What is it?

[chris]

Time for a little game... Here's a screen shot from my development version of Celestia, showing a minor new feature. Can anyone guess what's going on here?

Janus

--Chris

[Ricardo]

Ring's shadows!

[chris]

Ring's shadows!

That was quick!

The work is part of an effort to clean up and enhance the Celestia's shadow code. The generalization of ring shadows for bodies that aren't themselves encircled by rings was inspired by an entry in Emily Lakdawalla's blog:

http://www.planetary.org/blog/article/00002078/

The Celestia image is quite close to the actual Cassini image (I think frame 8 in the animation...) I think that the differences are due to slightly incorrect size, shape, and perhaps orientation of Janus. For this test, I used a SPICE kernel for the orbit of Janus--the default elliptical orbit isn't nearly accurate enough.

Here's another image of ring shadows, this time on Rhea:

Ring shadows on Rhea

Though I don't have a Cassini image to compare with this time, I think that the shadows should be a little fuzzier. Because Rhea is farther from Saturn than Janus, the apparent size of ring features will be smaller with respect to the apparent size of the Sun, which should tend to blur things out more. This effect is only incompletely accounted for in my model.

EDIT: Forgot to mention... The darker circular region on the lower part of Rhea is a partial eclipse of the Sun by Mimas.

--Chris

[duds26]

These are good improvements.
The generalization is a big step forwards.
The blurriness will make Celestia pretty realistic for the mentioned situations.

[Hungry4info]

That's amazing Chris! I'm glad to see this

[chris]

That's amazing Chris! I'm glad to see this

Thanks! I thought you might be interested--wasn't it you who pointed out the too-sharp shadows of Uranus' rings quite a while ago?

--Chris

[Ricardo]

I cannot see this feature in the roadmap (or I missed it?); this is for 1.6.x?

[volcanopele]

I agree, the issue of the non-point source sun should be addressed, as it would certainly be an issue for the outer satellites in particular, looking at that image of Rhea.

While you've brought this up Chris, is there any chance for the reverse, satellite shadows on the rings, in the new shadow code?

[symaski62]

http://saturn.jpl.nasa.gov/photos/raw/

[Hungry4info]

That's amazing Chris! I'm glad to see this

Thanks! I thought you might be interested--wasn't it you who pointed out the too-sharp shadows of Uranus' rings quite a while ago?

--Chris

Yes sir =)

[chris]

I agree, the issue of the non-point source sun should be addressed, as it would certainly be an issue for the outer satellites in particular, looking at that image of Rhea.

I've already mostly addressed it... It turns out that mipmaps (basically smaller, prefiltered versions of the base texture) give a reasonable visual approximation of the affect of a non-point light source on ring shadows. I wrote some new code that calculates the mipmap level to use based on the resolution of the ring texture and the projected size of the ring shadows. The graphics hardware interpolates between mipmap levels automatically, so the there's no 'jump' between mipmaps as the shadowed object gets further from the rings.

While you've brought this up Chris, is there any chance for the reverse, satellite shadows on the rings, in the new shadow code?

[/quote]

Yes, though some other problems need to be solved first. The shadows need to be volumetric, with the size of the umbra decreasing with increasing distance from the shadow caster. The effects of this are very evident in the Cassini's images of long shadows cast onto the rings near Saturn's equinox. At the moment, the same umbra size is used over the entire shadow--a decent approximation for shadows that aren't dramatically extended along the light-to-occluder direction.

--Chris

[Hungry4info]

Chris, what does that look like as the equinox passes? Is one side of the ring gradually dimmed as the other is gradually made brighter?

[chris]

Chris, what does that look like as the equinox passes? Is one side of the ring gradually dimmed as the other is gradually made brighter?

The abruptness of the transition depends on your viewing geometry. If the Sun is behind you, the transition from lit to unlit side is very abrupt. It's more gradual when the the viewer is facing the Sun. This is (I think) correct for those sections of the rings that are primarily backscattering. Forward scattering of light by the rings is not yet modeled, so there's more work to be done.

I've committed the patch to SVN, so you can try it out yourself now.

--Chris