Hi there,

Dene wrote:
>>>>Its a massive model (spans over 2 km/units in Blender) but has
>>>>not impacted framerate too much (yet)...still a lot of texturing
>>>>to do, but getting there...
>>>
>>>How are you finding the registration between the location of the models 
>>>and the terrain mesh?
>>>    
>>
>>Incredibly difficult, to say the least. Knowing the zero point of
>>the model is fine, but to have something to reference that against
>>has been my trouble.
> 
> Hi Chris,
> I have found two things that make placing objects difficult;
> - the algorithm used to reduce the triangle count from the base data can 
> eliminate significant landmarks eg Somes Island in the middle of 
> Wellington Harbour disappears in the 0910 scenery. IIRC the algorithm 
> eliminates vertices based on the elevation of vertices compared to those 
> around them. If the elevation difference is below a threshold then the 
> vertice is eliminated. In a tile that covers comparatively rugged 
> terrain the effect of this is more pronounced (the Wellington tile has 
> four fault valleys traversing it!).
> - there seems to be a N-S offset that shifts the landmass to the south. 
> This results in NZWN protruding into the harbour at the northern end 
> (where in fact there is approximately 100m of land between the end of 
> the runway and the sea)

You see the same effect at NZMC, where the southern end of the runway
sits in the Godley river. Also at YBBN where the southern end of the
secondary 14/32 runway sits partly in Moreton bay. The taxiway off the
end of 14 is not quite joined to the runway, so if I land a 737 on
that runway I need to backtrack to a taxiway that is joined, unless
I want to dip the nose of the 737 into the bay... :-)

> It will be interesting to see the effect of the landcover database on 
> both yours and my observations.

Robin Peels new apt.dat might help, but it sounds like that might only
improve airport layouts. The terrain mesh is too coarse around rivers
and lakes, possibly due to the way the algorithm you describe works.
Here is an example of how rough it is...

http://users.tpg.com.au/blobster/fg-vs-ge.jpg

Thats Google Earth/Maps at the top, and the default fg (plus someones
work-in-progress cbd model) at the bottom. You can see the difference
in the definition. Riverbanks, shorelines, and so forth are simply
a dogs breakfast. I don't doubt that it will consume more RAM to add
more accurate terrain, but surely there is a better compromise than
the current situation.

> Keep the screenshots coming Chris, I add them to the;
> http://denemaxwell.dnsdojo.net/users/ALBUM/Australia-Scenery/index.html
> page every time a new one comes out.

No worries. There might be 1 or 2 more coming up, but I almost feel
like completely halting the work I'm doing due to the fact that my
models will look silly sitting on the innaccurate terrain and airport
definitions.

Kind regards,

Chris Wilkinson, Brisbane, Australia.

-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys - and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
_______________________________________________
Flightgear-users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/flightgear-users

Reply via email to