Parallax uncertainty presentation

[Janus]

In looking through the Gaia data dump, I noticed something.

In addition to the parallax column, there is a parallax_error column.
The former in mas, the latter as well. (I am assuming most people reading this can translate mas to LY on their own.)

Essentially, milliarcseconds of their best guess, followed by how certain they are.
This results in a range instead of a simple distance. (Min LY, Max LY, probable LY.)
Normally Celestia ignores this uncertainty, displaying stars as something like dots at an exact Light Year.

What I am asking is if anyone besides me is interested in being able to visualize that uncertainty.
My proposal is to add a function to the stars & render objects.

Draw a line running from the minimum distance to a star, to the maximum distance, with a larger dot in the middle.
The purpose being that you can then view a selection of stars from the side, with uncertainty in position already graphed out visually.

In my own fork I added functions to lookup stars by RA & Dec, by distance from SOL, and distance from the currently selected star.
Those functions were added for my own use.
This is simply an addition to that base.

If no one else is interested, I will add and use it on just my own fork.
If there is interest, then once I have it working, I will send in the changes so they can be added.


Janus.

[john71]

Sounds interesting. By the way do you mean the Tycho-Gaia 2.5 million stars database or the 1 billion stars "big Gaia" database?

[Fenerit]

I'm interested in these things; at least to discuss and/or improve them after their check out. Unfortunately i'm not to sure to be able to code something within Celestia, then my suggestions should be just in "description" mode. For example, unless seen, I don't have figured out your error-bar method. Do you mean a cross centered on stars where the up line piece is the uncertain "forward us" and the bottom line piece the uncertain "toward us"? The line pieces left and right their relevant displacement uncertains? How do you set the thing for an observer who looks toward the star from a place in space beyond that star: the option it is not shown in this case (rightly imho)?

P.S.
You are welcome in this matter, then I can do further suggestions. I think a new kind of markers must be added to the usual ones, that is those common on sky atlases for deep space objects. An ellipse for galaxies, a crossed circle for globulars, a sparse tiny discs for open clusters and the circle with the four lines appendix for nebulae. To display them would remain in Celx code, at least.

[selden]

A problem with drawing stars as lines from min to max is that they obscure other stars, making the display hard to understand. Would it be possible instead to be able to separately select drawing all stars as points at their minimum distances, drawing them at their standard distances and at their maximum distances? And/or toggle between two of them?

[Janus]

Sorry if I was unclear, it was late after a long day when I posted.

From the perspective of earth/sol, nothing would change in the display at all.
It would only be visible from views outside this solar system.

My idea is to be able to select individual or groups of stars to show this way.

Warning: I am not an artist, and I do not wish to offend any real ones.
The basic shape would be this.
r = 1/2 Radius; R=Radius; Please ignore the underscores, they are for alignment only.
______R
rrrrrrrrrRRRrrrrrrrrr
______R

A cylinder at one half of the star's diameter, drawn from min to max.
With the star drawn normally in the middle of it.
Perhaps with the cylinder being a transparency.

The idea is not up for up close observation, but from a distance.
I came up with it when trying to compute old/future stellar positions based on motion data, which includes lots of uncertainty.

One of my long term ideas is to use gravity mapping from estimated masses, combined with motion data, and uncertainty, to find combinations of distance or mass offsets that work well together to make a more accurate 3D map.

After all, stars move.
So if proper motion is applied, there will be somewhere in the uncertainty range of distance that will work using gravity and align with proper motion data.

It is a long term goal, and something I am simply curious about.
What I learn working on this helps me in other areas.

I do not claim to be an astronomer.
I actually found it easier to make sense RA & Dec by reading celestia's sourcecode, than astronomy books.
I do not have it all yet, but I am slowly getting it.

I used to have a copy of Celestia I had modified to add three north and south pole stars to show me the plane of the solarsystem vs the galaxy via an added constellation.
I even added a pair of stars at 0,0 to show where that actually points at.
The big T shape put the whole thing in perspective quite easily.
In distinct contrast to chapters of math I can't read using concepts that make no sense without somewhere to start.
Sadly, I lost that copy when an HDD died unexpectedly.

I hope I made my idea for showing position uncertainty clearer.


Janus.

[Chuft-Captain]

Janus,

I don't know if you are aware of the "fading orbits" feature which was implemented (but never released) for Celestia 1.7.0.
It seems to me that his approach perhaps might be adapted for what you want to acheive (ie. a line centered on the star and fading out relative to the distance error).
If you had access to the 1.7.0 codebase you might even be able to re-use some of the relevant code, and this would probably be more efficient and performant than most other approaches.
If you haven't seen what these fading orbits look like, perhaps someone with the 1.7.0 codebase can post a screenshot for you.

CC

PS. It goes without saying that features such as this should be implemented as an "optional" feature which is disabled by default, and toggled on/off as Selden suggested.

[Janus]

No, I am not familiar with that.
However, you are correct, it does sound like it could be adapted.

I also agree that such changes should be off by default.

For now though, I am simply waiting until things get organized here.
This is a large project, and it takes time to break inertia.

I am also aware that my needs are different than others.
So not everything I do for my own needs will work for others.

Janus.

[Chuft-Captain]

FWIW,

This is what the fading orbits look like:

[FarGetaNik]

This is what the fading orbits look like:

That looks aktually cool and could be very helpful! Too bad I can't get Celestia 1.7

[Janus]

I like the basic idea.
However, I would want it to be brightest in the middle, while fading toward both ends.

I look forward to things getting organized here.
Until then, I will keep tinkering.

Janus

[Goofy]

If this can help someone, I have the fading orbits release, and I attach here the zipped .exe file.
It was released years ago by Guillelmo Abramson.
I don't remember if there were other files to be added to the 1.6.1 standard release, so ieventually ask me for them.
Bye

Goofy

[John Van Vliet]

for the fading orbit

edit ?/src/celengine/render.cpp line :1956
const double LinearFadeFraction = 0.0;
------- to ----------
const double LinearFadeFraction = 0.66;

and rebuild
the higher the value the FASTER the orbit fades

[Janus]

Thank you very much.

It is used later in the same file @2056 & 2071.
Appears to be a fade by amount since it uses 1-LinearFadeFraction in an IF statement.
Not sure why it was split, the call would work the same either way given how it is written.
I have seen other tweaks of the same type to account for older hardware that is helped by avoiding the extra math.

I will be playing with this.
Does anyone know which way the fade faces.
Dark behind fading in front, or fading behind dark in front.
I am hoping it is dark behind, that would follow visual logic better.

My initial thought is adding it to the menu selection criteria.
My next thought is adding it to the appropriate object(s) so it can be modified by a script.

I have already been looking at the ability to change orbit colors anyway, this will make a nice addition.


Janus.

[Goofy]

I have already been looking at the ability to change orbit colors anyway, this will make a nice addition.

Hi Janus.
Actually orbit colors change can be done in .cel scripts with a command line like this
setlinecolor { item "planetorbits" color [0.43 0.04 0.04]
and something similar works in celx, I think.
You can do it for
planetorbits
dwarfplanetorbits
moonorbits
minormoonorbits
asteroidorbits
cometorbits
spacecraftorbits
starorbits
equatorialgrid
planetographicgrid
planetequator
galacticgrid
eclipticgrid
horizontalgrid
ecliptic
too, each of them with a different color, if you like.
The same can be done for labels, using
setlabelcolor.
Hope this helps.
Bye
Goofy

[Janus]

Thank you.
I was unaware of that.
I will see how many other things the same pattern applies to, and keep my changes as consistent with it as I can.

It is funny, but one of the things I was asked to do is animate the zodiac.
To turn them on and off individually, and if I could, to give them each different colors.

It might be interesting to recolor the stars instead or as well.

Janus

[FarGetaNik]

If this can help someone, I have the fading orbits release, and I attach here the zipped .exe file.
It was released years ago by Guillelmo Abramson.
I don't remember if there were other files to be added to the 1.6.1 standard release, so ieventually ask me for them.
Bye

Goofy

Thanks, that worked! Orbits are now fading, it even fixed the spice bug with orbit rendering I described here:
http://celestiaproject.net/forum/viewtopic.php?f=2&t=17456
Windows was a bit suspicious with this exe but hey...
Just a few questions, it says Celestia 1.6.0 at launch, I hope all features of 1.6.1 are included?
And it brought back the bug described here, just that the squares are now only a few pixels wide and it affects now all graphic processors:
http://celestiaproject.net/forum/viewtopic.php?f=15&t=17487

edit ?/src/celengine/render.cpp line :1956
const double LinearFadeFraction = 0.0;
------- to ----------
const double LinearFadeFraction = 0.66;

and rebuild
the higher the value the FASTER the orbit fades

I guess it is not done by editing some of the files included in the 1.6.1 installation, but rebuilding an executable.

Added after 5 minutes 23 seconds:

Hi Janus.
Actually orbit colors change can be done in .cel scripts with a command line like this
setlinecolor { item "planetorbits" color [0.43 0.04 0.04]
and something similar works in celx, I think.

There was a recent discussion regarding this, everything discribed worked fine for me: http://celestiaproject.net/forum/viewtopic.php?f=2&t=17443