No Problemo, we are all in the Arm Waving phase now. I saw very distinctive features and snapped some Puzzle Pieces together and came up with this as _an_ initial interpretation. I've since tweaked it, and will continue to do so as we get more puzzle pieces. I do like your "coriolis ash cloud" idea, and your thoughtful write-up. The thing to remember is that we have milli-gal gravity and a micro-bar atmosphere and things can and do move at cm/second speeds, super sol-mo by our standards.

And look at the "duneoids" on the North Polar Plain... :blink

Check your Terracentric Baggage at the door, kids, this is an Odd New World.

--Bill

... ok, after considering a handful more approaches, I think, there is at least one mechanism which cannot be neglected: Erosion by solar wind.
Think of an Fe⁷⁺ ion of the solar wind impacting a dust grain. Let the dust grain replace one of the seven lost electrons by each of three neighboring grains below, and keep one electron lost, such that the grain remains charged with 1+ (equals about 1.6e-19 Coulomb), like the three grains below; three electrons come from other neighbors, the forces of which may be neglected for simplicity.
Now calculate (approximately) the repulsive electrostatic force dependent of the grain size (assume spherical grains of density 1 g/cm³ for simplicity), and compare it with gravity.

I'll leave this as an exercise, and try to find time to solve it, if noone else provides a solution until tomorrow.

Considering two grains, one on top of the other, each charged 1e, and find the ratio of electrostatic force to gravity:

Fe/Fg = 6*k*(e^2)/(pi*rho*g*(d^5))

Where Fe and Fg are the electrostatic and gravity forces, e is the charge, rho is the grain density, g is the acceleration due to gravity, and d is the grain diameter. Here are the values:

k = 8.9876e9 [N m^2 / C^2]
e = 1.6e-19 C
rho = 1000 kg/m^3 (1 g/cm^3)
g = 1.0720e-4 m/s^2 <-- from g = mu/(R^2), mu = 670 m^3/s^2, R = 2500 m

The ratio of forces is 1 at a grain diameter of 5.28 um, and rises enormously with decreasing size due to the d^(-5) proportionality -- at 1 um the ratio is up to about 4000 (see the attached log-log plot).

So any particles below 5 microns would start to be repulsed based on these assumptions, and it could be possible that particles even smaller could be accelerated along big enough trajectories to make some notable macroscopic features.

Edit: I've also added a figure showing the height at which grains would be at equilibrium. Although the force ratio is 4000 for 1 micron grains, they would only be in equilibrium at about 0.1 mm height -- whereas grains of about 100 nm would be in equilibrium at around 100 m. This is not to suggest particles would end up actually hovering at these heights, but the equilibrium height gives a sense of the potential energy gained by the interaction we're analyzing.

The difference though is that the spot I identified is #1 just exactly where we know Philae was at 5 minutes after 1st touchdown (assuming the 5 minutes after touchdown photo pre-set 'rumor' is a true one) but, more importantly, #2 matches not on just a couple of major black spots but literally on two dozen features all around the image.

I first matched it by just a 3-4 of the most obvious spots. And you can indeed find a number of more-or-less matches for 3-4 amorphous blobs. But when I uncovered the rock complex near the lower LH corner, which had been covered up in my original base photo, that matched, too (and rather exactly--it's a very distinctive rock formation). And when I then tracked down a higher-res photo of the area, the number of clearly matching features went way up. Both of those things make me think this is a real match and not just hyperactive pattern matching of random blobs.

In addition, the places that don't match can pretty much be explained by different sun angles between the various photos--some shadows become quite different, others disappear entirely, others appear that were not there before, from one photo to the other just because of the differing sun angle. See the attachment below showing sun direction in the base photo and CIVA photo overlay.

See attached:

1. A new animated gif showing the deconvoluted CIVA photo overlaying the base photo, with a higher resolution base photo and two dozen points of similarity indicated. You have to click on the attachment to see the animation, then zoom in quite a bit to really see the detail. As I mention above, I'm even closer to 100% certainty on this now, after examining it even more carefully . . .

2. Location of the post-first-impact CIVA photo shown in context of the higher-resolution base photo (photo location shown as red rectangle, Philae 1st impact location shown as red dot)

3. Approx. location of Philae at 5 minutes after 1st impact (white dot) along with the location of the CIVA photo (red rectangle), as visualized in STK.

4. Flat/nonanimated version of the '24 points of similarity' animated gif.

5. Base photo & CIVA overlay, indicating sun direction. Sun elevation for the CIVA photo is about 45 degrees coming from the direction dhown; for the base photo the sun appears to come from the direction shown (based on shadows) but I'm not certain of sun elevation (the image was taken on 14 Sept 2014 but I'm not sure the exact time that day).

Does the angular size of the CIVA image projected onto the comet model match the expected FOV of the camera?

Great, thanks a lot!

For the potential energies (how high would a grain jump?) I got a slightly different result:
h = 3 k e² / (8 pi r^4 rho g)
by using m g h, with h the height, as the potential energy in a homogenious field of gravity, and k e² / 2r for the electric potential energy between point charges e of distance 2r, leading to a (dynamic) equilibrium at about 4.1 mm height for a 1 um grain (r = 0.5 um), and 41 m height for a 100 nm grain (r = 50 nm).
Within an order-of-magnitude estimate enough to move fine dust even in the absence of sublimation.


Details about actual ionic charge states of solar wind, see e.g. this paper.

In a short article in a german newspaper from January 2nd, Holger Sierks the PI of the OSIRIS camera system says that Philae could not be found in the images taken in the 20 km orbit.


"Unfortunately, we could not see anything from the lander. Otherwise it would be a nice present for christmas eve."

According to one seemingly reliable source, the CIVA cameras have a 70 degree field of view. (Most sources say 6 cameras X 60 degrees field of view to cover 360 degree panorama - but I find it hard to believe they don't have any overlap at all? The source is MCSE, which claims to have developed the CIVA cameras, so I assume they know what they are talking about.)

Attached is the deconvoluted CIVA image, plus a screen shot of the scene visualized in STK with a 70 degree field of view.

Now, it turns out that the original CIVA image has a slightly wider field of view than the deconvoluted image--something got trimmed in the deconvolution process. The deconvoluted imaged shows maybe 60 degrees FOV, assuming it started with 70 degrees. But all that means is that Philae must have been a bit higher than illustrated here--or perhaps our 3D model is a bit off, or perhaps the CIVA camera's geometry is a bit different the STK. We don't have Philae's height above the surface constrained very tightly at this point (at least, I don't).

Point is, it is very close and well within the constraints we have at hand--all of which have a fair bit of play in them.

The difference between these values and those in the previous plot is expected since this is a different (and more sensible) analysis compared to my previous post. Here we are looking at the peak height for a particle sent on a trajectory where the initial velocity is defined by the electrostatic potential energy. Previously, I was looking at the height at which the electrostatic force and gravity forces are equal -- both analyses suffer from an assumed constant value for g, but nevertheless it looks like this erosion process would be process.

Ah, ok I see, where the discrepancy comes from:
0.001 mm = 1 um = 1,000 nm = 1,000,000 pm.
1e-10 m = 100 pm = 0.1 nm.

Argh, stupid mistake on my part! 1 nm = 1e-9 m...

You've put a lot of consideration into the comparison, and I am not saying it is incorrect (your gifs are very convincing and I could easily believe this is the correct location). However, a lot of assumptions have gone into the analysis and stating things like "exactly where we know Philae was at 5 minutes after 1st touchdown" and "I am 99.8% certain, perhaps a little more than 99.8%" makes me cringe! We don't know exactly where Philae was anytime after touchdown -- our best guess about the trajectory is based on aligning Rosetta and Sun vectors using the images at 15:35 and 15:43.

If you are relying on Malmer's model and a presumed trajectory that simply connects waypoints with spline trajectories then things get even more uncertain. Plus, you are not transforming the image at all but instead assuming it is only scaled and rotated, meanwhile the (also unknown) direction of the camera will likely not be directly downwards.

Note this is just constructive criticism, I really hate to sound like I'm bursting your bubble! After all, you might be spot on -- just wouldn't say 99.8% sure...

In ESA Blog: How & Where is Philae there is an OSIRIS montage. At the bottom of the image is a possible Philae (green arrow).
Click to view attachment
Roughly scaling the Philae Landing & Bounce OSIRIS montage to the same size by eye from the terrain and then comparing it's Philae at 15:43 (green box) the size seems OK.
Click to view attachment

To narrow down when it was taken, compare the shadows (left half ) with the OSIRIS Bounce montage (right half), they seem to be longer and at a bit of a different angle, also Rosetta has moved - so this means later than 15:43.
Click to view attachment

Nice catch, it would be great if we could figure out when that image was taken! Here is a rough attempt at projecting the mosaic onto a NavCam image without so many shadows. There are a couple of rocks jutting up that are most likely the bright spots seen nearby the potential Philae location -- I can't pick out anything that could be mistaken for the lander after assuming the obvious rocks are the other bright spots.

@Brian Lynch , your overlay is well placed and there does not seem to be anything there that could eliminate the Philae candidate.

The same area in cross eye stereo from NAVCAM 20141030
Click to view attachment

On the OSIRIS bounce montage, is it possible to estimate the height of 15:43 Post Bounce Philae by size ( relative to the other known heights at 15:14, 19, & 23 ) ?

I hate to be that guy, but doesn't that assume it bounced in the opposite direction to what has been established already?

Not really, it is easy to look at the OSIRIS landing sequence and assume that Philae is flying above the locations in the image, but in fact if you were to drop lines directly down to the ground you would find it is not even close. Attached is a figure I posted before but with the details emphasized a bit more:

- the blue line is Philae's trajectory (from SPICE data) up until the first touchdown
- magenta lines are drawn from Rosetta to each of the locations where Philae is seen in the OSIRIS images (red dots)
- the intersections of the magenta lines and the blue line are the actual positions of Philae at each of those times (blue dots)
- post-touchdown locations are treated similarly but with green lines and green dots (no SPICE data so it requires more analysis to determine the actual Philae positions, but this has been discussed previously in this thread based on Sun vectors and the shadow in the image at 15:35:32)

With the scene oriented so we're looking straight down at the image, the blue dots are the locations on the terrain over which Philae is really flying at each of the times in the OSIRIS sequence -- the huge difference is due to the position of Rosetta, who is looking down at the landing from a significant offset angle. As Philae climbs during the bounce trajectory, we should expect to see it appear lower and lower in the image -- although the comet's rotation and Rosetta's motion makes it hard to guess if this location makes sense.

If we figure out the time that this OSIRIS image was taken then I can look at the vector from Rosetta to the potential Philae location and see if it aligns with the expected bounce trajectory.


This might not be too hard, although it is a bit involved since you have to account for what I said above (ie. Philae's size will be related to its distance to Rosetta, not the ground).

On the other hand, this document suggests the side of the square is 60 degrees, and the 70 degree measurement is partway to the corners. There are some diagrams starting on page 8. To some extent there would be overlap since the camera field centers are pointing a bit below the horizontal (ranging from 15 to 25 degrees).

4th rock from the sun posted a great diagram of the CIVA FOVs here.

Where did you find the OSIRIS montage you posted with the green arrow? The blog post you linked to has only a tiny version, and the version on the image site has only about half the resolution of the one you posted. Your version really has more resolution, it's not just upsampled.

@fredk, apologies the high res version is from http://sci.esa.int/rosetta/54944-searching-for-philae/ - on the right "Also available as" are the hi-res png and jpg

http://sci.esa.int/science-e-media/img/a0/...ite_montage.jpg

The sci.esa.int always has the best resolutions - http://sci.esa.int/multimedia-gallery/3091...t=1012&cl=2

http://sci.esa.int/rosetta/54971-osiris-sp...ross-the-comet/ - has a 1849x1269 version of the landing montage and details of resolution and times

In trying to figure out when the "How & Where is Philae" OSIRIS montage was taken I have observed that some of it is very similar to some of the "Philae Landing & Bounce" OSIRIS montage and some of it is very different.

Peering closer at parts shadowed identically I observed this intruiging difference (1st is from Landing & Bounce montage ; 2nd is from How & Where montage )
Click to view attachmentClick to view attachment

Could it be a shadow ?

Making a gif demonstrates two similarly moving objects or perhaps 1 and its shadow.
Click to view attachment

If the "How & Where" image is darkened the landing craters become clearer - giving a different angle than the "Landing & Bounce" montage.
Click to view attachment

This gif compares the post bounce Philae insert with 15:43 Philae-in-flight and shows the image was taken before Philae got there - so this narrow strip is between touchdown and 15:43 which means Philae should be somewhere between the two.

Click to view attachment Click to view attachment

In fact, there is even the possibility this is the shadow for Philae at the 15:43 spotting -- although it conflicts with my and Malmer's estimates of a shallow bounce trajectory (I think we both based that on the shadow in the 15:35:32 image).

Searching for shadows in the OSIRIS landing images was discussed here, where the figures give an idea of the sun direction (yellow) at the various times Philae was spotted. Drawing a line from this new shadow in the sun direction may in fact intercept the green line that defines the line-of-sight from Rosetta to Philae at 15:43, but it would be quite high above the surface. However, I wouldn't rule it out given the uncertainty in the shape model and how I fit it to the SPICE data (ie. it could be that the terrain results in the shadow being cast far away even though Philae is on a shallow trajectory).

The fact there are two features that seem to move together suggests it might just be two artifacts in the image -- maybe we can spot them in other OSIRIS images? I can't imagine one of those objects is Philae since it would be much brighter, and there really shouldn't be anything else moving around (unless you spotted some kind of comet activity!).


Is that conclusion based on the change in shadows?

They are not quite in sync the lower feature moves a little bit more.
This could be consistent with the lower being the shadow of the upper moving feature, cast on an incline.
I have seen "snowballs"/particles in the air but never something dark - might be an imaging artefact, or perhaps something Philae broke off a cliff ?


Yes. I am comparing the shadows to the 15:43 inset. Click to view attachment

The shadows in the part of "How & Where is Philae" strip with the landing site seem shorter than the 15:43 insert ( based on 'rock' landmarks rather than apparent size ).

But I realise I am assuming the shadows are getting longer as the day goes on, and now I am not so certain whether this is 67P's morning or afternoon.

The earlier / later could be also be established by Rosetta's direction - the gif comparison shows that Rosetta is looking more directly down on the flat-top boulder, in the part from "How & Where is Philae" than the 15:43 insert - so Rosetta has moved toward Agilka - which would make this after 15:43.

The FOV depends on if you consider the circle or the rectangle of the viewed area: the rectangle is inside the circle, so the circle has a 70° FOV and the rectangle (the actual picture) has a 60° FOV.

The picture overlapping is quite complex because cameras do not point horizontally but are slighlty pointed downward (15° all cameras but 6 and 7 "stereo couple", 25° downward).

Orientation:
http://naif.jpl.nasa.gov/pub/naif/ROSETTA/...ions/ROS_V23.TF

FOV:
http://naif.jpl.nasa.gov/pub/naif/ROSETTA/...ROS_CIVA_V10.TI

This is my attempt of building a 3d model of Philae and CIVA Fields Of View in Trimble Sketchup:
http://lc84.altervista.org/rosetta/CIVA-3d-model.skp


Images are placed "randomly" in the model because I don't know actual distance of ground from cameras, but angles and FOVs are real and based on official kernel files:
Orientation:
http://naif.jpl.nasa.gov/pub/naif/ROSETTA/...ions/ROS_V23.TF

FOV:
http://naif.jpl.nasa.gov/pub/naif/ROSETTA/...ROS_CIVA_V10.TI


This allows better understanding of overlapping than 2d image:

Original:


Processed:


I wonder if the second image of the stereo pair has never been released.

I found a third moving dot in the images, and shadows direction does not allow supposing one dot is a shadow.

Great work.
There's some overlap between at least two images, so you can place those at a reasonably correct distance for the points closer to the lander.

ADMIN EDIT: There is no need to "quote" a comment that is directly before yours. This has been raised many times before.

Maybe there's something wrong in my model, because I can't obtain the image overlapping unless I go dozens of meters away from lander. I don't know if it depends on the only unknown data in the model: exact distances of cameras from lander edges (I estimated them)

I would think the inferred distances of overlap features would be quite sensitive to actual camera pointing directions. The actual directions will differ from the nominal ones. If you change the pointing of one overlapping frame relative to the other by, say, one degree, how much does that change the inferred distances?

This page has a CIVA image with overlap with CIVA 3 & 4 is it merely a skewed CIVA 3 or actual parallax ?
Click to view attachment
Click to view attachment

The image on that greek page just looks like the "welcome to a comet" image.

I see only a very few features in common between the two frames - those I've circled in black here:
Click to view attachment
The dark region circled in white is a possibility, but I think unlikely since the parallax is wrong: if the dark region was closer than the lighter region, it should shift in the opposite direction than it has.

But clearly there is real parallax here and not just image distortion/shear, since there are no obvious matches in the area just above the ellipses. So for the "welcome to a comet" stitch, those upper areas must have been spliced incorrectly. Of course nothing better could've been done with the images at hand.

ADMIN EDIT: There is no need to "quote" a comment that is directly before yours. This has been raised many times before.


If you match the images carefully you can actually measure a couple of distances there. there is a lot of depth there. And slightly over the ellipses there is an overhanging rock that is close and only visible in the right image. so it blocks further stereo coverage.

I'm experimenting with my 3d model of the cameras FOVs, but I'm really in trouble with getting a 60° large rectangle INSIDE a 70° large circle, there's something wrong here... maybe even in the SPICE kernel?!?
I must check with some trigonometry and geometry if it is actually possible to draw such a rectangle inside such a circle.

My math appears to say that:
- if 60° rectangular FOV was right, circular FOV would be 78°, not 70°
- if 70° circular FOV was right, rectangular FOV would be 52°, not 60°



Also manual measurement appears to confirm such results.

Unfortunately this does not match with SPICE kernel.

What am I doing wrong?

Yes, there seems to be some overlap at the bottom edge of the frames. Looks correct, since the cameras are tiled down.
Perhaps the FOV uncertainties come from that. Could be that the number given is the horizontal FOV at the middle of the frame...

I've noticed a very strong vignetting type effect in the corners of several of the CIVA images published. In fact it rather closely matches your image if you compare the 70 degree circle to the 60 degree square.

I wonder if 70 degrees doesn't amount to the max usable field of the optics. The square optical detector cuts off that max usable optical field in some places and exceeds in others.

Note vignetting in corners of these CIVA images:

http://sci.esa.int/rosetta/54965-frame-fro...iva-p-camera-2/
http://www.bbc.com/news/science-environment-30524429
http://sci.esa.int/rosetta/54966-frame-fro...iva-p-camera-3/
http://sci.esa.int/rosetta/54967-frame-fro...iva-p-camera-4/

I'm assuming that for images that don't show the vignetting effect (or only show it in some corners), either the effect is hidden because that portion of the image is all black, or the image has been cropped. The CIVA camera appears to return 1024x1024 images, meaning that some of the above have been cropped and/or scaled.

On edit: Scalbers makes that point upthread: "this document suggests the side of the square is 60 degrees, and the 70 degree measurement is partway to the corners." See diagrams starting page 9. Once you understand how it works you can see they have carried through that type of diagram on some of the other diagrams linked upthread--even in the SPICE ASCII diagrams. So I think your diagram does pretty much sum up the situation.

P.S.--Some truly amazing detective work from a number of people tracking down details of the CIVA cameras upthread. Wow!

I was tipping more in the direction of 'possible camera artifact' for those little dots, but finding a 3rd one really does push it much more in the direction of some kind of real objects flying above the surface. For one thing, Object #3 seems to have a fairly distinctive long, thin shape that appears to have rotated between the two images. To me, Objects #1 & #2 look more like possible 'bad pixels' on the camera but Object #3 doesn't look like a bad pixel at all, due to its shape, and particularly its changing shape from image to image.

I spent some time in STK 'observing' this area of the comet from the viewpoint of Rosetta and then comparing what I saw there to what I see in these two images, in the area of the three objects. A few thoughts based on that:

• "Montage" appears to be first in time and "Mosaic" a bit later in time.
• That means the objects are moving down and to the left in terms of the image
• That appears to mean the objects are moving close to due west, ie opposite direction of rotation. So, perhaps (?) a similar scenario to that we've been discussing upthread, where debris is spewed from vents; if they are expelled with any upwards velocity they tend to move westward relative to the surface of the comet because of the fast rotation rate of the comet surface and the fact that an object rotates at a slower speed than the surface the further from the axis of rotation it rises.
• The time between the two photos appears to be in the range of approx. 3 minutes to 5 minutes, maybe a bit more or less. It is definitely more than, say, 1 minute and less than, say, 15 minutes. That is just my impression based on stepping the view in STK by various time periods and looking; I'm not sure exactly what you could measure to nail this down more precisely.
• If you assume all of the above is true and then trace the direction of the objects backwards following a straight line path, the general direction is back across the small crater where we see Philae in the 15:43 photograph. To my eye, the objects might have passed about through the area covered in that montage by the 15:43 inset. (See Note below about apparent linear movement of the objects.)
• Rosetta's location is almost directly above the objects at the time this photo was taken. It is just a hair to the northeast at the moment of Philae's touchdown (see attached photo showing direction from Philae to Rosetta at that moment; these objects are just SW of that Philae touchdown point so Rosetta will be just a bit more NE of them); at 16:00 it's nearly overhead (just a hair to the north) and at 16:30 it is a bit to the NW. All are pretty close to vertical, though, presumably because the Rosetta team planned to have Rosetta above the Philae landing point a the moment of touchdown.

Those are all just impressions, further verification required and could easily be wrong. But an interesting starting point.

Further and more hypothetical guesses and speculations would be:

• If the three objects were expelled with slight different velocities in slightly different directions, that (mostly) explains the spread in distance we see among the three objects; whereas they mutual westward direction would be (mostly) due to the comet's rotation. Imagine the three objects were expelled generally upwards in a shotgun pattern from a point on the surface; now that origin point is rotating out from under the objects it expelled.
  
  If this is true, the fact that the three objects are moving together in a very nearly westward direction is (mostly) explained by the differential rotation of the surface vs objects rising above the surface, because the objects rotate at a slower angular rate the higher they rise. The objects have similar vertical velocities so they end up moving westward relative to the surface close together and with similar velocity. They have a slight spread because they had different initial velocities & direction. Further amounts of eastward/westward spread could be explained by the fact that objects had varying amounts of vertical velocity, resulting in differing amounts of this differential rotation effect. But overall, their own horizontal speed relative to each other is fairly slow compared with the fairly rapid speed of the rotational differential with the surface.
  
  The end result of all this is, the objects appear to be moving across the surface together in a group that relatively slowly spreads apart over time.
  
  
• It is very interesting that the area where the three objects are has so many long, thin, linear features that are lined up very close to the direction the objects are moving--and, perhaps not coincidentally, are lined up very closely with the direction of rotation. (They might also be in the direction of a gravitational fall line here?)
  
  
• We've previously noted the 'dunes' that seem to be oriented primarily perpendicular to the direction of rotation. There are some of those in this area, but also a number of rock bulges that are also oriented approx. perpendicular to the direction of rotation. I am wondering if both these & the dune features couldn't be created by buckling due to stress induced by differential rotation rates. A feature that extends from, say, 2300 m to 2500 m from the axis of rotation is going to have a pretty big velocity differential between top & bottom; that would induce to pretty large internal forces that could lead to buckling predominantly in the direction perpendicular to the rotation. </end massive speculation>

Again, all this is guess/speculation, but it gives some really interesting possibilities that could be checked out more rigorously.

As Brian notes, these objects could be almost any elevation above 67P--1 meter, 10 meters, 100 meters, 1000 meters--maybe even 3000 or 5000 meters. I think we can rule out some very low elevations because we would see shadows. The fact that they are traveling close together in almost the same direction/speed suggests to me that they might have been ejected together fairly recently, which argues for a relatively lower altitude.

It would be a very interesting project to try to project these three objects back in time to try to determine when & where they were ejected and if they were ejected together. For example, if you just follow the straight line of each point backwards in time, do they converge to a point or near-point? Or with more sophisticated analysis of speed, direction, rotation, gravity, etc etc etc do they converge back to one single starting point and time? The fact that they are traveling so close to together in space, time, and velocity suggests to me that a common point of origin is possible--or perhaps even likely?

They are a really interesting find regardless, but it could be even more interesting if (for example) it appears they were ejected together from a point along the 'fracture vent' that Bill identifies here. The 'fracture vent' does appear to be in the line of the objects trajectory if projected backwards. (!)


Note regarding apparent straightline motion of objects as seen in successive Rosetta images: Rosetta is moving in (very close to) a straight line in the plane of the photograph and so are any objects moving across the surface. Both Rosetta & any other moving objects will have orbital motion that implies a curve in the vertical direction WRT to the 67p gravity vector at that point, but due to the position & orientation of Rosetta, that vertical motion of both Rosetta & any object it is photographing will be (almost all) perpendicular to the plane of the photograph. The result is (in the plane of the photograph) that Rosetta has (very nearly) linear motion that is composed with the object's (very nearly) linear motion. Geometry says that when you project one point moving linearly through another point moving linearly onto a plane, the result is a third line.

End result: The object will appear to move in a straight line in a sequence of Rosetta photographs if it is in fact moving in a straight line across the surface (straight line meaning straight in the horizontal direction relative to the surface; in the vertical direction WRT to the gravity vector it will be curved, of course). We've observed this with Philae immediately before & after touchdown, where you can take the Rosetta photos of Philae from before 1st touchdown draw a straight line from one Philae loation to the other before. Similarly for the touchdown point & the two photos of Philae just after 1st touchdown--they are aligned pretty closely.

It's worth pointing out again that if the object is very high above the surface, its apparent position above the surface as shown in the Rosetta photographs can be quite deceiving--it might be quite a long distance from the object it appears (in the photograph) to be very close to and directly above.

See attached two views of Rosetta's position WRT to 67P center of rotation during the period 12 Nov 2014 09:20 to 12 Nov 2014 19:30, from SPICE data and processed via MATLAB. Rosetta made maneuvers just before and after this period but during this period there are no maneuvers. One view shows Rosetta from the 'top', the same approximate direction it is aiming to take photographs, and you can see its motion projected onto this plane is very nearly linear. The other graph is a 'side' view of Rosetta's motion and you can see the effect of 67P's gravity at work.

Yes, this document had been already linked one or two... pages ago; but at first look it appeared to me it says same thing of SPICE kernels. At a closer look, I see it does not: it says the rectangular area 60° large falls OUTside the circular area, unlike they say in SPICE kernel:

Image from document:




As the pictures in the document match with my 3d geometric construction, I think we can assume SPICE kernel is wrong and document is right.

Repeating image for better readibility of standalone message:

Good point.

I think this image obtained by overlaying a raw CIVA image to my "square & double circle" image solves the issue:


i.e., the CMOS is not completely inside the 70° circle (as stated in kernel file) AND it does not completely include the whole 70° FOV; it's "right in the middle" of the two cases.

This also means that any not-black object/detail/thing in corners of CIVA images is just an artefact.

Quick note: One theory I had about the three objects, is they could somehow have been kicked up by Philae when it landed, and now be moving away from the Philae 1st touchdown point. That seemed at minimum a reasonable theory given that the objects are pretty close to the 1st touchdown point and moving generally away from the 1st touchdown point at a time that might to be soon after the touchdown.

But, objects originating at the Philae 1st touchdown point would be following a line directly away from the 1st touchdown point in these photographs. These objects don't do that--not even close--so I think we can completely rule out that possibility.

Also--I'm a bit confused about when the "mosaic" image could have been taken. It seems to show Philae at a bunch of different times and yet I don't really see any obvious seams. Maybe their image preparation people are just that good.

And yet further questions--if there are three objects, maybe there are even more? Maybe some of the others are in the same general area but smaller/harder to detect? I can't see any more than the three on a quick inspection--but that doesn't mean there are none!

And why do we see the objects as purely black? Wouldn't any object seen at this sun angle have a lit side an a shadow side? I think perhaps we do see some of the lit side of the objects, but the color of the lit portion of the objects blends well with surrounding comet terrain behind it. Only the side of the object that is in shadow (ie, nearly black) stands out clearly.

Could this be a seaming / processing artefact in the Philae Touchdown Mosaic , just below the 15:23 insert
Click to view attachment

Nice work on the CIVA FOV issue.
Putting it all together, I get to following approximate crops (from the full CMOS raws) on the released images:


Click to view attachment Click to view attachment Click to view attachment Click to view attachment Click to view attachment Click to view attachment

Very consistent with vignetting. Reminds me of fish eye photos.

I'm sorry but I don't think they're right: the region between the two circles IS valid: it is part of the FOV and it is covered by CMOS. Insead, outside of the external circle we are out of the lens coverage, so the CMOS is detecting invalid data.

See comparison between your image and mine:


Additionally, I think this image is not part of the first sequence, as it was shot after the 35° rotation (don't know if clockwise or CCW):


I made again from scratch my Sketchup model of CIVA pictures.

This time I can eventually see the overlapping of the lower parts of the picture, but there's still something wrong as the intersection of visual rays on same object leads to a point 16 meters away from lander (lines are hidden for better reability, you must enable hidden geometries from menu).

I also added "scenes", which you can access by clicking the tabs: first 6 scenes are overviews from far away; next 6 scenes are taken directly from real cameras positions; last 3 ones shows how the model foot match the pictures. These scenes result in a very immersive effect!

http://lc84.altervista.org/rosetta/CIVA-3d-model-3.skp

Probably true, this line is in the SPICE kernel under the section "Field of View and Detector Parameters":

"The author of the Kernel is not yet sufficiently informed to give the most relevant value to IFOV."

This is mentioned since the computed per-pixel angular size in the kernel doesn't match that in reference [7].

What makes you say that? I thought all the civa frames we've seen are pre-rotation.

All those images are from first sequence. I think that the second sequence turned out to be in darkness and the rotation was done much latter.

About the useful FOV, I stand by my interpretation.
There are details almost up to the limit, so the lens edge must be farther out. See the third image (the one with the leg and well exposed).