Reprocessing Historical Images, Looking for REALLY big challenges? Options

[tedstryk]

Don, excellent work!

They really were remarkable for their time. The fact that they were 8-bit cameras that selectively converted the used portion of the dynamic range to 6 bits for transmission results in using the 6 available bits very efficiently.

[edstrick]

The Viking lander cameras had two smallish design defects that might have improved their results, one significantly, one a larger amount.

The cameras scanned the scene with silicon photodiodes looking up at a nodding mirror which rotated in azimuth. Since ONE detector takes an entire picture or color channel, the image quality is extremely stable photometrically. However, the detector, during brief moment when it was looking at nothing (black) inside the camera housing, would periodically take a "dark current" measurement. Every 64 scans (I think) in monochrome scans, every 64/3, I think) in 3-detector color scans. This reset the dark count, but the short sample time resulted in the sampled value jumping around noticiably in low-light imaging, putting annoying vertical bands in the image that changed each dark count resampling. The defect was that the count wasn't measured and transmitted with engineering data, preventing easy and almost perfect removal of the stripe pattern.

The other defect, more significant, was that the camera had absolutely no automatic exposure adjustment capability. You pre-selected camera gain and zero-level offset, had the commands coded and sent up to the spacecraft, and waited some days or more after the exposure was set to see if you were right. Normally, it worked OK, but during the dust storm season for both landers and when northern winter ice clouds were varying illumination at the surface as cold fronts passed for Viking 2 (I think), they'd have illumination drop and badly underexposed pics, they'd adjust exposures, have the illumination brighten, and the commanded image would be unusably saturated in brighter areas or everything except actual shadows. That caused real science data loss.

Still....., Marvelous cameras.

[Ian R]

Fascinating stuff Ed!

[djellison]

I'd love to know what the very first pancam - the rotating push-broom affair first proposed by Sqyures et.al. for Pathfinder - was really like - what the results might have been like.

Doug

[DDAVIS]

Although 'pushbroom' cameras are grand for capturing static views I would ideally prefer the approach of capturing portions of the visual fields instantaneously in RGB. If, for instance, the first Viking surface photo could have managed a 'snapshot' rather than a gradually scanned image the white streaks at the start of the image would have appeared as the edge of a small dust cloud kicked up during the landing. Pathfinder, building up wide panoramas from many small frames, captured each view as a 'snapshot' which included dust devils in some instances.
The approach of carrying color video cameras on board future rovers is enormously exciting. The PR value of being able to capture a landing and dust devils, time lapse clouds, guysers etc. is enormous. If anything the trend of evolution of 'added value' imaging from Mars should move toward full 30 frame per second full HD video. The data transmission would be daunting but the technological trends seem to be steadily improving that aspect of the problem. There are things you need good 'temporal' and visual resolution to study.

[monitorlizard]

"full 30 frame per second full HD video"

I've always wondered if you could generate intermediate images by computer, using pairs of sequential images from a rover, to create something like 30 frame/second video without the spacecraft having to transmit 30 frames/second. Something to lessen the data rate on very valuable spacecraft transmitting resources. If you transmitted, say, 15 frames/second and generated an image between each set of sequential frames, you'd have 30 frame/second video but only use half as much spacecraft bandwidth.

It would be computationally intensive, of course, but the computer resources on the ground would be so much greater than rover transmitting resources. I guess it would all depend on how much value a smoother, longer video sequence would add to public outreach.

[djellison]

Well Mastcam is something like 1280 x 720 up to, I think, 5fps. To be honest, I can't imagine 30fps on Mars being THAT usefull. Be it your rover moving, or something happening on the surface of Mars, not much happens in a 30th of a second

Not saying it wouldn't be awesome - but 5fps HDTV res is enough for me for now (and enough for MRO relay as well I'd guess )

Doug

[helvick]

The MSL Mastcam's 720p video resolution mode @10fps would generate a raw data rate of 100Mbps+ depending on how many bits per pixel it's built to produce. I suspect that since it uses a Bayer filter pattern it will actually be capturing pixels at 25% of that rate spread across the Bayer 2G:1B:1R range and interpolating back to full resolution before compression. In the end the video camera and in situ processing could produce very high quality video (720p HD level) for a cost of around 0.5 to 1 megabytes of data per second of MPEG-2 video.

MARDI is definitely also going to be capturing to MPEG-2 color using a similar 720p capture mode but at 5fps ("500 images in 100 seconds")

Can't wait.

[DDAVIS]

... on the surface of Mars, not much happens in a 30th of a second

In general you are right, indeed obtaining color time lapse sequences over extended periods of lighting changes and clouds could be considered a more efficent use of such a system. High 'temporal resolution', such as is obtained bt 30FPS would be important in a minority of occasions such as the landing itself, and for dust devils or guysers, future sample return launches and such. I have shot a lot of dust devil videos and I often see motion near the base which is just captured in 30fps, which if obtained on Mars could aid in studying their dymanics. BTW, The difference between 24 and 30 FPS is obvious when seeing swift dynamic things like water sprays and splashes, whirling dust, and flames. Even NTSC resolution 30 FPS video would be useful, especially if it is compressed less. Interpolation between limited frames per second is a lot of work to perform well if there is something moving swiftly, with hand retouching sometimes required. Even things like shadows creeping along rocky surfaces are a pain using such methods.
The upcoming 10 FPS Martian surface video will be similar in frame rate to the Apollo lunar surface videos, and of incomparably better quality!

Don

[ElkGroveDan]

MARDI is definitely also going to be capturing to MPEG-2 color

But only on Tuesdays.

[DDAVIS]

Well, I felt compelled to complete the Viking one color coverage so here is the camera one view, a work in progress. I am working on merging the two, and later adding detail with high res grayscale images. I really wish a DEM of the lander region existed, however years ago I drew a meter grid on acetate overlays over the large wall map editions of the panoramas using the stereo derived partial contour maps in the Viking Lander Atlas of Mars. I can create a 3D model from such material however it is quite laborious. There should be an automated way to do that.

Don

[mcaplinger]

I really wish a DEM of the lander region existed...

http://nssdc.gsfc.nasa.gov/database/Master...g?ds=PSPA-00335

This work was done, I believe at the Stanford AI Lab. It's digital contour data and would have to be interpolated to form a DEM. The whole dataset could be reprocessed using more modern techniques, I guess, but I'm not sure if anyone considers it that scientifically useful.

[stevesliva]

But only on Tuesdays.

rimshot.

Don, a beautiful image! The horizon is so annoyingly close, but that's not your fault...

[edstrick]

"...I can't imagine 30fps on Mars being THAT usefull. ..."

I'm sitting here imagining the MER rovers doing a test drive like one of the Apollo 16 astronauts did with the lunar rover while the other astronaut filmed him with the 16 mm camera set to a more or less normal film speed... rover bouncing along, maybe all wheels off the ground, throwing roostertails of dust in the air...

(Air... what air?.. oh. on the moon... YEAH. RIGHT!).

[4th rock from th...]

Don, that's beautiful! Thanks for sharing. I really like the overall color, very nice ballance between realistic and visually striking. Just one thought, the sky darkening introduced by the cameras on the top of the image is somewhat strange, now that we have good sky coverage from MER mosaics. What's your take on that? Did you correct the Viking images?