Labels overlay on Moon

[brokfn]

Hi,

I thought that the labels on the moon were designed to not overlay each other.
Meanwhile, here is what I've got (using only locations files provided with basic package of course) :




Hard to read indeed !

Bye,
Brokfn

[selden]

Well, that is where Apollo 17 landed, so there are many named features in a very small area.

I suspect that they also need to have more precise locations associated with them.

[t00fri]

Hi,

I thought that the labels on the moon were designed to not overlay
each other. Meanwhile, here is what I've got (using only locations files
provided with basic package of course) :
...
Hard to read indeed !

Bye,
Brokfn

That /of course/ can only work on a statistical basis! The criterion is
that the number of labels/area = density of labels is kept ~
constant for any zoom level. An importance weight was assigned
to every location based on an almost perfect Gaussian distribution of
the crater sizes on the Moon.



The above extreme local density fluctuation of locations near
the Apollo 17 landing area is clearly falling out of such a distribution,
yet in a statistically irrelevant manner.

Everyone is invited to study my detailed, commented theoretical
derivation that is available in my original thread on the new moon
locations. Reading it first might have prevented that sort of "complaint".

http://www.celestiaproject.net/~t00fri/images/moon_weights.pdf

Bye Fridger

[brokfn]

Well...

Let's zoom a little closer :


Can you tell me what you read or is it statistically irrevelant ?

Bye,
Brokfn

[t00fri]

Well...

Let's zoom a little closer :

Can you tell me what you read or is it statistically irrevelant
?

Bye,
Brokfn

This particular example may well be funny, but you are
missing the point and this image is certainly NOT generic.
That's all one can do without "hand sorting" every single
label, which NOBODY wants to do, given the more than
100000 labels that we administrate in Celestia

Perhaps, it is worth counting the fraction of such label
overlaps compared to the total number of ~ 8600 moon
labels. I bet that number is pretty SMALL...That's what the
term "statistically irrelevant" implies.

Another instructive exercise is to compare the original list of
~8600 labels WITHOUT attached importance factors. There
you will have a lot of fun, for sure.

Here is another one for you: among our 10000+ galaxies
you will also find some where the spirals wind in the
wrong sense as compared to the photographic images.
Also this is statistically irrelevant and intrinsically
unavoidable unless you design every single galaxy
individually

Bye Fridger

[selden]

Well, it's not all that hard to determine if there are any duplicates: sort the file by LongLat, then sort deleting duplicates, then compare.

Hopefully there are few enough that it isn't too hard to determine why they exist.

In the case of the Moon, there are 10 duplicated LongLat values, varying between 2 and 7 entries for each. As best I can tell, the problem is simply that more precision is needed in the values. Three significant digits just aren't enough. Finding a database that has the necessary precision is a problem, however.

Here's the list of 7 entries that are at the same location near the Apollo 17 landing point:

LongLat [ 30.8 20.2 0 ]
All are Type "LF"

Location "Cochise" "Sol/Earth/Moon"
Location "Light Mantle" "Sol/Earth/Moon"
Location "Powell" "Sol/Earth/Moon"
Location "Shakespeare" "Sol/Earth/Moon"
Location "Sherlock" "Sol/Earth/Moon"
Location "Trident" "Sol/Earth/Moon"
Location "Van Serg" "Sol/Earth/Moon"

[t00fri]

Well, it's not all that hard to determine if there are any duplicates: sort the file by LongLat, then sort deleting duplicates, then compare.

Hopefully there are few enough that it isn't too hard to determine why they exist.

In the case of the Moon, there are 10 duplicated LongLat values, varying between 2 and 7 entries for each. As best I can tell, the problem is simply that more precision is needed in the values. Three significant digits just aren't enough. Finding a database that has the necessary precision is a problem, however.

Here's the list of 7 entries that are at the same location near the Apollo 17 landing point:

LongLat [ 30.8 20.2 0 ]
All are Type "LF"

Location "Cochise" "Sol/Earth/Moon"
Location "Light Mantle" "Sol/Earth/Moon"
Location "Powell" "Sol/Earth/Moon"
Location "Shakespeare" "Sol/Earth/Moon"
Location "Sherlock" "Sol/Earth/Moon"
Location "Trident" "Sol/Earth/Moon"
Location "Van Serg" "Sol/Earth/Moon"

Selden,

since I used nothing but the "official" location sources
(USGS) I naturally assumed that this organization knows
what they are doing . Perhaps they are also
"understaffed"?

Anyway, you just proved that these 10 bugs out of ~8600 are statistically irrelevant

Bye Fridger

[dirkpitt]

Would it be a good idea for Celestia to never draw labels overlapped, but rather if several labels existed for a point
those labels were drawn distributed around the point rather than on top of each other?

Sort of like this:

Code: Select all

Cochise
Light Mantle
Sherlock
+

Or this:

Code: Select all

 Sherlock   Light Mantle
Cochise  +   Powell
 Trident    Van Serg

If there were two points very close to (but not identical to) each other, something more complex may be required.
Perhaps this paper (pdf) might be an interesting start?