Outerra forum
Anteworld - Outerra Game => Terrain & terrain data issues => Topic started by: KelvinNZ on April 14, 2012, 10:25:38 pm
-
Hi Cameni,
what height elevation data is the standalone data? is it 90m SRTM and is this the highest possible data that is available for OT?
-
It's derived from 90m SRTM. Thus far it's the only global dataset available. There's also ASTER that should be 30m, but it's got too many defects to be usable yet.
-
Ok. I have been reading quite a lot about the various data today and was wondering if OT would utilize higher resolution. I was reading the details about a newly released elevation data package for Microsoft Flight Simulator and Prepar3d which explains some manual adjustments made to the data to get it looking more accurate. See the link below if you are interested:
http://www.fly2pilots.com/cms/Ui/Pages/Products/MainPage.aspx?id=2d768af2-35f2-4107-8d22-5ddf9e193b47 (http://www.fly2pilots.com/cms/Ui/Pages/Products/MainPage.aspx?id=2d768af2-35f2-4107-8d22-5ddf9e193b47)
-
Sure OT could use a higher resolution of data, and we don't even have to resort to some hacks as the whole handling is in our hands. There are of course better source data that are commercially available, that we could use if it was economically viable for us as well.
-
Cameni, if there is a website where you have seen this commercial elevation data can you point me to it?
-
A couple of links can be found, for example, here (http://www.pathloss.com/pwiki/index.php?title=Terrain_data). Else just search for elevation dataset [commercial]. Some references are at http://vterrain.org/ (http://vterrain.org/)
-
Hi Cameni,
Can you confirm if the SRTM data used in OT is the same as noted here:
http://srtm.csi.cgiar.org/ (http://srtm.csi.cgiar.org/)
it appears to be version 4, is this the same version used in OT?
What do you think about this site:
http://www.diva-gis.org/Data (http://www.diva-gis.org/Data)
it has all sorts of data per country (global data). Is it possible and practical to draw information from the downloadable files?
See the attachment for the areas that are covered.
-
No. That one doesn't have a suitable license for us. Besides, it also doesn't contain data for areas above 60° north. We used http://www.viewfinderpanoramas.org/dem3.html (http://www.viewfinderpanoramas.org/dem3.html) data which is based on the original SRTM.
-
Cameni,
Can you tell me how I need to be reporting the elevation issues using xnview and the plugin? The only data I have is the Southern Alps not New Zealand in its entirety. If you could let me know what you need to be seeing from this then i can start trying to note all the incorrectly elevated areas in the data.
EDIT:
I was wondering about a more practical way to determine elevation problems in the SRTM data maybe by running a side by side application to compare using OT
-
You can download the tiles from http://rmw.recordist.com/ (http://rmw.recordist.com/) specifying a lat&lon window. Xnview with the plugin (http://outerra.com/xnview/) is one way to view the terrain data. If you find an improper terrain in OT, you can locate it in the source files using the lat&lon coordinates.
Then we would need to fix it. A full fix can be costly (in terms of time), but a simple fix could be possible - depends on the type of the bug. I'll ask ME-DEM guys what software could be used to fix these issues the easiest way.
-
That sounds good. If it is merely a repetitive process and just time heavy then I wouldn't have any issues working through the errors; given that no major programming needs to be done to fix each error. I have quite some spare time I can invest and happy to help, not to mention that i have always had an interest in this sort of thing.
-
Aren't SRTM files already fixed? I mean, an attempt has been made to fix them.
SRTMv4 has been 'fixed' according to the various step described in the link but there will still be problems. I've not encountered any in Outerra but I've no doubt they exist.
http://www.cgiar-csi.org/data/elevation/item/45-srtm-90m-digital-elevation-database-v41 (http://www.cgiar-csi.org/data/elevation/item/45-srtm-90m-digital-elevation-database-v41)
I think you could get away with using Photoshop in png 16 bit format if you had auxillary dems that were not available to the folks who produced the v4. Blend the two terrains together. I'm not familiar with XNview -could you use that as a 3D viewer? Either way, you'd definitely need a 3D viewer to view your Photoshop work.
You could convert between the png and the native via Wilbur, 3DEM, or I could do it for you here painlessly via Global Mapper.
You could of course go track the areas down in Google Earth and eyeball their commercial terrain, and fix them manually but you'd need a 3D terrain package for that. I'd recommend Leveller, World Machine Pro or GeoControl. Wilbur is very capable and free!, but you're limted to 2D. Of the three, Leveller is the most paint friendly and there's a 64 bit version available now.
I've basically done this kind of thing before. I blended manually between our project data and real world data. You wouldn't need the Photoshop stage (but then you might well already have that app), but I'd recommend Leveller to do it.
A programmatic fix would be possible but I think they've already covered all those bases in v4 anyway.
If you feed me the lat long coords or links to files I could open them up, merge them and export them in a useful format. Any chance of posting some pics of the dodgy areas Kelvinr?
monks
-
Just FYI, Global Earth Coverage Available in 2014, TanDEM-X Global Digital Elevation Model, see here :
http://www.astrium-geo.com/na/1068-tandem-x-global-dem (http://www.astrium-geo.com/na/1068-tandem-x-global-dem)
But what price and which datas size for such an accuracy ? I wonder
:D I got the answer for the data size LOL :
" lying in close formation only a few hundred metres apart, the two satellites are imaging the terrain below them simultaneously, from different angles. These images are processed into accurate elevation maps with a 12-metre resolution and a vertical accuracy better than 2 metres. The amount of data generated by the satellites will grow to 1.5 petabytes within three years "
They have 3D Elevation10 (or 30) samples there too...Barcelona area
-
petabytes...? pfftt! ??? I'm thinking though that the dem alone won't come to that. I mean at 100m res the whole Earth is around 15-20 GB? So 10x res should give you 150 GB surely. Hmmm, no, it's squared...so it's x10^2...which is massive but still "only" 1.5 TB.
monks
-
I guess they mean the collected data from multiple runs, not the dataset size.
Raw input data at 90m come to ~80GB, resampled to ~76m & packed are ~12GB though we are losing some details by the bilinear filtering there.
-
Aye, + I imagine they collect other data too which they can use for stuff like the MODIS etc. It'd be kinda wasteful if they didn't while they were up there!
monks
-
To get proper results they have to image several times the complete land surface area of Earth.
There is an explanation why they have to do so at the bottom of their page here
http://www.astrium-geo.com/na/2953-first-tandem-x-coverage-completed (http://www.astrium-geo.com/na/2953-first-tandem-x-coverage-completed)
What is good is that they can record the full Earth many many times, so quality with accuracy will improve more and more.
As long as the SATs remain operative...
We can assume that at the end the dataset will be lower than Petas... but I wonder how big though?
-
I read somewhere that they were using 3 sats now but they're using two here. Maybe them being so close together will enable them to only use two to get the accuracy.
That is some crazy good dems there. Makes me go all fuzzy...What I'm wondering is once these come online, will better dems than SRTM 4 become freely available?
monks
-
Yes you ask the good question :P
I do not think that those European companies (Astrium, which belongs to EADS, which belongs to French and Deutsch Gov and Lagardere Group and others...)
will make anything freely available.
I am from Europe and when you see that our politicians have been able to crash down the fusion between EADS and BAE (British AeroSpace),
you can understand where we are here >:(
-
Monks? aren't you the ME-DEM's man? from Europe too, GB, are you?
-
Thanks!- that's a good link and very informative. Yes, I think 4 is just 3 with fixes. I do share peoples' fascination and love of high quality dems- lookng at dems just does something to my head- can't explain ;D, and to some people, they don't even need a game to enjoy it. Having said that, having tried modelling a large "realistic" terrain I've also become more aware of what's possible. If you're not careful you can end up always chasing the next wave (higher res or whatever), and never actually releasing anything. In ME-DEM we *could*
go to 50m res, or possibly even higher but scripts would need to be rewritten and there would be delays, not least with the extra processing time. Much better to get out a version 1.0 that's as good as you could feasably make it, and accept that it ain't perfect.
monks
-
haha!..you made me laugh "ME-DEM's man". Yes, from GB mate. You're from France I read. ;) You seem to know a fair bit about dems and GIS stuff, sources and what not.
monks
-
;D I'm French but I'm in Luxembourg, "crushed" between France, Germany, Belgium.
Thks from the compliment but I'm not so versed in GIS and other DEM, DTM, DSM stuff : I'm sure I'm much less skilled than you there !
I just like 3D technos, Comp Graphics, simul, games, mil. stuff and many many other things.
I'm an "old" Robotics/CAD/.... guy with eclectic interests (does eclectic means anything in English?).
-
They still manage to get those radio signals out though- radio Luxembourg has kinda entered the cultural psyche over here.
I'm not really skilled, just persistent. :)
Yes, I can see why you would be drawn to Outerra- well I can see why anyone would, but the
possibilities for sims in Outerra is great plus it kinda straddles a couple of territories- games and GIS.
Eclectic, yes gotcha...is that another word that we English stole? ;D Sounds like could be French origin.
monks
-
Do you mean that you listen to Radio Lux ( RTL or RTL 2 ) in the UK?
as long as I know RTL2 in German is still diffusing good 70s/80s rock/pop songs and here they have many auditors from FR-DE-BE-NL
However I am pretty sure you have much better radios there in the UK for that kind of good music.
Anyway Luxembourg is more famous for its Fiscal and Banking system than for anything else.
Yes I fully agree about OT having great potential for GIS, real time Simul, games, etc...
I am not so interested in game engines, I'm more focused on professional applications or serious games (for example Prepar3D in flight simulation)
Please can you explain what are you doing at ME-DEM ?
Cheers - Denis
-
BTW I have just quickly tested the TANDEM-X sample GeoTiff from Barcelona Spain,
they say it is 10m resolution and the file can be seen and downloaded here :
http://www.astrium-geo.com/na/1194-sample-imagery (http://www.astrium-geo.com/na/1194-sample-imagery)
Then I imported it in GMapper, re-exported it to KMZ low res,
then made 2m vector contours lines and exported them to KML without compression.
( I must admit that it is the 1rst time that I use GMapper... so maybe I did not use it at max perfs ??)
There are funny little ridges that you can see in the blue regions : in fact they correspond to sur-elevated highways,
so when they say that is Digital Terrain Model, according to me, these man made structures should had been cleaned...
Then I wanted to test it and compare in Google Earth, so :
I imported both files in it :
I am very surprised to see that GoogleEarth terrain in that area is quite good, it seems that Earth is at or close to 10m there ?
as long as I understand GoogleEarth is using a topo source from the Catalunian Region, so the good res... and they also have "elevated roads" artefacts on their datas, ;D
I wanted to look to importing in OT the corresponding datas (collada importer) but unfortunately I can't, cause an issue on my old GPU...If someone could take a look, of course, the goal is absolutely not to criticize because we know what OT terrain is (resolution + fractalisation), just a question of curiosity.
Below my screen dumps related to the below procedure (GoogleEarth images zoomed proximity to the San-Climent-de-Llobregat town) :
(http://www.cadasys.com/files/GlobalMapper-with-DTM-elev-10m-Barcelona-TANDEM-X.jpg)
(http://www.cadasys.com/files/Gearth-with-DTM-elev-10m-Barcelona-TANDEM-X.jpg)
(http://www.cadasys.com/files/Gearth-with-lines-10m-Sant-Climent-de-Llobregat.jpg)
-
Luxembourg radio...yes, I think it's just a bit of a novelty- that we can pick it up! ;D
We're building a dem of Middle Earth. Simple as that really. That includes textures of course, and once
the dem is released for Outerra, we hope people will add stuff to it- buildings, etc, whatever.
We have all of the core data in Global Mapper, dem + vector maps for stuff like roads and vegetation distribution.
It's georeffed according to what Tolkien stated in his letters, Minas Tirith = Florence, Hobbiton = Birmingham, UK.
The dem is currently at 100m res. Not much point in going larger than that until we get the first version out
the door. Also, Outerra fills in the small scale details which helps greatly. One of the original objectives was to have good hydrology on the dem- that is for the main rivers on the Tolkien maps to flow correctly in a physics
sim. It's a tall order and Robes (one of our members) has written a physics sim so we can test it. It won't be possible though for all of the rivers, just potentially the main ones.
In Global Mapper did you have the simplify contours slider set? You might want to check that that was set to 0.
Hmm, yes, they are very straight- they're definitely not natural and they're not artifacts produced in the export.
I've never had anything like that when using GM. So really it's a DSM. I don't see how analysts could miss those.
Maybe this is still not an exact science. I mean they really have to cobble together full coverage dems from what's out there. It must be difficult. If it's difficult for Google, imagine what it's like for everyone else! The terrain is great though in Google- so are they using 30m res?
monks
-
We're building a dem of Middle Earth. Simple as that really. That includes textures of course, and once
the dem is released for Outerra, we hope people will add stuff to it- buildings, etc, whatever.
Hummm, nice project indeed !
In Global Mapper did you have the simplify contours slider set? You might want to check that that was set to 0
You're right, simplification was at a default value of 0.1 I think
I have just quickly put it at 0 and re-exported the contours, but does not change a lot
ve never had anything like that when using GM. So really it's a DSM. I don't see how analysts could miss those.
I know that this Tandem-X goetiff sample is on the Astrium website since quite long, so I suspect it comes from a very early recording without real check,
and I hope that they normally provide better 10m DTM ?
as for what I'm seeing in GoogleEarth, for that specifi location, I would say that they are at the same resolution.
If it was at 30m, they would not have the same artifacts on the roads on bridges... would they?
What is sure is that they do manage well in having different resolutions : for example their DEM in the Luxembourg area is pretty bad, I even wonder if they are at 30m here.
-
Yes, they're early images, and if it takes up to 4 years processing them to commercial standard...! The analysis will no doubt correct those artifacts.I imagine blending between different resolutions of dem data is a bit of a nightmare- is there any evidence that Google are doing that? I thought that Google is used for some scientific work- the Google pro package? Dunno.
Hey, do you remember NASA WorldWind? We were looking for someone to get the ME-DEM data into that because it allows user created dems (Google does not). We had some interest from a guy who works for NASA (in UK I think) a few years ago and he said it would be ok if we put it on a server there.But we did not have the dem to good enough standard- until now really. :( That would have been Fab!
We have a guy on the project who set this up. We're intending to add more map layers to it. :)
http://66.172.33.110/openlayers2.html (http://66.172.33.110/openlayers2.html)
monks
-
I have no sufficient time to check properly what Google Earth is doing in resolution, but I think they do have different resolutions !
If you look at Honolulu, HI, National Cemetery of the Pacific (it's inside a crater), I think they are at much better than 90m or 30m.
It is not so hard to mix various resolutions if you get the ressources to do so...
Middle Earth in OT time line is ?
-
That is surprising, not for typical use of it, but from more scientific use of it. I suppose it's something of an inexact science dealing with dems. Mind the gaps!
HA! Punchbowl crater really is bizarre! Looks like they got their own SuperBowl.And I never knew just how crowded Honolulu is. Fascinating place indeed. Yes the imagery is more detailed around that whole area but move around the island to Makaha Valley and it's less.
Middle Earth? Oh, say 5000 years ago from present Earth or something. So translate that and you'd get a time in OT. Have you got any plans to add any models to Anteworld?
monks
-
;D yes I agree, Hawaï is fascinating ! I'd like to go there once again :-X (I've been there years ago during wedding trip to the Cooks Islands, I saw nothing from Hawaï, it was a quick refuel stop for the plane, at night : LOL)
Would be nice to find out where Google is at its best resolution... worth laung few searches to see if someone's talking about this.
Plans to add models in Anteworld : yes we should in particular test and import collada cities samples ASAP; but we lack time and need to upgrade our CPU/GPUs, we have AMDs and NVidias but oldies ones, so step by step.
I'm also professionally engaged on several things...
-
I hear ya ;), I only have a nvid 460 here- I mean it's pretty middling really. Yes, at some point I'd like to try the importer out too- apparently it's very robust, even for an early development. ...heh, too bad about Hawaii. It was very inconsiderate of them to arrange their wedding like that >:( haha.
monks
-
------
-
main workstation here is having Sapphire ATI Radeon HD2600Pro 512MB with two DELL 24" HD monitors, but this board is now too slow
we should go for an Nvidia GTX660 or better, but I think it is worth buying a new full CPU... we'll see
we also have other computers with Nvidias, desktops and laptops, but they are worse ;D
BTW I've just played a bit with the flight simulator which is in GoogleEarth and it is not bad at all, I had never looked before, but it is a good surprise.
I'm also cheating a bit in offline using the Gearth cache files at their Max, LOL.
I look forward to playing with Anteworld on a stronger platform :P
-
Was that first post moderated? ;D
Those two dell monitors sound really nice. I'm on an old 21" CRT...if we had an earthquake the house would fall down around it. lol
I'm still in the process of downloading the Outerra Earth data...been at it for weeks! I'm lucky in that I have the terrain data for ME-DEM on disk so I never get any lag with that. Couldn't find any info on highest res data in Google- but I'm sure it's out there someehere. I suppose if you follow the Google Earth updates closely you'll how a good idea.
monks
-
Well the maximum disk cache size of Google Earth is at 2 Go which is much too low for good testing ...
I played a bit with places where they have large 3D models (for example the Los Angeles area, and the Bordeaux city in France)
but the viewing experience could be much better if they would increase these cache limits. OT should be much more efficient in that kind of usage...
Yes, it's a pitty :( we have nice monitors but slow CPU/GPUs ... time to upgrade !