just after it appeared - and it would have been great if I had photometric filters. But my real problem in that imaging session was that the supernova was saturated in the images - so it was very hard to get any brightness info at all - and that would have been true even with Sloan filters. The other thing that motivated me was when UCAC4 came out and included all the APASS sloan magnitudes for many stars. You can still just use one star and do the equivalent of a G2V calibration - but with Sloan filters. But typical fields have many stars with sloan magnitudes - so you can do a fairly accurate photometric calibration of the scene using many stars - and I wrote code to do that automatically.

So this approach is not for everyone and that notch to block light pollution does have benefits - but as LED's become more popular maybe it will lose its usefulness.

Frank

#11 Lucullus

Lucullus

    Apollo

  • -----
  • Posts: 1,482
  • Joined: 14 Feb 2012

Posted 04 March 2016 - 02:37 PM

Maybe a stupid question but I have no experience yet whatsoever with colour and photometric filters, as well as colour calibration and weighting etc. To get rid of the light pollution when using Sloan filters, can't you just subtract narrowband images of filters corresponding to the pollution's wavelengths?

 

Lucas



#12 freestar8n

freestar8n

    Vendor - MetaGuide

  • *****
  • Vendors
  • Posts: 11,740
  • Joined: 12 Oct 2007

Posted 05 March 2016 - 12:21 AM

Hi-

No you can't really subtract pollution images because while you expose the color images the light pollution will be shining down on the ccd during the exposure - and that will show as background skyglow - and that will contribute noise. You can 'subtract' the overall glow - but you can't subtract the noise associated with it.

You *could* combine any of these filters with another separate light pollution filter - and that should work except you would have two filters in the light path - which isn't ideal. But light pollution filters don't tend to be very narrow - so you would corrupt the transmission curve of the photometric filters and they wouldn't represent the true response they are supposed to have.

If you are at a very dark site, the fact that there is no light pollution blocking with photometric filters shouldn't be a problem. But the total transmission of light from the objects would be much greater with sloan filters because there is a wide spectrum covered with no gaps - at nearly 100% transmission. Other photometric filters like Johnson-Cousins have broad peaks with drop offs that would reduce the transmission - but they may match the eye response better.

In terms of true vs. natural color - a lot depends on how you do the color balancing and what you are trying to achieve - and how you want any narrow band features to appear. Astrodon and other filters are designed to split the Oiii band into both B and G channels so the color is distinct from HBeta - which shows as pure blue. But that's just one way of doing it. A lot of the result and the color balancing people do is based on convention and aesthetic choice. I am using Sloan filters to avoid that issue so the colors are calibrated and have interpretable meaning that is consistent across objects and object types. And the whole scene can be calibrated accurately using the many available field stars that have sloan mags. in UCAC4.

Frank

#13 MartinMeredith

MartinMeredith

    Viking 1

  • *****
  • Posts: 999
  • Joined: 01 Feb 2015

Posted 05 March 2016 - 03:47 AM

Frank and others, thanks for your contributions to this really informative thread. I'm very keen on seeing 'true' physical colours in the way you describe. Your NGC 206 has a very subtle pastel look that I feel is sometimes lacking in astro-images. Do you have more details of how this image was captured and processed? I was just wondering how many stars you need to get a decent calibration using the technique you describe in post 10. Presumably the more the merrier, but I'd expect the benefits of adding more to asymptote fairly quickly?

 

Martin



#14 korborh

korborh

    Apollo

  • *****
  • Posts: 1,125
  • Joined: 29 Jan 2011

Posted 05 March 2016 - 01:57 PM

I just ordered Astrodon Sloan r',g',i' filters inspired by this thread and Frank 's images/techniques.



#15 freestar8n

freestar8n

    Vendor - MetaGuide

  • *****
  • Vendors
  • Posts: 11,740
  • Joined: 12 Oct 2007

Posted 05 March 2016 - 05:28 PM

Frank and others, thanks for your contributions to this really informative thread. I'm very keen on seeing 'true' physical colours in the way you describe. Your NGC 206 has a very subtle pastel look that I feel is sometimes lacking in astro-images. Do you have more details of how this image was captured and processed? I was just wondering how many stars you need to get a decent calibration using the technique you describe in post 10. Presumably the more the merrier, but I'd expect the benefits of adding more to asymptote fairly quickly?
 
Martin


Hi-

Thanks for your comments. There are clearly some people who like really bright and saturated colors, while others prefer things more subtle. I'm mainly trying to capture objects in a way that shows them as they are and without imposing my own interpretation of what they should look like. The use of photometric filters and calibrations is one part of that, in terms of how the images are processed, but the other benefit of Sloan filters is the use of a very wide passband at nearly full transmission. This will be a disadvantage if you have light pollution - but cutting out a big chunk of the spectrum is a disadvantage with or without light pollution.

With normal g2v calibration people use one star and match its values to calibrate the color channels for white balance. That should do something - but it won't have as much accuracy as using many stars. In my case I wrote code to find the stars that have Sloan mags in the field - and there are quite a few provided by the APASS survey info - but those mags aren't intended for rigorous photometric work. But they are still pretty good - and I did a study of how useful they are a couple years ago in this thread:

http://www.cloudynig...hl= calibration

For the ngc 206 image there were probably over 100 stars available for calibration - and the calibration for the r' channel is shown in the plot below. Stars brighter than mag 12 are saturated and excluded from the calibration - and the fainter ones are fit to a very good linear match with the catalog mags.

You wouldn't need all those stars for the fit - but at least two in each channel would be good. I have my own code for doing all this stuff automatically - but you could do it manually with fewer stars.

Some objects and star fields will look fairly drab in these colors - while others will show a strong range from red to blue.

Once you have the calibrated channels - you can create the color image with varying amounts of alteration:

1) Combine as linear R, G, B and do a linear stretch
2) Leave the hue alone and stretch intensity - to see more dynamic range without altering the hue. That is mainly what I do.
3) First stretch the saturation to exaggerate the color range - but without "shifting" it toward the blue or red. Then stretch in luminosity

Each of these will keep some tie to the calibrated values so the hue has information in it that tells you something about the fundamental measurement. If something is blue - it really is stronger in g' than i' - and it isn't just because I shifted the hue to make it blue.

But keep in mind that for emission lines like H-alpha - in i' r' g' color mapping it will show as green and not red or purple. This is just a different way to show it - and it has some advantages, particularly to reveal Hii regions in galaxies and distinguish them from orange/red stars clusters.

Frank

Attached Thumbnails

  • rpcal.png



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