[X] My Assistant

[djellison]

Jim Bell's agreed to do a brief Q'n'A in a few days time, so similar to the one I did for Steve, I want your questions! Try and keep it quite 'current' if you can, as we're going to make this more a 'news' outlet than a look back type chat.

We're going to try, if this one works, to do these every couple of weeks or so, a bit of Rover news and a bit of Q'n'A each time, but we'll see how this one goes first!

Fire away people

If they're all crap, don't worry, I've got LOADS in mind.


Doug

[algorimancer]

For the next "Jim & Doug Show", questions for Jim...

I've been looking over the spec's of the equivalent of the pancam for the MSL,

http://www.lpi.usra.edu/meetings/lpsc2005/pdf/1214.pdf

These spec's are really exciting, but a few questions arise.

The MER pancams are separated by 30 cm, which makes for pretty good stereo imaging. On MSL this is only 20 cm. If anything, I would have been inclined to scale up to 40 or 50 cm, particularly on a larger vehicle. Curious as to what factors led to scaling this downwards rather than upwards (if it is indeed settled).

The MSL mast cameras include telephoto zoom (10X!) and focus capability. Whereas I really like the additional capability, I find I'm a bit concerned at the potential long term reliability, considering that MSL, like MER, may last for several martian years. I have dealt with a personal digital camera whose autofocus mechanism broke, and the same on Mars would be really irritating. Of course there're 2 cameras, so that helps. Safe to assume these will be tested to ridiculous extremes? There's mention of them using legacy actuators from MER.

Hi-def movies - considering how long it takes to get fixed images down currently, are we likely to see many of these? Not that I'm complaining or anything, quite the opposite. No Mars Telecommunications Orbiter

Have the flight cameras actually been constructed yet? Calibrated? Camera models generated - and if so, do they continue to use the CAHVOR model, and how will that cope with the zoom capability? [this latter may be a bit technical]

I see that the cams have a "Bayer Pattern Filter" ccd, which has red/green/blue filters overlaid on the ccd pixel sensors, as with typical personal digital cameras. Is this the only CCD, and if so are there any issues using it with the fixed wavelength filters?

[mcaplinger]

For the next "Jim & Doug Show", questions for Jim...

You can ask Jim, but at this point I think I know more about the MSL cameras than he does.

The MER pancams are separated by 30 cm, which makes for pretty good stereo imaging. On MSL this is only 20 cm. If anything, I would have been inclined to scale up to 40 or 50 cm, particularly on a larger vehicle.

Why, so your eyes would try to pop out of your head when you tried to fuse the stereo, like they do with the Viking cameras? A 20 cm baseline was judged to be good enough for the near field. For the far field we will get stereo by moving the rover between frames, as no fixed baseline was large enough. And at any rate, I think the flight baseline will be a bit larger than 20 cm for reasons of the mast mechanical design.

The MSL mast cameras include telephoto zoom (10X!) and focus capability. Whereas I really like the additional capability, I find I'm a bit concerned at the potential long term reliability... Safe to assume these will be tested to ridiculous extremes?

I'm concerned too, but those are the requirements the PI gave us. 3x mission life is the requirement, that's six years equivalent. And we are using slightly more robust materials than your digital camera.

Have the flight cameras actually been constructed yet?

No, of course not, not for a year or more. We'll use whatever camera model is appropriate. Zoomed in there is very little distortion; zoomed out there's a lot.

I see that the cams have a "Bayer Pattern Filter" ccd, which has red/green/blue filters overlaid on the ccd pixel sensors, as with typical personal digital cameras. Is this the only CCD, and if so are there any issues using it with the fixed wavelength filters?

Yes, this is the only CCD. In some wavelengths (beyond 750 nm) the Bayer filter is essentially transparent anyway. In the visible wavelengths some of the pixels don't return usable signal levels and are discarded. If it were up to me I would leave the filter wheel off, but you can take that up with Jim.

[algorimancer]

Yes, this is the only CCD. In some wavelengths (beyond 750 nm) the Bayer filter is essentially transparent anyway. In the visible wavelengths some of the pixels don't return usable signal levels and are discarded. If it were up to me I would leave the filter wheel off, but you can take that up with Jim.

Thanks for the detailed reply Here's a related question for you: At least a decade ago there was a report in Science about a new CCD design in which each pixel could simultaneously detect multiple wavelengths, in fact my recollection of the paper was that each pixel could detect a pretty broad spectrum of colors. This was later commercialized by a company called Sigma, and they currently have a couple of cameras which use this type of sensor, called the SD10, and the sensor is called the "Foveon X3® direct image sensor".

http://www.sigmaphoto.com/cameras/cameras_...amp;navigator=1

The obvious nice thing about this sensor is that you can at least triple your resolution per color channel for the same pixel count (and conceivably replace a whole stack of filters). When I first read the intial paper on this sensor my expectation was that it would be really useful for astronomy and spacecraft, but as far as I am aware it has never been used beyond the commercial market (though I understand it is highly regarded there). So, the obvious question is, are you aware of this type of sensor, does it have limitations which make it inappropriate for use as a rover camera, or is it simply a matter of too young a technology which has not been space-rated?

[mcaplinger]

So, the obvious question is, are you aware of this type of sensor, does it have limitations which make it inappropriate for use as a rover camera, or is it simply a matter of too young a technology which has not been space-rated?

We have, of course, been tracking this technology since it was introduced (see http://www.foveon.com). There are two major issues: first, the company was not very forthcoming with samples of and details about the sensor in response to our inquiries initially (we haven't checked for several years, though; they initially were selling entire custom camera setups for studio use, not just the sensor) and the sensor isn't truly electronically shuttered (it's a CMOS rolling shutter design and is typically used with a mechanical shutter). Being CMOS, it is likely also susceptible to single-event latchup (CCDs aren't because they are NMOS). I don't know what the actual noise performance of these sensors is (all they appear willing to say is that they're "low noise"). And there really hasn't been a lot of commercial acceptance of this sensor technology yet; the Sigma SD9/10 is the only DSLR available that uses the sensor that I know of, and I've never seen one. For MSL, we believed that a CCD system was a better choice.

It's also not clear that you "triple" your resolution. A Bayer filter has two green and one each of red/blue for each 2x2 pixel group. With proper resampling, there is a fairly small impact on effective luminance resolution except for pathologically-color-patterned scenes. The MSL cameras will use good resampling and then compress in YUV space, so color artifacting should be pretty minimal for RGB color images. For narrowband color, we may take a resolution hit for some bands, but this mode is mostly only useful for fairly gross spatial characterization of color differences anyway.