Cel URLs not precise?

[selden]

I've been trying to generate a Cel:// URL showing the NEO asteroid 1994 XM1 flying close by the Earth. The asteroid is rather small, so the viewpoint has to be close to its surface to make it look dramatic. I've been positioning it only a meter away. Unfortunately, every URL I've captured results in a viewpoint with the asteroid nowhere in the field of view and typically 70-80 meters away.

I've tried all sorts of viewpoint options and none of them seem to make any meaningful difference. Even combining phase lock betwen xm1 and the earth and tracking (either of earth or of xm1) doesn't help. Tracking xm1 is the only way I've been able to keep it in sight, but then the earth is out of view.

See http://www.lns.cornell.edu/~seb/celestia/minor-planets.html#3.2.2.1 for my most recent attempt.

sigh.

[Calculus]

Try mine
it seems to work cel://Follow/Sol:1994%20XM1/1994-12-09T17:41:44.93?x=B4eanaq3EQzBDA&y=Hb54iaOFMQI&z=UEJEFcs2Yc3w/////////w&ow=-0.229387&ox=0.287453&oy=0.712594&oz=0.597464&select=Sol:1994%20XM1&fov=45.000000&ts=0.000000&rf=791&lm=32

it works perfectly!

[selden]

That's right.
Tantalize me.

How did you do it? Hand edit it?

When I simply follow the asteroid and save the url, the result is way out in left field.

[Calculus]

I don't see any way to hand edit a Cel:// url .
I just did a CTRL INS and a paste as usually with Celestia V1.3.0pre4
I don't understand what cause you this problem.

[selden]

I don't understand it either.

Maybe somehow it has something to do with the graphics card?
(Although I don't see how: location in space should be entirely controlled by Celestia itself, I would expect.)

Anyhow: this problem is seen on this system:
256MB, 2GHz P4, NT4 SP6a;
32MB ATI RAGE 128 Pro, OpenGL v1.2.1652
Celestia v1.3.0pre4

I haven't tried it on my home system yet, which has an Nvidia card.

[chris]

Hmm . . . precision of time stored in the URL is one hundreth of a second. How fast is the asteroid moving?

--Chris

[selden]

Chris,

Yup: that's it.

Apparently the viewpoint vectors recorded in the URL do not correspond to the timestamp in the URL. My guess would be that to do it right would require setting Celestia to the time that's actually going to be used in the URL and then recording that vector state. Is that possible?

Alternative suggestion: have the "pause" command stop on a 1/100 second boundary (or whatever precision you decide URLs should use). This would help somewhat by hiding the problem.

According to Horizons, the asteroid was travelling at 11.3 km/sec relative to the Earth. I dunno what that'd be in heliocentric coordinates, but it's the right order of magnituude (113 meters in 0.01 sec). I had tried just doing a "pause" but that didn't help: obviously it's hard to stop at a 1/100 second boundary!

However, it is possible to get the desired viewpoint if you do it by generating two URLs. I'm guessing this may be how Paul managed to get the desired result -- he started from one of my URLs.

1) pause, roughly position objects, generate url, goto URL.
This puts you at the 1/100 second boundary and stopped.

2) position objects precisely, generate new URL.
This second URL does position the viewpoint correctly, although time is left stopped.

It'd be much better, though, if one could generate precise URLs while time is progressing at whatever rate is appropriate.

[Calculus]

However, it is possible to get the desired viewpoint if you do it by generating two URLs. I'm guessing this may be how Paul managed to get the desired result -- he started from one of my URLs.

Yes this is how I did it (but not on purpose!)

[selden]

Bad news: the increased precision of time in pre5's URLs doesn't really help. One of the URLs I captured while time was progressing normally was off by over 300 meters.

I fear that Celestia really does need to be set to the time that's going to be put into the URL so that the other vectors are generated correctly for that instant.

Doing the sequence of commands pause, take 1st url, exit Celestia, goto 1st url, adjust viewpoint, take 2nd url does result in the desired viewpoint. Presumably this is because, when Celestia is restarted the second time using the first URL, time is set to the corrrect instant and stopped (not paused). Trying to use the same sequence of steps but not exiting Celestia leaves time "paused" (not stopped) and the resulting 2nd URL still puts the viewpoint in the wrong location.

[chris]

Bad news: the increased precision of time in pre5's URLs doesn't really help. One of the URLs I captured while time was progressing normally was off by over 300 meters.

Ugh . . . I think I forgot to make Celestia actually read more than just two digits after the decimal point. I look into it after work today, but I'm almost certain that's the problem.

--Chris

[selden]

Don't forget to have it read the full year, too.
v1.3.0pre5 is obviously discarding some of the year digits.

While trying to find a good viewpoint to look at the ISS, I wasn't paying any attention to the date. I would up in the year 13,000 or therabouts without realizing it. I took several example pictures at that date. ( See the thread
http://63.224.48.65/forum/viewtopic.php?p=13369#13369 )

Unfotrunately, selecting the Cel:// URL for that date puts you in the year 1313, although the year in the URL itself is clearly in the far future.
[url]cel://SyncOrbit/Sol:Earth:ISS/13149-02-09T06:13:18.79824?x=HZZjdULFBVzGDA&y=TR37fILHqg0&z=NF9dvh9hYXoN&ow=-0.466530&ox=-0.827366&oy=
0.248584&oz=-0.189793&select=Sol:Earth:ISS&fov=45.000000&ts=0.000000&rf=71559&lm=16[/url]

[chris]

Here's a new Celestia EXE to test, which should have fixes for all cel: URL time problems:

http://www.shatters.net/~claurel/celestia/files/celestia-win32-1.3.0pre5.1.zip

(somewhat whimsically called 1.3.0 prerelease 5.1)

Let me know if this solves the precision problems and the future dates problem.

--Chris

[selden]

Yes, it does fix both problems!