I was thinking of posting this as a reply to Ken’s detailed post about transformation just below in the VPhot forum but thought it might get lost in that long thread, so I decided to make it a new topic.
Although I am starting to get into photometry with a monochrome camera and photometric filters, most of my observing now is done with a Seestar S50, from which I submit untransformed photometry results gathered from the green channel (TG) images using Tycho Tracker to do the work. My question is, is it worth trying to compute transform coefficients for these observations and submit them as transformed V data? I took the CCD photometry class and did the transform examples from there, so I think I understand the basics of transformation (though my hands-on experience is very limited so far). The AAVSO DLSR manual talks briefly about transformation in the DSLR context but doesn’t go into too much detail on it and I found it a bit hard to follow. Is the basic idea that you just take the green channel and either the red or blue channel and follow the same steps in the Transform Generator that you would if you had V and R or B images? I realize that transforming TB and TR results can be problematic for many stars (as the DSLR manual points out) but I wouldn’t be trying to submit the blue or red results, just the transformed green.
Brian: In answer to your question above, YES! In VPhot and TG it just requires renaming the color filters (TR, TG, TB to R, V, B). IMHO, if you first make efforts to reliably separate the Bayer Array channels and then improve your precision by stacking/averaging the Seestar 10 sec sub-images, it is worth improving accuracy by transforming.
This continues to generate debate (worth it or not?) so you really need to try with your targets and determine what works for you OR take the simple choice of reporting only V.
IF you have Seestar images of a standard field, try to work through your transform coefficients using my posted guide. Ultimately, the Smart Telescope Working Group (STWG) will publish their guidelines and tools. Wait or try yourself?
So it is feasible BUT whether it is worth the effort often depends on your objective, i.e., only temporal analysis AND/OR always generating the ‘most’ accurate data? I’ve answered this question for myself BUT everyone must ask the same question and be willing to practice the appropriate procedures to achieve it. This is supposed to be fun!
There is another useful factor in the consideration of TG versus transformed magnitudes which I mention every time this discussion comes up.
A while back Andrew Pearce posted the Tv_bv for his SeeStar S50. It was quite small. I can’t remember the exact figure, but say it was 0.04.
Subtract the B-V of the target from the B-V for the comp and multiply by 0.04. The absolute value approximates the absolute value of the error of your magnitude estimate due just to the B-V difference. Thus if B-V target minus B-V comp = |0.25|, the error of your estimate will be only |0.01|. The relationship is linear, so a delta B-V value of |0.5| will still yield an error of only |0.02|.
If an observer is using a DSLR camera (not a SeeStar) the situation is quite different because the Tv_bv is likely to be a much larger (negative) figure, perhaps somewhere between -0.1 and -0.2. Transformation is thus a more necessary step for such cameras.
By error I mean difference between the true magnitude and the measured magnitude.
I followed Ken’s process and derived transform coefficients for my Seestar about 18 months ago. I found that the transformed B and R estimates were not close to other observers results who used Johnson filters really regardless of star colour. This is not surprising when considering spectral response in the B and R channel compared to the spectral response of Johnson B and R filters. Therefore I do not report transformed B and R magnitudes. It’s been some time since I tried and I did not measure lots of stars with varying colours but that was the general conclusion I came to. As Ken rightly says, it depends on your objective. I mainly perform long time series of southern eclipsing binaries. Can I determine times of minima OK with untransformed TG magnitudes or do these need to be transformed? When you have over a thousand images in one time series, it can be time consuming to derive TB, TG and TR magnitudes required to derive transformed values rather than just work with TG (particularly when as Roy notes TG and V are very close for my Seestar). I suspect everyone will have a different view.
Thanks to Ken and others for all the useful information in response to my initial post. In response to Ken’s question, I do always stack numerous 10-second exposures before doing any photometry - depending on the magnitude of the target I usually collect anywhere from 40 to 200 exposures, which I debayer and do aperture photometry on using Tycho 12.5.6. I rarely do time series of short period variables; mostly I look at RCB-type stars to check for fading and also some Mira and semi-regular variables as well.
As Roy and Andrew note, the Seastar’s TG channel seems to be pretty close to reference V magnitudes. I have done some testing with mine with the standard field Melotte 111 using a set of comp stars with various B-V values and got an average of 0.01 difference between the Seastar’s TG and the reference V magnitudes. The results were best for stars with B-V values around 0.5 and worst for those with B-V values over 1.0 (though even the latter had an average diff of only a shade over 0.02).
It looks like it will be clear tonight so I will try to image NGC 7790 and start doing some transform computation on that, assuming I get good images.
