Jump to content

Timmo

Members
  • Content Count

    11
  • Donations

    $0.00 
  • Joined

  • Last visited

Posts posted by Timmo


  1. Here are my thoughts.

     

    FSX is a very capable scenery engine- Apart from a couple of issues related to the period in which it was developed (i.e. lack of proper multi-core threading, under utilization of GPU, 32 bit pipeline etc) it still holds its own in the current day IMO, not least of all because of the ability to upgrade most things within the simulator install.

     

    X-Plane 'feels' more like a flight simulator than FSX does....but the scenery engine isn't quite up to par yet (the lighting is infinitely better than FSX and flying very low is generally more immersive but at mid-high altitudes, their procedural rendering can't match the realism that FSX's tiled aerial textures + Autogen give) and I really dislike the interface.

     

    Flight had some really nice additions to the FSX code and would make the perfect base for a new simulator with the addition of global landclass, terrain and Navigation data- The lighting is far nicer (i.e. on autogen objects especially) and the flight model seems far better too....It would have been a hit if MS had simply provided their own marketplace where payware developers could have integrated their products into the 'store' and paid a small fee. This would have provided the post sim sales revenue that was lacking with the FSX release.....they swung too far the other way.

     

    In other words- My ideal simulator would have the flight model of X-Plane, the scenery data of FSX, the scenery engine of Flight + procedural terrain refinement and lighting of Outerra.

     

    Where does this leave Outerra?  Currently Outerra has the capability to provide the scenery base- The new 'biomes' in development should allow more accurate representation of the world (i.e. landclass) along with the other parts in development (importing of vector data/roads, water bodies etc). As mentioned, the use of 3rd party physics/flight modelling libraries provide that portion of a sim and it can already import common 3D object formats. This leaves a decent proportion of the requirements for a flightsim not (yet) filled which begs the question, what should Outerra provide and what shouldn't it provide? If the consensus is that a fully fledged flightsim is too large a project for a small dev team, then the project moves to being split between the base engine (Outerra) and 'everthing else' (possibly created by the community)

     

    Obviously 3rd parties can develop aircraft and improve the scenery/Airports...leaving items such as weather modelling, Airways/ATC needing to be developed within the 'Outerra Flightsim' (or at least the ability to add it)....this does run the risk, however, of a 'moving target' for development- I.e. the base engine breaks older data requiring it to be re-compiled.


  2. It should also be pointed out that, because ORBX uses their own landclass values, textures and autogen, looking at their data to try and understand how to make your own landclass in the Dominican may lead you astray- It is best to try and find out what is happening in your own region by using something like TcalcX to discover which files cover the area you want to change and then examining them in TmfViewer (the default scenery .bgl viewer which comes with the SDK) As Luis mentioned, even trying to find which landclass value matches which texture can be difficult- An easy (ish) way to find out what landclass value looks like (i,e, texture and autogen in all seasons) is to make a grid of landclass squares (say, 5km by 5km) of each landclass value placed over your country. You can then fly over each square and decide which landclass values to use when creating your final coverage.


  3. High resolution terrain, such as that collected from LiDaR does (in its raw form) contain information about building heights which can be compiled with the resample tool into a bgl file.However, there are several problems with this approach:ortho-rectified aerial photos don't contain information to render the sides of buildings (or indeed any face too far away from horizontal)Users would need to run very high mesh complexity and resolution settings....and even then, the highest accuracy LOD doesn't get loaded until very closeSo.....not really useful.


  4. In principle anything's possible but at a cost. Microsoft will have set a budget and the developers will have to work within that. Are sloping runways the best use of the budget compared with all the other featutes that have appeared on wish lists?
    It's also a case of computing cost- rendering an object along a basic 'flat' vector is fairly simple and so doesn't require many calculations to do. Making an object read the terrain multiple times and offset itself from it requires a fair amount of processing power.....power that may be best used in other places.
×
×
  • Create New...