MESSENGER News Thread, news, updates and discussion Options

[ugordan]

There's something weird with the Where is MESSENGER page. Here are two screenshots of the same simulated time:


Notice how Earth's position shifts from right of the Sun to its left in the second image. Even taking into account the second image appears slightly rotated CCW w/respect to the first, it still doesn't look fit. Earth appears at two different positions at once. A bug?

[remcook]

Seems OK to me. Looks just like parallax when you move towards mercury in the centre of your view.

[ugordan]

How can you have parallax if your viewpoint (MESSENGER) is fixed?

[remcook]

hold a pencil in front of you and turn your head.

[ugordan]

You do realize that by turning your head, you're shifting your eyes, do you? MESSENGER is a single, fixed point in the simulation. Only the look vector changes.

[remcook]

yes, but I think it's probably a very similar effect that comes out of the simulations. Not sure though, but it looks like it. Angles stay the same, but the projection doesn't, because you look in a different direction, so your plane of projection has a different angle. Will try to make a drawing.

[remcook]

Here's the drawing. green and red show two objects at different distances. lines show middle and edges of field-of-view. If you have a projection plane perpendicular to the middle line at a distance c, a will not equal b for the two cases. Not sure this is how exactly they made the plot though.

Attached thumbnail(s)

 

[ugordan]

I'm not sure I get that plot, but I understand what you're saying about the distances. However, I'm talking about the angle between Earth-Sun-Mercury, see how it changes in the two instances. That does look like a bug. I have a fair amount of experience playing around in Celestia/Orbiter and never have I encountered such position shifts, even for large FOVs. To drive my point further, here are two plots from the Solar System Simulator:
Earth centered
Mercury centered

If you disregard the rotation (the sim probably aligns to the target north axis), you can see how the angle stays the same and is consistent with the first screenshot I posted, viewing Earth. Earth stays on the same "side" of the Sun in both cases, as one would expect.

[Bjorn Jonsson]

Having a fairly extensive experience playing around with 3D software and even writing a 3D renderer myself I cannot see how this could be anything but a bug. The views are supposed to show the Earth and Mercury as seen from Messenger at the same time and therefore the same location. The field of view is identical (or at least very nearly so), judging from Mercury's distance in pixels from the Sun. However, not only is the Earth's location relative to the Sun different in the two images, its 'distance' is also different and that difference is big. The field of view also isn't very wide (~30°) so distortion isn't significant.

Assuming the views generated by the JPL Solar System Simulator are correct the view showing the Earth from Messenger is correct (or very nearly so) while the one showing Mercury is not.

[remcook]

ah, I think I know what you mean now. sorry, I misunderstood you - I looked at the sun and Mercury, not the angles between them all.

edit - what I think I was trying to say is that if you have a flat projection plane you might get things distorted, whereas you have a spherical projection plane the angles would always stay correct. I guess the same problems occur with e.g. a fisheye projection. angles of things relative to eachother may change depending on where you aim it. But if the fov is not very wide...

[Malmer]

it has to be a bug.

There is no way that any projection can move objects around in respect to each other that way.

it looks to me as if these pictures are rendered with an equirectangular projection (ordinary 3d camera) and that the position of the camera has changed. Im 99.9% sure.

they probably have some strange offset. (like if you move out from an object in celestia and then turn the camera to look at something else)

The simulations look s bit substandard to me... not like something that you would expect from a project that cost millions and millions of dollars.

[remcook]

Had another think I see now I was just being plain stupid In my mind I had my projection plane some distance from the rotation point (Messenger), which would obviously give you a parallax. But maybe the programmer was similarly stupid Man, sometimes I wonder why i get up in the morning

[Phil Stooke]

On the date in question the spacecraft and Venus were some little distance apart. It looks to me like one simulation was viewed from Venus and one from Messenger, just a simple mistake in entering the viewpoint before doing the rendering.

Phil

[ugordan]

Nope. Not even Venus as viewpoint brings Earth to the other side of the Sun. Image.

[ugordan]

MESSENGER Mission News
December 2, 2006

http://messenger.jhuapl.edu/news_room/stat...t_12_02_06.html