28-Megapixel Camera to Monitor the Night Sky

StupendousMan writes "Scientists at Sonneberg Observatory in Germany will start monitoring the sky every clear night (starting in October), using a curious combination: a 28-Megapixel CCD camera behind a 30-mm f/3.5 fisheye lens. As the first light test images show, the device can record nearly the full sky down to ninth magnitude (about 20 times fainter than the naked eye limit) every 5 minutes. The goal is to create a permanent record of bright objects, which could be used to discover comets and novae."

Low resolution?

[ka9dgx]

The specs say 7k * 4k pixels, but the JPEG files don't go anywhere near that. Also, they talk about the pixels being binned 4*4. So... what's the real resolution?

--Mike--

Re:Low resolution?

[DJGreg]

Binning 4*4 makes the res 1750*1000, 1.75Mpixels. Binning or combining the values of a group of elements in a CCD array is used to increase the light sensitivity of the CCD. That's how they got that much from such a short exposure. Typically to get the kind of results that they showed, you need at least a 30 minute exposure.

Re:Low resolution?

[__aaaaxm1522]

Also, don't forget that for astronomy purposes, FITS is pretty much the standard file format. Images were converted to JPEG for the average web-surfer's convienience...

Re:Color

[dmatos]

Well, having colour would mean they would need a grid of filters, and then do some interpretation between pixels to guess at what the colour level there would be. In other words, if the grid looked like:

G B G R...

R G B G...

Which are Green, Blue, and Red sensitive pixels (a fairly standard pattern), then there would be no measure of how intense the blue light was on a green pixel, or vice versa. Some extrapolation would be required, and you would essentially be reducing the effective number of pixels on your device. My suspicion is that they wanted max resolution, and weren't overly concerned with putting pretty pictures on the net for us to gawk at, like Hubble can do.

On another note, who else thinks that this device has 2-3 dead pixel columns? Note the vertical black lines on the full-sky pictures. I'd be a little upset if I bought one of these babies (for $$$ probably) and ended up with dead columns.

Re:Color

[Caid Raspa]

Which are Green, Blue, and Red sensitive pixels (a fairly standard pattern)

In astronomy, the chips are generally uniform and colour sensitivity is achieved by using a filter wheel (or several) in front of the detector (the wheels may also contain grisms, slits or calcites for spectrometry/polarimetry). Colour sensitive pixels sounds more like a digital camera. Astronomical chips are also sensitive to IR radiation, and the images the Germans show are most probably 'white-light', with significant IR contribution.

Another interesting thing is that they are using Peltier cooling. Liquid nitrogen would reduce the dark current considerably, allowing longer integrations. Maybe they get confusion-limited (not enough spatial resolution to see faint sources => multiple overlapping faint sources => confusion), or light pollution is a bigger problem.

On another note, who else thinks that this device has 2-3 dead pixel columns?

I think they are not that upset. After using this for a few years, they will probably have dozens of dead columns, hot/dead pixels etc.