#89 three visual anomalies seen with stars

[Selden Ball]

I know the third problem has been reported before,
but I don't recall seeing reports of the first two. Maybe I've just forgotten....
I noticed these while looking at Medusa's Sheliak STC.

1. From some viewing angles, the orbital path of a binary star system has a bright segment. The length of the bright segment varies depending on the viewing angle.

2. From some viewing angles, one of the stars in a double star system has a bright-white annulus. Which star is affected and the width of the annulus depends on the viewing angle.

3. Stars have bright white central spots in Star Style: scaled disc and vertex shader render paths (including OGL2)

An Addon demonstrating these is
http://www.lepp.cornell.edu/~seb/celestia/files/star_anomalies.zip

It includes the STC and an HTML file with images to demonstrate the anomalies.

s.

[Chris Laurel]

Logged In: YES
user_id=159994
Originator: NO

1. The bright segments appear when the star glare halo is rendered in front of the orbit path. I'm not sure what can be done about this except to render the glare as a post-processing effect (implying at least some limited form of HDR rendering.)

2. The annulus seems to be gone now that there is better positing of the glare halo.

3. This is obviously a problem with the scaled disc star style. I'm working to fix it.

[Selden Ball]

Logged In: YES
user_id=662401
Originator: YES

I've reviewed the symptoms using the updated star rendering code, building Celestia from cvs on December 11, 2007 at about 7:30AM EST.

Unfortunately, they're all still present, including symptom #2.

Here's a URL which demonstrates the annulus when using the Addon previously mentioned:
cel://Follow/Sheliak A/2007-05-04T22:32:23.60212?x=AJxjnWrpJ9fabbYJ&y=AIZzrCTDYwbY/HYs&z=AEy32kAf6w/9+mQc&ow=0.116547&ox=0.224411&oy=0.955892&oz=-0.149421&select=Sheliak A&fov=40.650936&ts=1.000000&ltd=0&p=0&rf=2463159&lm=512&ver=2

It displays Sheliak A with a white ring around a blue center on my system in Basic and Multitexture render paths. Paths which use vertex shaders look fine. Selecting different star styles makes no difference.

[Toti]

Logged In: YES
user_id=1387492
Originator: NO

The annulus is also the result of the proximal part of the star emerging from the other star's glare.

[Chris Laurel]

Logged In: YES
user_id=159994
Originator: NO

Yes, the annulus is the result of the star showing through the glare. The problem is, I don't know how to fix it. We have to pick some z value at which to render the glare, and no choice will be perfect. The same artifact can be seen in games, where sprites use to render smoke, fire, etc. intersect objects.

With HDR, glare is rendered in a postprocessing step, eliminating artifacts caused by placement of the glare sprite. Aside from that, I don't know what can be done to make this problem go away. Actually, I can think of solutions, but they're terrible hacks that require significant amounts of coding, e.g. rendering the glare sprites in front of both stars.