›

With SPCC do we need to collect RGB data in addition to Narrowband?

Kevin SheaKevin Shea
in PixInsight
Adam,

Thanks for your videos on SPCC.  I love using it and the results I see.  Recently, I processed an object and collected only narrowband data.  I mapped it to HOO.  I found a tutorial (can't remember from whom) on how configure the settings in SPCC when one only has narrowband data.  I followed the instructions and was very satisfied with the outcome (color of the stars).

I didn't have RGB data to compare it to so I really have no idea if I would have had better results if I had collected separate RGB data and used it to capture the color for the stars, run SPCC, and then combine them with starless narrowband data.

Perhaps, someone has done such a comparison. It would be great to be able to use (and collect) only narrowband data at times and leave it up to SPCC to get the stars' colors correct.  But I have a feeling I am missing something or misinterpreting something.

Any thoughts or insight would be appreciated.

Kevin
Madison WI
Rodeo NM

Comments

  • Kevin SheaKevin Shea
    Asked this question over on the PixInsight forum, will see if I get some information there.
  • Mark Mayer
    This is a great question and I hope Adam responds. With an SHO filter set, typically one maps Sii (deep red) to red; Ha (red) to green; and Oiii (cyan) to blue. I've never been happy with the star color that SPCC narrow band option outputs, and wondered if it can be improved.
  • BlockHeadBlockHead
    Hi Guys,

    Sorry, I guess I do not understand the question.
    If a song (music) is played on a piano.. you need to have all keys/notes available to play the entire song.

    If you "filter out" most of the keys and only leave three of them available, you cannot play the song. 

    Stars are continuum emitters (blackbody spectrum). You need the full visible spectrum (all of the piano keys) to have faithful star color (the song). SPCC *will not* create proper star color... literally the information is not there to do it. 

    So I agree with Kevin.. you are missing something..the rest of the spectrum! 
    If you want real star colors- you need to acquire separated RGB data.
    I plan on showing compromise in a new video soon. It is somewhat possible to produce color stars that give a sense of how blue or red they are (based on the Ha and OIII information)- and by putting an intermediate green channel.. you get something that looks like RGB Stars..but it isn't their real color. 

    Please let me know if I am on the right track.

    -the Blockhead
  • Kevin SheaKevin Shea
    Perhaps if I understood (at a simple level) what SPCC does and doesn't do, I would understand why one needs to collect the RGB data.

    My understanding is that it knows the spectrum of the stars from the Gaia observations, i.e., the stars' colors. Thus, if it knows the color, does SPCC apply it to the stars it has identified regardless as to what color the star is pre- SPCC processing? Or is it doing something else?

    I guess what I'm missing is what is SPCC really doing?  If it knows a star is red isn't it "coloring" it red?
    On the other hand, if I carry this out ad absurdum, would a mono camera, luminance filter light frame of a star field processed by SPCC come out fully colored?

    I have friends who live in light polluted skies and find it only practical to shoot in narrowband.  They find it very difficult to collect RGB data.  Their nb data stars when mapped to the Hubble palette become effectively full spectrum stars (don't they?).  If they could use SPCC to correct their stars' colors they would have stars with their real color displayed (or maybe this is where I am fundamentally misunderstanding what SPCC does).

    Regardless, it appears to be an awesome tool to a novice like me.

  • Ray Hansen
    Hi Adam,
    I'm looking forward to the video. I understand the premise, but when you read the Vicent and Juan SPCC reference document, it seems to suggest you only need narrow band images. It would be very helpful if you refer to that document and explain where you think it needs more explanation.

    best regards,
    Ray
  • BlockHeadBlockHead
    Can you please point at the language you are referring to?

    -the Blockhead
  • BlockHeadBlockHead
    Kevin,

    GAIA (and SPCC) does not know the integrated color of a star per say- though the information is contained in the spectrum. Parts of the spectrum are analyzed (values are added) as described below.

    When you compare a BAND of wavelengths you can measure the total of amount of light within that REGION of light. Red (filter) is NOT a color. It is a group of colors that comprises 1/3 of the visual spectrum (roughly). The same is true for Green and Blue. What SPCC does with the GAIA catalogue is it measures a total of amount of light (flux) in that part of the spectrum that corresponds to the RED filter you are using. Now you have a ratio of expected instrumental response.

    But your REAL stars in your data are also affected by the actual Filter (which might not be truly ideal), the atmosphere, the sensor response to light...etc etc. So SPCC plots a ratio of the responses of the filter bands from your data compared to the expected ratio. All of this ratioing get around worrying about absolute photometric values. 

    Now consider the bandwidth of the H-a filter.  My Chroma H-alpha is 3nm. The bandwidth of the Red filter is 70-100nm (I think). So... if we go to GAIA we can measure the H-alpha line and compare this is my Chroma H-alpha filter and get a ratio. Great. But what next? Now measure the star in your data. Due to the conditions, sensor and so on we will get a corrected brightness of this tiny part of the spectrum.

    But this is only a tiny fraction of the star's output of wavelengths. OIII, for example, is another line (a tiny sliver) and we can get the corrected flux for it. But we cannot take the ratio of the H-alpha flux over the OIII flux and hope to get properly colored stars! Stars, based on temperature and other physics emit so many other lines you are not sampling enough of the spectrum to reproduce a BROADBAND coloration. 

    All we get from SPCC is the true line strengths in brightness for your data (and thus the corrected ratio of the H-alpha line with respect to the OIII line or any or emission line). 

    R,G, and B filters together sample the entire swath of light from these stars. That is why a ratio of their brightnesses help us color stars by plotting lots of stars and fitting a line. Stellar spectra are more complicated than pure blackbody emission. I do not think you can take the ratio of H-alpha with OIII can produce all of the colors of the stars. 

    Just as an example... lets say a star has pretty weak emission in OIII (for example)...so it basically does not show up in OIII images. Now when SPCC tries to measure that star in OIII... a very small number results. So do we color this star pure Red, Orange, Yellow? From just the H-alpha line there isn't enough information  for this star. When we plot lots of stars... you do not get a nice correlation like we do when we sample more wavelengths of light across the entire visible spectrum.

    -the Blockhead
  • Kevin SheaKevin Shea
    Adam,
    This explanation helps quite a bit.  I confess I do not fully understand it.  I guess the one issue I am not comprehending is what SPCC looks at and what happens when we map SHO data to RGB.
    • When PI combines narrowband data to produce a color image does it take, for example, Sulphur and give it the full spectrum of what would have been captured with a red filter?
    • When SPCC is applied does it "work" on the processed color frame or does it look at the components that compromise the image, apply what it knows about the filters which were used to produce each of those three channels and do its adjustment to those channels and then re-combine the data to output the color corrected image?
    I really appreciate you taking the time to explain this, and I hope it might benefit others in understanding and appreciating what is happening under the hood.
    kevin
  • BlockHeadBlockHead
    Kevin!
    Please watch my SPCC videos.... 
    Specifically this one:

    (But you must watch the first part also:

    These videos are located here:

    Did you watch them?
    Please let me know if this answers your question.
    ZERO people have taken the time to explain how SPCC works. Not even the developers... except the documentation which I think doesn't treat it in the same way I do in the video.

    -the Blockhead
  • Kevin SheaKevin Shea
    I have watched both of them but I think I can get more out of them listening to them again with the focus of this discussion on the forefront of my mind. When I first watched them it was within days of their posting. My focus was on the mechanics of using SPCC, not how it worked.

    I’m hopeful a second viewing will answer the question in my last post.
  • Ray Hansen
    Adam - the doc is in the PI forum, announcements, SPCC. The link is buried in there.


    I can understand why someone reading this would think there are contradictions.

    RH
  • BlockHeadBlockHead
    Ray,

    Yes, I know... I read the document. It does not say in there that star color is derived, approximated or anything else regarding narrowband images. That is why I asked you to highlight the particular language you felt said otherwise. All the document says is that the flux ratios (line strengths) will be corrected by using the GAIA/SP reference and NB working mode of SPCC.

    -the Blockhead
  • Ray Hansen
    OK. Let's back up here. I don't have a dog in this fight. When I read the reference document, with attention to NB processing mode, I don't see any prescription of a need for RGB stars. If you create an instructional video that posits a need for RGB stars to get correct color, without addressing how/why you are or are not in agreement with the reference doc, some people are likely going to be confused by what appears to be a contradiction. I think that is a valid point.

    RH
  • BlockHeadBlockHead
    No Ray, this is not what I said.
    You need RGB stars for color calibration of broad band images with respect to a given white point.

    For NB images, you can use the Stars (spectral data) to correct for the line strengths at a given emission wavelength. So this is just the first part of the step that is done for RGB data (where you measure the flux at the wavelength of interest in the GAIA data and then correct (calibrate) the flux measured in your data given the filter used, the camera..etc. There is no ratioing of color channels for purposes of coloring broadband emission. 

    I think the misunderstanding is that the GAIA data is a SPECTRUM of star light. You can sample a small part (one line) or large swathes of the GAIA spectra. When you use the GAIA data to compare to stars in your images no matter if they are broadband stars or NB filtered stars. SPCC compares the same part of the spectrum in the GAIA database as is recorded in your data (broad or narrow). It isn't that "RGB" stars are needed... just STARS are needed.

    BUT, if you want to color balance broadband emission by ratioing color channels with respect to a chosen white point- your DATA needs to be RGB stars- your data needs to be broadband. That is all . 

    There isn't a contradiction in my statements as far as I know. If you have a timestamp in a recording that appears to be a contradiction please let me know.

    -the Blockhead

  • Ray Hansen

    Insert RGB Stars into Narrowband Image (Primary Method)

    Hi Adam,
    In this video, you're demonstrating a method to add RGB stars to a starless NB image. I don't see that prescribed anywhere in PI documents. If I've missed that procedure somewhere, please give me a direction. Given it would add at least several days, assuming perfect weather, for acquisition and preprocessing, I'd be interested to know what benefit the extra work provides. Is the resulting color correction significantly different?

    RH
  • BlockHeadBlockHead
    edited March 2023
    I am adding the stars into a NB image to augment the image because you cannot create true colored stars from NB data alone. I am not adding the stars into the image in order to measure anything or in some way correct the color of anything. I am adding stars into a narrowband image in order to make it "prettier" ! 
    This is a kind of composite image in that sense. 

    For an NB image- you either have 

    1. Stars that are colored based on the real signal strengths of the emission lines. This is what SPCC gives you..and it can be kinda ugly for aesthetic reasons.

    2. You devise some method of arbitrarily assigning colors to the NB stars. I have a video I would like to produce on this soon.

    3. You remove the NB stars and throw them away..and add/blend/composite additional RGB stars. 

    I like #3 because now I have two elements that are true. I have an NB image with emission lines strengths (channel ratios) that are either true as SPCC outputs or modified to produce more intermediate colors AND I can insert RGB stars... which are also a true thing... even though you never see NB and RGB stars together naturally. 

    I believe you are assuming I am inserting the RGB stars for color calibration purposes (of the NB data) ...
    -the Blockhead

  • Ray Hansen
    Perfect! I think if you explain it this way in your video, it will eliminate any confusion between your #3option and the SPCC reference documentation.
Sign In or Register to comment.