I’m beginning a serie of articles dedicated to this little imaging bomb that is the ASI224MC camera from ZWO. It’s quite different from the PLA-Mx that I have been using this past years, first because of its pixel size… Let’s take things by the beginning: the setting of the sampling!
The ASI224MC is currently THE planetary camera. This is quite an event as the last time a color camera was the best imaging tool for planets, it was some fifteen years ago with the good old ToUcam Pro. Its arrival in my equipment is an occasion to get back talking about a very important notion in astrophotography: the sampling. I talked about it already on the blog: What is the sampling?
There was an important problem with my telescope. For a few years I have been using a 250 mm Gregorian with a very long focal length at prime focus: 8 meters! This can be nice because I can get rid off any Barlow lens to image planets, but as the primary focal length of a telescope can’t vary, it’s necessary when choosing a camera to evaluate how does its pixel size fits well or not. For my telescope, the sampling can be evaluated as follows:
Minimal theorical sampling: 0,24″(arcsecond) per pixel, or two pixels for the resolving power (RP) of 0,48″for a 250.
Maximal theorical sampling: 0,16″/pixel, or three pixels for the RP. This is often the chosen value by planetary imagers because it maximises the capacity of the chip to detect the smallest variations of contrast, and so the smallest details.
In practice, the 8 meters of focal length makes the PLA-Mx and its 5,6 micron sized-pixels already quite oversampled since the value is 0,14″/pixel. It makes the setting of the camera and the focusing quite touchy already, it also asks for an intensive use of WinJupos derotation technics, but the camera is performing very well. But since I can’t change the focal length, it could be a problem with the ASI224MC, because the 3,75 micron-sized pixels require a shorter focal length than the PLA-Mx. The sampling at prime focus of the Gregorian is now getting awfully hudge: 0,09″/pixel, so no less than 5 pixels for the RP !!
Facing the risk of a massive over-sampling, I have decide to test a well-known technic that is the binning mode. When binning the image of the camera the pixels are re-arranged by group, for example 2×2 for binning 2x. The image is then two times smaller, but four times brighter, and the sampling of my telescope goes down to a theorically perfect 0,18″/pixel (or 2,7 pixel for the RP).
But to be sure I have of course tested the idea on the field and here are the results after several imaging nights. So far however, I don’t know if the process of binning itself can be responsible in some way for the differences.
Using the camera for color imaging
At left is a comparison between two RGB images of Jupiter taken on April 1st, 2016, in binning 1 and binning 2. To make the comparison more easy, both images have been mapped with WinJupos and gif-animated.
Of course the variations of conditions can always explain differences but I have made this test several times and the results were consistent :).
To my eyes the image is noticeably better in binning 1. I would have a loss of resolution in binning 2x (remember in theory I should not)
If this loss is not due to the process of binning itself, people would explain this by the fact that color cameras have the reputation of require a longer focal length than black & white ones. The idea have however been critisized by american observer Dan Llewellyn in a recent article in Sky & Telescope : Redeeming Color Planetary Cameras.
Use of the camera for near infrared imaging
Despite beeing a color camera, the ASI224MC is none the less very performing in IR, a very interesting wavelength for planetary imaging.
The sampling conditions, however, are now different: the resolving power of a telescope is inferior in near infrared. If the theorical RP is 0,48″in green light for a 250, it fells to 0,61″in near infrared at a wavelength of 700 nanometers. The useful focal length in infrared is shorter than in visible light. Under such conditions, is the 2×2 binning getting closer to the high resolution imaging ?
The comparison now is not that clear. The images look very close at the center of the disk but the northern hemisphere is noticeably more resolved in binning 1x. The seeing may have introduced a bias.
Taking into account how does the RP vary with the wavelength, it looks like the differences in RGB is not artificially biased by the process of binning the image. If it was the case, the problem would be clearly visible in IR as well…
As a first conclusion…
The ASI224MC must require a very long focal length to reach high resolution. The “3 pixels for the RP” rule could now be looked at as a minimum and not a maximum. Fortunately, the very high sensitivity of the camera makes it easier to go along with the massive over-sampling: even in bin 1, on Jupiter, I was still able to run it up to 100 frames per second, very close to the theorical minimum of 130 fps!