Discussion What is the most efficient way to crowdsource the 3D system coordinates

[Harbinger]

OMG.. can't believe anyone else haven't mentioned it yet....

The Galaxy Map now only shows distances with TWO decimals.... not three....

That really sucks....

I'm gonna put the TGC on hold a bit - Until we figure out how to make that work
(I'm gonna be running alot of tests...)

You know, I never even noticed due to being so tired last night.

That'll be why I was seeing variances of up to [+|-] 0.005 when I was checking distances to just 4 of our traditional reference stars. At 2 decimal place precision, 4 distance measurements definitely isn't good enough to get an accurate fix on the coordinates.

 

[wolverine2710]

From another thread were disappearing station were discussed.

Originally Posted by Morbad
They intentionally swapped everything around so you'd have to find it all again, otherwise they would have left it the same.

Response by Michael Brookes.

Michael Brookes
No we haven't. The various systems used to generate contents and economies have been changes and the data re-exported. So only hand authored objects will remain - everything else will change. This will happen as part of the gamma reset as well.

Michael

 

[TornSoul]

Just did a quick test where I took the recorded distances we have - and rounded them to 2 decimals.

The bad news is that 95 of the "non-canonical" systems can no longer be resolved to a 1/32th coord.

I even went so far as to check each coordinate around the point trilateration reported, in a 7*7*7 cube, to see if one of those would fit all reported distances.

About a handful even had several coordinates that would fit the reported distances (as far as I can tell - that only happen for those stars very far away)


so.. uck...

amazing how much difference on lousy decimal can do...

 

[RedWizzard]

Only if you know where you want to go.

Trade route planners still need the data - As they need to incorporate it in the calculations of where you *want* to go.

Yes, though that's less of a problem (IMHO) because they can still display the X highest profits and let you use the galaxy map to figure out which one to use (and I kinda subscribe to the idea that perfect trade assistants will detract from the game so I'm not bothered too much if they are somewhat limited). Anyway for that purpose 100% accuracy is not really that important, so the pragmatic thing to do for trade planners is probably just to get "good enough" locations and not worry about it further. But that's somewhat unsatisfying and certainly less interesting as a problem to me.

 

[RedWizzard]

Just did a quick test where I took the recorded distances we have - and rounded them to 2 decimals.

The bad news is that 95 of the "non-canonical" systems can no longer be resolved to a 1/32th coord.

I even went so far as to check each coordinate around the point trilateration reported, in a 7*7*7 cube, to see if one of those would fit all reported distances.

About a handful even had several coordinates that would fit the reported distances (as far as I can tell - that only happen for those stars very far away)


so.. uck...

amazing how much difference on lousy decimal can do...

I think that cube searching technique is the way to go for refining the trilateration output. If it works for 90% of the cases where our previous trilateration algorithms don't then the upshot will just be requiring more references in the remaining small number of cases - no big deal. What concerns me a bit is the possibility that a system on the edge of the sausage might not be locatable at all with references in the sausage (just as Sag A* was not locatable from the Pill).

I'm not surprised this particular one decimal makes a big difference. We've gone from having distances with 30 times the precision of the grid (0.001 vs 0.03) to only 3 times the precision. You could say we've lost 90% of the relative precision of the distances.

 

[TornSoul]

I came up the following empirically in Excel for B2. I haven't tried to calculate a "best fit" axis (and a best fit would not necessarily be correct - we need the surrounding out-of-pill stars to get an accurate axis).
z = x * 0.45 + 7
y = x * 0.025 + 35
A 33 Ly radius cylinder using the above axis contains all pill stars (more or less). I was kind of expecting the radius to be 30 Ly but I don't think it is that small.

From patch notes
- Play area expanded (capsule 350ly long, radius 37 ly - 2400+ systems)

 

[CMDR Generic EventHandler]

From another thread were disappearing station were discussed.

Michael Brookes
No we haven't. The various systems used to generate contents and economies have been changes and the data re-exported. So only hand authored objects will remain - everything else will change. This will happen as part of the gamma reset as well.

Michael

So how do we tell which systems are handcrafted, are they the ones from Beta1?

 

[Biteketkergetek]

From patch notes
- Play area expanded (capsule 350ly long, radius 37 ly - 2400+ systems)

Perhaps it is another floating point error?

 

[wolverine2710]

So how do we tell which systems are handcrafted, are they the ones from Beta1?

I think you mean SB2 instead of Standard Beta1. From what I did understand from Michael Brookes. The SB2 partial list contains the star systems which are NOT procedurally generated (PG). As in those PG ones are more difficult to obtain - costs FD time and resources. In TOR (RW's reference JSON file) you can see what was supplied and what was crowd sourced.

 

[RedWizzard]

From patch notes
- Play area expanded (capsule 350ly long, radius 37 ly - 2400+ systems)

Yes, I was talking about the beta 2 pill. I think I saw somewhere that it was supposed to be 30 Ly across but I can't find that reference now. It's definitely not 30 Ly across but I thought perhaps they meant 30 Ly radius.

I have no idea whether the beta 3 long axis is the same as the beta 2 long axis.

 

[RedWizzard]

I think you mean SB2 instead of Standard Beta1. From what I did understand from Michael Brookes. The SB2 partial list contains the star systems which are NOT procedurally generated (PG). As in those PG ones are more difficult to obtain - costs FD time and resources. In TOR (RW's reference JSON file) you can see what was supplied and what was crowd sourced.

I seem to remember Michael saying that list was the occupied systems, not the non-procedurally-generated systems. My guess is that they had already extracted that list for some reason.

There were different levels of procedural generation in FE2/FFE and I think it is the same in ED. A procedurally generated system could have it's name overridden or it's contents overridden. System like the Wregoe/Wredguia systems in beta 2 were clearly procedurally generated but apparently even Styx was at least partly procedurally generated because it's changed in beta 3.

 

[wolverine2710]

the route planning problem is pretty well solved in game. That was my main interest.

I must have missed something. When I for example go from Eranin to YanYan a route is shown. Which is say 6 jumps. I was hoping that by selecting YanYan it would show me the first HJ jump and when exiting that one the next is shown till I arrive at YanYan. For me that doesn't work. It seems I have to write down the route and then make the jump. With Trade Dangerous it shows me a route and with alt-tabbing I can find out what next jump to make. Again. It seems I have missed something in ED or....

 

[Michael Brookes]

I think you mean SB2 instead of Standard Beta1. From what I did understand from Michael Brookes. The SB2 partial list contains the star systems which are NOT procedurally generated (PG). As in those PG ones are more difficult to obtain - costs FD time and resources. In TOR (RW's reference JSON file) you can see what was supplied and what was crowd sourced.

That's not correct - the list was systems that have had economies attached to them. As a heads up generated systems will change again at the beginning on gamma.

Michael

 

[TornSoul]

I'm a but unsure if the TGC is worth pursuing still?

Some kind of paradigm shift is needed then at least (seeing as how we can't pin down the coordinates accurately for a large portion of systems)

 

[TornSoul]

That's not correct - the list was systems that have had economies attached to them. As a heads up generated systems will change again at the beginning on gamma.

Michael

Economies - aye. How I recall it as well.

It's interesting to see that a HUGE volume of stars, outside the sausage, now also have economies (at lest they are colored as such in the GM)


Michael,if you're still reading, any idea of the chances of getting distance with 3 decimals again?
2 decimals is a bit of a gamebraker for us.

 

[Harbinger]

@Michael Brookes: Although things will inevitably change in the Gamma release is it possible to get a csv listing the coordinates of the B3 systems having an economy?

 

[wolverine2710]

@Michael Brookes: Although things will inevitably change in the Gamma release is it possible to get a csv listing the coordinates of the B3 systems having an economy?

Send him a PM this morning and also told him about the 2dp issue - and if 3dp would be possible in a patch.
Haven't got a response yet. I guess the SB3 release is consuming a LOT of his time. Fingers crossed.

 

[Chromatix]

The populated/economic system list would certainly be useful to trade-planners, even without the intermediate, unpopulated systems.

The latter are still useful to refine the results, of course, since fuel consumption depends on the availability of intermediate systems to shorten the jump legs.

 

[CMDR Generic EventHandler]

That's not correct - the list was systems that have had economies attached to them. As a heads up generated systems will change again at the beginning on gamma.

Michael

So what we really need are.

1. A display of the co-ordinates on the map screen
2. Return to 3 decimal places for distance
3. A list of known fixed systems that are not positioned by the generator.

Pick 2.

Personally 2 from that list would make me happy and I guess a lot of other people too.

Without any from the list, the only way I can see of deducing if the System is a "Fixed system" is calculate the distance from each star to Sol and then see if the current distance matches up with the beta 2 distance. Once we find out which systems are fixed, then calculate all the other stars from these reference points. Further out stars from a list of precalculated close by stars that have 10+ agreeing positions.

GEH.

 

[wolverine2710]

So what we really need are.

1. A display of the co-ordinates on the map screen
2. Return to 3 decimal places for distance
3. A list of known fixed systems that are not positioned by the generator.

Option 4. ED writes in a file (smilar to netlog.log) the coordinates in 5dp when exiting a HJ jump and entering a system.

I've asked Michael Brookes if that would be possible. His response:

Michael Brookes. We're busy with the game features so it's not something we can provide at the moment.

Of course quite understandable.

Getting 3dp back in the GM would solve our problems and give us a fighting change. An awful lot of time, energy, theory has been put into this thread. We finally nailed the algorithms, were ready for TGC and now its basically to no avail or back to the drawing board. It was in 3dp in the GM to start with. I did ask Michael if it could be put back to 3dp in a future patch. Did not have a response (yet). I'm really hoping we can get it back......