Meanwhile...back on- topic...

Any thoughts here? A noticeable darkening of images on Sol 1785 between 11.42 and 11.48 local time.
All navcam L0. 5 frames, the last two seem to darken. Illusion? Passing cloud?


Astro0

We see similar things from time to time. I'd guess it has to do with the fact that these are subframes. Probably some portion of the rover was in the field of view, but has been cropped away now so you can't see it. As the series of images proceeded, the sun moved and by the end of the sequence there was perhaps a bright glint of sunlight off the rover part. That resulted in a shorter exposure for the final frames.

This can't be exactly right, since the jpegs we see are supposed to be autostretched. But if you look closely at the bottom of these frames, you can see a few rows of white pixels. So the autostretching isn't working properly - it's biased by those bright rows.

The real question is where do those white rows come from. I still suspect my description above is more or less right, and that the rows of white pixels are increasing in absolute brightness (before stretching) as the sequence proceeds, which would explain the darkened later frames.

Astro0, is this a readiometrically corrected image? If not, the autoexposure might be affected by areas outside the subframe. If Im not mistaken the autoexposure works on the entire frame, not just the subframe.

Paolo

The images are just straight off the exploratorium site as delivered there. Stretched yes, altered in any other way(?), no.
I thought about it being some kind of reflective surface elsewhere lightening the earlier frames, but the frames only cover a period of 6 minutes - could there be that significant a change in that time? The dip in light levels seems consistent across the frame too.
I'm not suggesting anything here, just curious? The more technically minded among you will know the answer I'm sure. Fredk's response sounds right.

Astro0

The one event that would explain a real "darkening event," it seems to me, would be the very close passage of a dust devil. (Or at least the passage of a dust devil between Spirit's immediate surroundings and the Sun -- i.e., in the DD's shadow.)

Anyone check the solar panels since that observation? Or the watt hours?

-the other Doug

Apparently fredk had already suspected that, but, that is good to know. It hadn't occurred to me that the auto-stretch algorithm might be automatically applied to the full frame of a subframe image.

O'Doug: I suppose it could be a close encounter of one kind or another, but is this a fun place to hang out, or what?

I saw two earlier images from sols 1511 and 1517 come down today. For a moment I thought, "OMG, Spirit is almost on top of HP again.

I think he means the exposure duration that happens on the rover, not the brightness stretching that is applied to the JPG images we get online, CR. But then it seems like the brightness stretching should tend to hide that, so I don't know.

Yes, I mean the exposure setting is determined on board based on the full frame image. I have no idea what algorithm is used to do the 12->8 bit conversion. If it is just a scaling factor, it would not compensate correctly and if it is a full linear conversion, the algorithm might get fooled sometimes. Granted, the difference presented by Astro0 is quite large and am not sure if it can be explained this way. Mine is just a guess. I haven't heard anybody from Power screaming in joy or in terror. That might be a more reliable data point.

Paolo

I love these "rock pile" shots... makes you want to reach into the screen and just grab one, doesn't it?

Hit the nail on the head. The white rows are related to electronics noise generated by the way "fast dust devil" subframes are generated. The rover uses the capability of the readout hardware to generate a subframe that is transfered into memory, rather than transferring a full frame and then cropping. This is faster, but has side affects at the readout end of the frame. The side affects are temperature dependent, and so increase with time as the electronics are used. The actual scene brightened very slightly over the sequence of images. Most of the frame is not affected, so cropping the bottom couple rows masks the effect (an old example is here: http://www.lpl.arizona.edu/~lemmon/mer_dd/...nced_1239a.gif).

These images didn't use the 12-->8 bit conversion (I think). So the electronics generated 12 bits, these were downlinked, and then a linear scaling was done (tossing the brightest 1% and darkest 1% of pixels, I suspect). If the 12-->8 scaling had been done on board, that would be non-linear--sort of like square root encoding with the most commonly used look up table. Also, since these images use hardware subframing, the autoexpose software never saw the rest of the frame--but it normally would. It was happy with a consistent exposure, allowing the fastest image acquisition.

Well, that sounds pretty authoritative to me

Very good to hear all the subframing details; Thanks Deimos!

A few things worth mentioning in recent images. First, look not just at the rover deck, but at the ground in the foreground of this sol 1784 image:

http://marsrovers.jpl.nasa.gov/gallery/all...89P0665R0M1.JPG

To me it appears more mottled than usual (there's a high-contrast mix of lighter and darker patches). I think we're seeing the effect of wind clearing some patches on the ground. The ground in that new navcam reminds me of how the ground has looked after previous cleaning events, for example see this image from sol 424:

http://marsrovers.jpl.nasa.gov/gallery/all...B3P1705R0M1.JPG

or this image from 1250:

http://marsrovers.jpl.nasa.gov/gallery/all...CMP1665R0M1.JPG

Either I'm starting to get a pretty good feel for how the Gusev ground should look, or I've been staring at Mars for too long!

And it looks like we've dug up some more of the tasty white stuff:

http://marsrovers.jpl.nasa.gov/gallery/all...89P2598L7M1.JPG

Finally got my computer to crunch out the full Bonestell pan in 3D colour.

Enjoy your stroll around - and try not to trip over any rocks.



James

Thanks so much, James, that's absolutely stunning! It's also great to see real 3D, not the odd constant-depth anaglyphs we've seen recently from official channels.

What do you mean? I've made a lot of them, viewed a lot of them, and I don't know what that means.