First cultivation tests...

  • You need to close some tags:

    Code
    1. <[list_tmsimulator_view_position][view_positions][]
    2. >
    3. <[list_tmsimulator_scenery_cultivation][cultivation_files][]
    4. <[tmsimulator_scenery_cultivation][element][0]
    5. <[string8][filename][cultivation]>
    6. <[bool][auto_height][true]>
    7. >
    8. >
    9. >
    10. >
  • very nice job for Lisboa.


    I havn't much worked yet on lights, but there's still much to do. For instance deeper use of osm street maps to set different layers of lights.

    Since there's no ground traffic in AFS2, we should cover highways with a wise mix of red and white dots, because dense traffic is what you see best in night flight. What is not alight must be pitch black.


    Cheers

    Antoine

    Config : i7 6900K - 20MB currently set at 4.00GHz, Cooling Noctua NH-U14S, Motherboard ASUS Rampage V Extreme U3.1, RAM HyperX Savage Black Edition 16GB DDR4 3000 MHz, Graphic Card Gigabyte GeForce GTX 1080 8GB, Power supply Corsair RM Series 850W, Windows 10 64 bit.

  • Thanks Antoine


    The draw distance in AFS2 is amazing.

    Here, i'm leaving Lisboa behind



    By what little I could understand of SceneProc, I think that it can be done, with some patience. It would also be cool if we could randomise slightly the color and intensity of lights in the same category



    Nuno

  • You need to close some tags:

    Code
    1. <[list_tmsimulator_view_position][view_positions][]
    2. >
    3. <[list_tmsimulator_scenery_cultivation][cultivation_files][]
    4. <[tmsimulator_scenery_cultivation][element][0]
    5. <[string8][filename][cultivation]>
    6. <[bool][auto_height][true]>
    7. >
    8. >
    9. >
    10. >

    Thanks ! Il give it a try.

  • Great. Those can be considered quite small file sizes when compared to the FSET higher definition files that I have been doing lately. I am eagerly awaiting seeing some of your work on my pc. :)


    This is such a dramatic improvement to the nothingness that we had before. Absolutely amazing. Thanks again for your work.


    Regards,

    Ray

  • Yes , wow, it looks like this scenproc is a real game changer, at least for us the user/tinkerers. Really sad we have to duplicate so much work.

    I mean this is being incrementally better every day. First Lisboa, looked good, at night looked unbelievable. Now Munchen with different light colors and Saanen with custom textures.

    And if a city like Munchen takes 150MB, then we can have a huge rural area fully populated for the same thing.

  • It's too bad that the quantity of OSM data for the US is very spotty. I did my area at Oceanside California and unless I set the filters wrong (just took defaults), I got just a few housing subdivisions and lights just for major freeways. However, the ScenProc process worked fine and does really help - I love that all the buildings are registered so well - and it will be nice to try different scripts. I would really like to have Munich posted for test purposes - I would like to see how such a dense area slows down the frames per second since I like to fly with VR where I can't afford too much slow down. It is just amazing how many polygons and textures our Nvidia / AMD cards can handle at 100 fps or so!


    Dave W.

  • by compiling Switzerland with some 10 millions buildings in tiles if 1 deg x 1 deg I keep 250+ FPS, but still a freeze of ca 2 seconds when loading the next zone, because each zone laods in 1 step.

    Moreover, currently cultivation must be linked to an airport, thus the load zone must be adapted to the tile size, otherwise the neighbouring tile won't load before you're overflying it.

    These are hopefully things IPACS will improve when defining the new binary file system.


    The most interesting thing with current tests is people gain experience with ScenProc, which is good to make cultivation coverages once IPACS release their binary format...


    Cheers

    Antoine

    Config : i7 6900K - 20MB currently set at 4.00GHz, Cooling Noctua NH-U14S, Motherboard ASUS Rampage V Extreme U3.1, RAM HyperX Savage Black Edition 16GB DDR4 3000 MHz, Graphic Card Gigabyte GeForce GTX 1080 8GB, Power supply Corsair RM Series 850W, Windows 10 64 bit.

  • by compiling Switzerland with some 10 millions buildings in tiles if 1 deg x 1 deg I keep 250+ FPS, but still a freeze of ca 2 seconds when loading the next zone, because each zone laods in 1 step.

    Hi Antoine,


    I believe the freeze is due to the size of your tiles.

    From what I could observe in the log files, AFS2 loads all sceneries (.tsc files) that are in a radius of about 25 kilometers from the aircraft position. I would recommend not making tiles bigger than 15 to 20 kilometers.

  • Hi Antoine,


    I believe the freeze is due to the size of your tiles.

    From what I could observe in the log files, AFS2 loads all sceneries (.tsc files) that are in a radius of about 25 kilometers from the aircraft position. I would recommend not making tiles bigger than 15 to 20 kilometers.

    Hi Sylvain,


    The freeze is definitely due to the size of the tiles, because they're currently loaded as a pack.

    In the tsc files you define the load radius with the airport size, I assume it's in meters. For the 1 deg x 1 deg tiles I tried the of 60'000 which is a little bit bigger than half a cell size.

    Flying around various places the next tile seems to load before being visible , but the lack of slew mode makes efficient scenery testing difficult.


    Setting the value to 5'000, the next tile won't load before you nearly reach the center of the next slide. Thus you fly without cultivation and, suddenly, it loads below your buttocks.


    Of course, if AFS2 keeps with such a loading behaviour, the only solution would be to dramatically reduce the tile size, probably similar to MSFS LOD13 autogen tiles (ca. 1.2 x 1.2 km) and load/unload them according to the LOD radius scheme.


    Knowing that in Aerofly FS2 current cultivation concept you'll need for each TOC file a corresponding TSC file defining the center point of the tile as a ghost airport and a radius, that would be definitely a no go. A small country like Switzerland represents approximately 25'000 tiles of LOD13 size, forget it.


    The only solution - and IPACS is obviously working on it - is to adapt the concept, how TOC files are saved (binary ?), loaded and unloaded, something similar to current default vegetation which works fine.


    The problem with binary toc file format is, if we need a compiler for the ExportTOC step, we also need a decompiler ImportTCO function, otherwise there's no way a cultivation zone can be edited.

    Without such a function, it will be impossible to add anything, be it a simple 3D landmark, without conflicting with cultivation underlayer.

    It will thus also be impossible to add an airport, to clear inaccurately placed cultivation (e.g. trees on highways) or to correctly merge interfacing tiles where 2 adjacent sceneries meet, because you never know what data the other editor used and where they stop.


    The only workaround would be a system of exclude polygons, provided that it allows to place new cultivation on top and make it visible, thus work with priorities.


    Hope IPACS will take this into account to make it a success.

    Cheers

    Antoine

    Config : i7 6900K - 20MB currently set at 4.00GHz, Cooling Noctua NH-U14S, Motherboard ASUS Rampage V Extreme U3.1, RAM HyperX Savage Black Edition 16GB DDR4 3000 MHz, Graphic Card Gigabyte GeForce GTX 1080 8GB, Power supply Corsair RM Series 850W, Windows 10 64 bit.

  • Folks, please don't speculate too much about the cultivation file and it's shortcomings yet. Cultivation is a work in progress and the ability for you to create text files has just been done to allow experiments. The text format is too big and slow for real world scenerarios.


    As mentioned before, expect lots of changes in the future until we have decided on how to handle multiple cultivation files.