Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[Varldsligist]

First of all hello! I'm new here.
I bought the engine a year ago after I saw a post on Reddit, and let me just by saying it's truly a work of art. I know it's far from finished and we're still just seeing the early stages of development, but I think it's going to be spectacular once you guys implement clouds, proper trees and a few other basic elements.

The main reason I registered here is that I couldn't find an answer to a question that has been bugging me lately experimenting with the engine; how are buildings supposed to be added? I realize there probably isn't one answer to this question, however I would love to hear what you have been pondering so far. Traditional modeling can be very time consuming and takes years to master and to make even the most basic city by hand [CAD] can easily take a months of hard work, let alone making enough cities to populate a planet at the scale of the earth. This could of course be aided by "crowd" modeling (people working together) but frankly the tools aren't there yet and my guess is the result would be uncoordinated and messy. They would also be forced to work toward the same vision. So, in keeping with the spirit of the engine, is there a procedural solution to making builings, cities and other man-made objects? I have seen a couple of videos of people choosing this route and I came away less than impressed, take this for instance: There are a few other examples out there, but they all have that sterile, generic look to them.

I also had the idea of importing buildings from Google Maps as illustrated here: http://www.google.com/earth/learn/3dbuildings.html but quickly found that the models (surprise!) were lacked proper detail. I have yet to try out this route in Outerra, but I have done it in Unity with mixed results using models that others made before. Note that this is quite time consuming and doesn't really fit this application. The third and last suggestion/question concerns Google's Project Tango, which in short is (please correct me if I'm wrong) a way to use optics and other sensors to map your surroundings, for instance a building. This has been done before using expensive professional equipment, but they seek to get the technology in the hands of ordinary consumers by early next year. Presented here: .

What do you, the Outerra dev. team and modders think? What's your vision for Outerra regarding buildings? Will we some day be able to import and to freely roam around an exact replica of our home town/village? 

Let me end with a quick random question: is there a way (simple way) to change the time of day? I seem to always end up somewhere where there's night.

[M7]

I saw this guy solution a while ago
http://paranoidpuppets.com/001/index.php?option=com_content&view=article&id=107:uberbuilding-parametric-buildings&catid=47:tools-and-houdini-works&Itemid=111

It looks really promising if you ask me. A parametric tool  to create buildings then a procedural  tool for creation city. Unfortunately, it looks like he abandonned his project but i find these are the kind of tools that could make convincing cities possible in outerra.

[Levi]

Let me end with a quick random question: is there a way (simple way) to change the time of day? I seem to always end up somewhere where there's night.

Click on 'Locations' button, or simply press F2, then click on 'Sun' tab and adjust the sliders.

[HiFlyer]

I always figured that with OSM being mentioned from time to time, that the city implementation would be something similar to what X-Plane 10 is doing now. In a flightsim there's no reason for buildings to be over detailed, but in a first person shooter, buildings might have to be detailed enough to be entered.

I suspect that Outerra might be eventually divided up into custom worlds (or maybe sectors) suited to individual applications.

[PytonPago]

I always figured that with OSM being mentioned from time to time, that the city implementation would be something similar to what X-Plane 10 is doing now. In a flightsim there's no reason for buildings to be over detailed, but in a first person shooter, buildings might have to be detailed enough to be entered.

I suspect that Outerra might be eventually divided up into custom worlds (or maybe sectors) suited to individual applications.

... well, having a such deep procedural generation, that would go to interiors and their details, would be great. Especially if there was some parameter to say, on witch level to stop for certain applications.

[Varldsligist]

I saw this guy solution a while ago
http://paranoidpuppets.com/001/index.php?option=com_content&view=article&id=107:uberbuilding-parametric-buildings&catid=47:tools-and-houdini-works&Itemid=111

It looks really promising if you ask me. A parametric tool  to create buildings then a procedural  tool for creation city. Unfortunately, it looks like he abandonned his project but i find these are the kind of tools that could make convincing cities possible in outerra.

That looked really cool, albeit it's not replicating real life buildings, but rather interpreting or imagining new ones. I was dissapointed to note that according to his description, he now works at Ubi Soft which in turn means that he's probably busy doing that. I also don't know what Outerras policy is concerning buildings - accurate replicas or interpretations of real life citiies? I would hope for the former.

I always figured that with OSM being mentioned from time to time, that the city implementation would be something similar to what X-Plane 10 is doing now. In a flightsim there's no reason for buildings to be over detailed, but in a first person shooter, buildings might have to be detailed enough to be entered.

I suspect that Outerra might be eventually divided up into custom worlds (or maybe sectors) suited to individual applications.

From my [very basic] understanding this engine has the potential to do high detail building without it being too taxing on your system. From a flight sim:ers perspective this is not needed (but perhaps wanted?). I would hope for Outerra becoming a basic, solid platform which you, or a noob like myself, can easily mod to fit your specific needs.

@Levi
Thanks! I'm embarrased to have overlooked that before. Also, I love your work on the Basler. I imagine flying that once my DK2 arrives

[Levi]

@Levi
Thanks! I'm embarrased to have overlooked that before. Also, I love your work on the Basler. I imagine flying that once my DK2 arrives

No problem man
Thanks! Glad you like it Also, OculusRift users had some problems with vanishing instruments on that plane (maybe on the Ilyushin Il-14P too). But I think I know how to fix that now. Just tell me if you have troubles with that, and I'll try to update it when I can.

P.s. Sorry for the off-topic

[cameni]

Ideally there will be a procedural generator that can take external vector data from sources like OSM (floor polygon, height, type) and produce buildings in several LOD steps, starting from a very rough one, good from the distance & air, through one that's sufficient for flight sims, to gradually refined ones suitable for FPS, with interiors and all.

That looked really cool, albeit it's not replicating real life buildings, but rather interpreting or imagining new ones. I was dissapointed to note that according to his description, he now works at Ubi Soft which in turn means that he's probably busy doing that. I also don't know what Outerras policy is concerning buildings - accurate replicas or interpretations of real life citiies? I would hope for the former.

Trying to create replicas usually means creating custom models and textures. That's a huge amount of data for any usable detail. Automated processes exist, but they are costly to acquire and costly to render and stream, and/or else their resolution is insufficient.

OT is not (and it isn't meant to be) a streaming engine, so we definitely want to go the procedural route, with generators tuned to produce typical localized output. That has the ability to be refined down to any level of detail required, while also producing all levels of detail needed for efficiency.

Of course, even though it will be creating believable cities, it cannot be considered making "accurate replicas". There are ways how to meet the two approaches, though. We are in contact with a company acquiring and processing city imagery, and there's an idea to develop an algorithm that can select the best fitting procedural model (i.e. a procedural generator type and parameter set), that best matches the captured 3D content, reducing (compressing) it to a few procedural parameters.
You can always place custom models for unique and widely recognized structures, which usually have a unique pattern anyway, and it wouldn't be viable to parametrize them.

[2eyed]

Let me end with a quick random question: is there a way (simple way) to change the time of day? I seem to always end up somewhere where there's night.

I know it's OT,but:
If it's not too complicated, please cameni give us two keyboard keys for forth and back adjusting TOD.

[HiFlyer]

I'm still wondering about implementation of bridges/overpasses: or will all the roads in Outerra be flat ribbons ala' FSX?

There are so many approaches to the subject! I wonder where Outerra will go.

http://nccastaff.bournemouth.ac.uk/jmacey/MastersProjects/MSc2010/06ChaitanyaKapu/thesis.pdf

http://wiki.openstreetmap.org/wiki/OSM2World

http://ggg.udg.edu/publicacions/UsersWebs/ProceduralBridgesSIACG2011/

[Varldsligist]

Ideally there will be a procedural generator that can take external vector data from sources like OSM (floor polygon, height, type) and produce buildings in several LOD steps, starting from a very rough one, good from the distance & air, through one that's sufficient for flight sims, to gradually refined ones suitable for FPS, with interiors and all.

That looked really cool, albeit it's not replicating real life buildings, but rather interpreting or imagining new ones. I was dissapointed to note that according to his description, he now works at Ubi Soft which in turn means that he's probably busy doing that. I also don't know what Outerras policy is concerning buildings - accurate replicas or interpretations of real life citiies? I would hope for the former.

Trying to create replicas usually means creating custom models and textures. That's a huge amount of data for any usable detail. Automated processes exist, but they are costly to acquire and costly to render and stream, and/or else their resolution is insufficient.

OT is not (and it isn't meant to be) a streaming engine, so we definitely want to go the procedural route, with generators tuned to produce typical localized output. That has the ability to be refined down to any level of detail required, while also producing all levels of detail needed for efficiency.

Of course, even though it will be creating believable cities, it cannot be considered making "accurate replicas". There are ways how to meet the two approaches, though. We are in contact with a company acquiring and processing city imagery, and there's an idea to develop an algorithm that can select the best fitting procedural model (i.e. a procedural generator type and parameter set), that best matches the captured 3D content, reducing (compressing) it to a few procedural parameters.
You can always place custom models for unique and widely recognized structures, which usually have a unique pattern anyway, and it wouldn't be viable to parametrize them.

Music to my ears! I'm also thrilled about the highlighted part - if you can get rid of the main bulk of buildings and structures that leaves the fun parts i.e landmarks and such to be done by hand. I don't know if this is relevant, but the future of high-detail maps looks bright: http://www.engadget.com/2014/08/10/high-res-imaging-satellite-launches-august-13/ Although I assume OT already has it's raw data in place. Have you considered incorporating laser mapping/Project Tango or something along those lines? I'm aware of the fact that it's currently prohibitively expensive, but do you ever see that taking off and becoming a useful tool?

[cameni]

Well, we don't plan to get involved in the mapping business, and satellite imagery and point cloud scans are not (directly) useful for us. Brute force scanning produces tons of data that have to be reduced before becoming usable. For companies that have large server farms and huge internet pipes at disposal it's currently easier/cheaper to process the data only relatively lightly, with small compression ratio, and streamed to browsers or light clients.

For us (for the detail and quality required for games and simulators) these data would have to be heavily processed - doesn't matter whether automatically or by crowd effort. OSM and similar vector data sets are much more usable for use, even though they are usually not made with a procedural generation in mind and the data are often incomplete and/or using ambiguous schemes.

[PytonPago]

 ... such hi-detail level laser/satelite surface scanning of the earth is still at beginning. It will take some time till the data is in landmass coverage great enough (and probably some thousands of UAVs around the world) and processed rightly for performance and data compression/reduction for some implementation of this kind. Also, there will still come a lot of companies in big and small scales, that will lead this mapping technology further, so waiting for them to make this data finished for just implementing by 3-rd parties will be better anyway.

... or were start a OT crowd-fund for a space program and a 6 times the size fleet of (live feed weather and surface mapping tech fitted) UAVs as the UK and US army have. And not to mention to get involved into the space-race for Moon and Mars !

Now ... who is going for the first manned mission there to scan and measure all the new craters ?