There has been a lot of comment recently in the Forum on photometry with smart telescopes.
These instruments use OSC (one shot colour) sensors, and therefore the principles of DSLR photometry would generally apply.
One of those principles is that images should be slightly defocussed, to spread the star image over more pixels because of the issue created by the Bayer matrix where only one in four pixels is blue (or red) and only one half of them is green.
However, smart telescopes track the sky using astrometry, and thus I presume would work best with in-focus images.
What are smart telescope users actually doing? Has the smart telescope working group discussed this?
I don’t have a smart telescope, but out of curiosity ran tests on in-focus images with my DSLR imaging through a 200mm f/3.5 lens. I was surprised to find that transformed magnitudes measured using photometric standard stars were pretty good when averaged over a number of images (15 in the tests).
I’ve been experimenting with a Seestar S50 to measure T CrB over the past week. At least from my site, the FWHM of stacked images is around 3-4 pixels, so my guess is that the light is spread out over plenty of pixels. If the system were producing images of FWHM ~ 1-2 pixels, then I might worry.
On the other hand, there might be another reason to defocus: a 10-second sub-exposure causes T CrB (at V ~ 10) to have peak pixel values not too far from saturation. If one is observing a star of similar brightness, one might defocus to provide a larger window of safety to prevent saturation.
The S50 does not allow shorter exposures?
Wow! I almost ordered one. Thanks for saving me.
I suppose you could expose and calibrate with neutral density filters.
No, one CAN take shorter exposures. The 10-second value is simply the default. At some point, if one observes very bright stars, decreasing the exposure time can cause problems, and so choosing to defocus becomes a good idea.
Yes, you should defocus and not just for very bright stars. Since you are separating your color channels, defocusing assures that your stars are adequately sampled. I used my Seestar S30 on an eclipsing binary last night that was 10.1-10.6 mag. I defocused by adding 50 to the autofocus position. That gave me a FWHM of 9 pixels for my target and comp stars. I used an exposure of 30 secs and it gave me a SNR of ~55.
Yes, the AAVSO Smart Telescope Working Group has looked at this. The ability to defocus in a controlled way varies across the different smart telescope vendors. We’ve found that you can get good photometry (+/- 0.018 mags stdev relative to Landolt reference stars with averaging across 20-30 raw 10-second images) with or without defocusing.
But (and this is a big “but”), if you don’t defocus then we’ve been forced to change the way we do aperture photometry. With sharply-focused images, we’ve been extracting four photometry channels: one each green, blue, and red, and one luminance channel. We don’t “deBayer” the image in the traditional sense (couldn’t achieve good photometry with well-focused images), and instead do what we’ve started to call “masked photometry” from the original image, which solves some undersampling issues. (Masked Photometry is aperture photometry on the original image that ignores (masks off) pixels from the “wrong” Bayer colors. The important advantage is the handling of partial pixels is much more correct in Masked Photometry than in traditional aperture photometry of deBayered images.)
The other thing we’ve found (and we’re still debating this one) is that the best aperture radius seems to be different between the green channel and the blue/red channels and different between the green channel and the luminance channel. (We define “best radius” as the radius that gives the best least-squares match to the Landolt standard stars.) We’re pretty sure that this is related to the three different pixel densities of the luminance, green, and red/blue channels, but we don’t fully understand the physics of why this works.
– Mark Munkacsy, AAVSO Smart Telescope Working Group
I think it’s important to provide a definition of what “good” photometry is and in what context (i.e is slightly less accurate data relative to reference stars acceptable for time series photometry of eclipsing binary stars or other short period variables acceptable provided the consistency is OK versus more singular TB, TG and TR estimates of longer period stars, etc).
TBH, I haven’t noted a significant difference in results between debayered focused and defocused images from my Seestar certainly in the context of eclipsing binaries, however beauty may be in the eye of the beholder! Defocused images can create other problems such as losing the ability to plate solve and align images as an example
It would be useful to provide some guidance on what “good” photometry really is for smart telescopes.
This was the issue that prompted me to start the present thread. A while back I saw some time series photometry results from a OSC camera (not a smart telescope) which I thought looked pretty good. The description and image implied the images were in focus. The thing is, it was a 16 bit camera. I didn’t note the URL. Has anyone done photometry with a 16 bit colour camera?