Bright standard star system for CMOS RGB photometry

American Association of Variable Star Observers (AAVSO)
Thu, 04/01/2021 - 04:38

On tonight's astro-ph, along with several April Fool's spoofs, is a potentially interesting paper about adopting photometry from the Johnson 13-color system (a little-used intermediate-band system) to develop a new standard system for CMOS detectors:

The idea is to define a camera-independent system close to the mean of a couple dozen commercial detectors.  The 1300-odd reference stars cover the whole sky and have a wide range in colors.  One would still have to transform data, but the coefficients would be small, and make use of all three filters.


Bright standard star system for CMOS RGB photometry

Thanks Brian,

I've just skimmed through this paper but it looks very interesting. I'll read it properly when I get some time. The 1346 stars are all quite bright and spread fairly evenly in the northern and southern hemispheres, so suitable for DSLR cameras with short focal length lenses. Of course differential extinction would need to be considered. I wonder if AAVSO would adopt such filter system?

A lot of DSLR photometry is done using a telescope with much smaller field of view. I presume it would be possible to develop a catalog of fainter secondary standards (e.g. the existing standard clusters) using these 1346 stars as primary standards. Sounds like a lot of work though. Cheers,


Information for RGB photometric system

Hello Mark, Brian and interested observers,

You can find information and tables at:

Our idea is to avoid transformation to Johnson photometry, for instance
and to have a reference photometry frame closer to amateur RGB instrumentation
(DSLR, color CMOS, smartphones, monochrome cameras fitted with RGB color filters etc.)

We had in mind the wonderful AAVSO-DSLR observing manual (cited in the text).

It would be nice if some observers could calibrate their cameras using these bright standards.

Thanks and all the best, Jaime

American Association of Variable Star Observers (AAVSO)
Where to spend AAVSO $$

Mark et al:

Improving the accuracy of the magnitudes submitted to the AID in the Johnson Cousins, Sloan or RGB filter bandpasses is clearly an appropriate effort! A casual glance at the LCG2 plots exemplifies the typical scatter in light curves and the (in)accuracy of various photometric systems. Getting observers to conduct appropriate transformation continues to be a struggle.

Unfortunately, as you allude too, adding comp star magnitudes in ALL of these bandpasses into the current AAVSO comp database takes time, effort and money.  I think doing this for the Sloan bandpasses would be the appropriate next step as those filters are used/preferred by many professionals. Hopefully, the addition of the Sloan magnitudes generated in the APASS program will occur in the not too distant future.

The addition of a suitably curated RGB comp database would probably be well received due to the popularity and cost effectiveness of the Bayer array cameras. However, how well would this be received by professionals? Should that be a critical need? Which would be more useful? Thoughts?


Variable Stars South (VSS)
New CMOS photometric standards

Hi Brian,

Interesting paper you've flagged.

Coming from an astrophysics group at the Complutense University of Madrid (see Wikipaedia), it describes a new synthetic RGB standard for colour CMOS cameras. I’ve tried to grasp some of the basics. A full understanding is beyond me, mainly because of the mathematics.

Spectral sensitivity curves which are the medians from 28 commercial CMOS cameras define the RGB passbands. Thus no one camera will have filters that correspond precisely to the standard. Transformation to the standards will therefore be necessary.

The paper lists a set of 1346 bright stars defined as standards, all from the Yale Bright Star Catalogue. They include some variable stars. Obviously, the latter could not be used as photometric standards, although the amplitude is relatively low (usually less than 0.1 mag in the small sample I scanned). Note that these are designated spectrophotometric (not just photometric) standards.

An important part of the process was the determination of the absolute flux densities in each of 13 passbands for all of the standard stars. Thirteen filters (hereafter 13C), from the medium-narrow-band photometric system published by Johnson and others between the late 1960s and early 1980s (references in the paper), define these passbands, and cover the spectrum from about 3,300 to about 12,000 Å.

The authors write: “The first step before attempting to estimate the synthetic RGB photometry of the bright star sample was the fit of the spectral energy distribution of each star to the stellar model spectrum that best matched the available 13C photometric data.” A particular set of stellar model spectra from the literature was chosen by the authors. The 13C photometric data was then converted to flux density values (erg s−1 cm−2 Å−1). I think that what then happened was that the model spectrum for each standard star was used to calculate the new standard RGB magnitudes, based on spectral flux densities for the median spectral sensitivity curves of the RGB filters noted near the beginning, with the zero point defined in the AB magnitude system (see Wikipaedia).

This is an attempt to understand the basics of a complex paper. There may be faults in my interpretations. If anyone recognizes them, please correct them.

The full table of the standards, including errors, is provided at the URL given by Jaime Zamorano, the second author, in a recent post in this Forum. It is also provided in the paper itself:


Bright RGB standards

Hello Roy et al.

Your summary is correct.

It is important to note that the 1960  and 1980 observations were of very good quality. Some of you are old enough (as I are) to remember the times of photoelectric photometry. 

From the 13 photometric points one can obtain a spectral energy distribution (low spectral resolution) and then fit the stellar model (high spectral resolution). With the spectrum you can determine the magnitudes in any bandpass using synthetic photometry (the integrals in the paper).

In the process we can detect problems with some stars and reject them from the list.

The authors are ready to answer any question and to help potential users to use the system.

Thanks and all the best, Jaime



Variable Stars South (VSS)
Bright RGB standards

Thanks Jaime.

Mark has raised the issue of secondary standards for equipment setups with narrow fields of view.

I think the fact that the new primary standards are based on CMOS RGB filters means that CMOS cameras would be used to define secondary standards.

That is not the case for existing photometric filter systems. Standards can be created only within the photometric system to which data from DSLR and other colour CMOS cameras must be transformed.



Variable Stars South (VSS)
RGB CMOS Standards - Distribution of star colours

I have selected a subset of 63 stars between RA 10h 13m and 15h 56m, and DEC -16deg 13m and -70deg. That's a region from which I could currently theoretically determine transformation and extinction coefficients.

I don't think any stars are included where proximity (e.g., stars in annulus) could be an issue, nor any stars labelled as variable or NSV.

It turns out the sample is heavily weighted to bluer stars. The distribution is as follows. For each set of numbers, the first is the B-V (Johnson) range and the second the number of stars.

<0.0  37; 0.0-0.2  7; 0.2-0.3  2; 0.3-0.4  3; 0.4-0.6  1; 0.6-0.8  1; 0.8-0.9  1; 0.9-1.0  5; 1.0-1.1  4; 1.1-1.3  2; >1.3 (=1.36)  1.

My comment is that fields would have to be selected carefully if not tediously to include a good range of star colours for transformation and principal extinction coefficients. For the latter, the comment assumes the method used would be the one where TCs are known, and extinction stars of various colours are imaged over a range of airmass values.





Variable Stars South (VSS)
Selection of RGB CMOS Standards for Transformation Coefficients

In the previous post it was suggested that it may be tedious selecting RGB CMOS standard stars with a sufficient range of colours for transformation coefficients. This is not correct.

A subset of standards for the part of the sky near the meridian at my location (27.5 deg S) at a reasonable hour yielded 10 stars over three photographic fields with a 50mm lens on a DSLR camera that were acceptable.

The B-V (Johnson) values ranged from -0.16 to 1.26. The airmass for individual stars ranged from 1.00 to 1.10, with 7 stars at airmass of 1.04 or less. The other three airmasses were 1.07, 1.08 and 1.10. Altitude at airmass 1.1 is about 65 degrees.

CCD observers who use M67 for the determination of TCs may use many more stars. Personal experience with DSLR photometry on images taken through a camera lens indicates that 10 well selected stars are sufficient to determine good TCs.


American Association of Variable Star Observers (AAVSO)
CMOS standard fields


If you come up with good standard fields for the wide field CMOS/DSLR cameras I can help get them added to the AAVSO std fields pages:  


Variable Stars South (VSS)
Transformation coefficients from CMOS RGB stanards

I have determined transformation coefficients for a Canon EOS 500D camera and Canon EF 50mm f/1.8 STM lens using the new CMOS RGB standards. Exposures were 15 seconds at f/2.5 and ISO400, 10 images per field for averaging. Defocussed images were taken across three fields during one night (i.e., the targets were distributed over three fields). Calibration (darks and flats) and photometry used AIP4Win.

The targets were 12 stars, V magnitude range 2.682 - 5.979, B-V index range -0.177 - 1.262. (These values are given here because the CMOS RGB magnitude values are quite different from the Johnson/Cousin magnitudes for the same stars). Airmass values for individual stars were 1.00, 1.00, 1.01, 1.02, 1.03, 1.03, 1.05, 1.08, 1.08, 1.09 and 1.12. The last value corresponds to an altitude of about 63 degrees. Atmospheric extinction was ignored for this exercise.

The quality of the plots was good, with small residuals distributed uniformly along the plots.

Transforms plotted and the TCS are as follows. The first five use the new CMOS RGB standard magnitudes and colour indices. The last two use Johnson V magnitudes and B-V colour indices, for the same 12 target stars, and thus represent typical values for DSLR TCs.

b-g/Std_B-G   1.323
g-r/Std_G-R   1.256
Std_B-g/Std_B-G   0.256
Std_G-g/Std_B-G   0.012
Std_R-r/Std_G-R  -0.190

b-g/B-V   2.610
V-g/B-V  -0.108

I can't work out how to post a secure link to screen shots of the results. If anyone wants to see any of the data (including images if you have a Dropbox link) send me a private message and I will send you what you want to see.

The next exercise will be ensemble photometry of slightly fainter stars down to 6th magnitude in V, using the same images, and the above TCs.


Variable Stars South (VSS)
Ensemble photometry of CMOS RGB standards

Ensemble photometry was performed on two fields of CMOS RGB standards, with 5 standard stars in each field. Ten DSLR images of each field were obtained through a 50mm focal length lens. Two sets of calculations were performed. First, each of the stars in Field 1 was selected as a target in turn, with all 5 stars in Field 2 used as the comp stars. Second, the process was reversed, with each of the stars in Field 2 selected as a target in turn, with all 5 stars in Field 1 used as comp stars.

Instrumental magnitudes were averaged across the 10 images prior to calculating the transformed BGR magnitudes and errors.

In the tables below note that B, G, R, B-G and G-R are catalogue values.

Errors in the table below represent measured parameter minus catalogue parameter. Overall, about 50% of the measured values have errors of 0.02 or less. In the second set of results (Field 2 targets), about 50% of the measured values have errors of 0.01 or less.

Field 1 Targets and errors.  Airmass 1.04 - 1.08
HR    B    G    R    B-G    G-R    Err B-G    Err G-R    Err B    Err G    Err R
4216    3.143    2.765    2.501    0.378    0.264    0.001    -0.009    -0.017    -0.018    -0.010
4167    3.894    3.834    3.819    0.060    0.015    -0.010    -0.019    -0.007    0.003    -0.031
4023    3.723    3.798    3.907    -0.075    -0.109    -0.026    -0.031    -0.005    0.021    -0.028
3886    5.324    5.504    5.692    -0.180    -0.188    -0.020    -0.026    -0.045    -0.025    -0.094
3786    3.688    3.631    3.618    0.057    0.013    0.048    -0.026    -0.016    -0.064    -0.093
Field 2 targets and errors.  Airmass 1.01 - 1.04
HR    B    G    R    B-G    G-R    Err B-G    Err G-R    Err B    Err G    Err R
4232    3.758    3.208    2.820    0.550    0.388    0.003    0.035    0.009    0.006    0.057
4287    4.639    4.167    3.834    0.472    0.333    -0.004    0.022    0.017    0.020    0.070
4382    4.168    3.682    3.343    0.486    0.339    0.003    0.029    0.000    -0.003    0.042
4405    4.065    4.071    4.117    -0.006    -0.046    -0.006    0.023    0.000    0.006    -0.021
4514    5.173    4.772    4.497    0.401    0.275    0.010    0.003    0.064    0.054    0.107

Prior to performing the above exercise, ensemble photometry was performed using the same equipment on other fields using Johnson B and V values in the calculations. The accuracy for measurement of 4th to 6th magnitude stars (V) was very encouraging, with 4 of 8 stars having V mag errors of <0.01 (zero to 0.008), 3 stars having V mag errors of 0.021 or 0.022, and 1 star having a V mag error of 0.035.

Given the above, my assessment is that reasonably accurate wide field DSLR photometry down to at least 6th magnitude (in V) can be performed using CMOS RGB standards, and that with care, secondary standards may be measured and listed.


Variable Stars South (VSS)
Improvement in CMOS RGB photometry results

On Sunday May 16 I posted tables of results.

I was not happy with a few of the less accurate R magnitudes. Turns out there was an error in a spreadsheet formula. It has been corrected, and the amended results appear below. The least accurate measures have improved.

Just to recap: ensemble photometry was performed on images of CMOS RGB standards from a Canon DSLR camera through a 50mm Canon lens at f/2.5, 15 second exposures, ISO 400, each result the average of measurements/calculations from 10 images.


FIELD 1                                        
HR    B    G    R    B-G    G-R    Err B-G    Err G-R    Err B    Err G    Err R
4232    3.758    3.208    2.820    0.550    0.388    0.001    -0.009    -0.017    -0.018    -0.009
4287    4.639    4.167    3.834    0.472    0.333    -0.010    -0.019    -0.007    0.003    0.022
4382    4.168    3.682    3.343    0.486    0.339    -0.026    -0.031    -0.005    0.021    0.053
4405    4.065    4.071    4.117    -0.006    -0.046    -0.020    -0.026    -0.045    -0.025    0.001
4514    5.173    4.772    4.497    0.401    0.275    0.048    -0.026    -0.016    -0.064    -0.038
FIELD 2                                        
HR    B    G    R    B-G    G-R    Err B-G    Err G-R    Err B    Err G    Err R
4216    3.143    2.765    2.501    0.378    0.264    0.003    0.035    0.009    0.006    -0.029
4167    3.894    3.834    3.819    0.060    0.015    -0.004    0.022    0.017    0.020    -0.002
4023    3.723    3.798    3.907    -0.075    -0.109    0.003    0.029    0.000    -0.003    -0.032
3886    5.324    5.504    5.692    -0.180    -0.188    -0.006    0.023    0.000    0.006    -0.016
3786    3.688    3.631    3.618    0.057    0.013    0.010    0.003    0.064    0.054    0.051

American Association of Variable Star Observers (AAVSO)
Magnitudes transformed to RGB system

Good evening, dear observers, I don't see why not to determine RGB magnitudes with the m67 cluster itself, which has been traditionally used for this work, in relation to those that use a narrower configuration. As the RGB bands do not correspond to any particular model, but median curves to the set of 28 models, it would be necessary to determine the synthetic magnitudes with spectra of the m67 stars themselves (I suppose). because also have to transform the magnitudes too RGB , for this it would be useful a software in which I come I have been working for a long time, with which to transform RGB magnitudes to the standard bands, at this moment I am incorporating the option of synthetic filter conceived by Roger Pieri, this would be useful for the task, I leave you a video of the operation of the software and the address of the work of the new system and the summary.

download software

videotutorial 1
videotutorial 2

by the way, in which direction do you access to download the spectra of the UCM?

RGB photometric calibration of 15 million Gaia stars


The promised paper with more standard stars has been accepted for publication at Monthly Notices of Royal Astronomical Society.

RGB photometric calibration of 15 million Gaia stars
Cardiel et al. (2021) MNRAS (in press)

You can read it at

Thanks and enjoy the new catalog.

Keep us informed if you use it .
Do not hesitate to contact us in case of problems.

All the best, Jaime




Variable Stars South (VSS)
Thank you Jaime.

I have…

Thank you Jaime.

I have only just now been able to look at your paper on RGB standards from Gaia stars. I suspect that accessing them at this time will not be possible for most AAVSO observers.

I'll try to run the rgblues Python script, and will report back here if I can get it to work. I have only a very rudimentary understanding of Python.


Variable Stars South (VSS)
It is clear that I don't…

It is clear that I don't know enough about Python to make this work.

I'm sure there are many like me in the AAVSO. Before we can access this expanded dataset of CMOS RGB standards, there will need to be an interface for observers who are not able to use Python.

Perhaps the mentioned access via Vizier will be the way forward.


Hello Roy,

We are planning…

Hello Roy,

We are planning to create a suite of tools to help observers to use the RGB photometry.
There will be also hands on manuals and the information that the potential users seems that is needed.

Any help from AAVSO members is welcome.

Thanks and all the best, Jaime

PS: The paper can be accessed at

RGB photometry UCM extended catalog using GAIA


The Universidad Complutense de Madrid RGB photometric system
has increased the number of standard stars using GAIA data.

"Photometric Catalogue for Space and Ground Night-Time Remote-Sensing Calibration: RGB Synthetic Photometry from Gaia DR3 Spectrophotometry “
Josep Manel Carrasco, N. Cardiel, E. Masana, J. Zamorano, S. Pascual, A. Sánchez de Miguel, R. González & J. Izquierdo   Remote Sens. 2023, 15(7), 1767;

  • RGB magnitudes for ~200 million objects
  • Synthesised from Gaia DR3 low-resolution spectra
  • Including objects without extinction restrictions
  • Accurate RGB estimates with reliable uncertainties
  • Data available through Python code rgbloom

You can find the catalogue  at RGB photometric calibration of 213 million Gaia stars : II/374 or create the list of standard stars (and a chart) in your field using

We can help any interested observer who wish to calibrate the observations made with colour CMOS or DSLR.

A light curve example for SN2023bee can be found at

Thanks and all the best, Jaime