1. #1
    In leading up to the release of BOB, I was just re-reading my favourite Spitfire resources (including The Spitfire, Mustang and Kittyhawk in Australian Service) when I came across the section describing the internal structure of the Spitfire.

    The following paragraph describes the wing structure.

    "Despite the apparently conflicting requirements of light weight, fineness and strength, a very strong wing was designed for the Spitfire and this required a fairly complicated all metal internal structure comprising a single spar made of spar booms joined by thick web plates; forward of that was the area providing most of the wings strength, a very strong torsional box leading edge which could be easily strengthened simply by substituting thicker gauge alloys. The wing was covered in Alclad and control surfaces (including the rudder and elevators) were fabric covered."

    "My GOD!" I yelled! This is classified stuff!! If those Jerries find out this they'll be directing fire on the leading edges of our wings leading to catastrophic wing failure.

    I wonder how faithfully BOB will model the internal structures of the aircraft?

    I know it's a trade off between fidelity of the models and performance but it'd be good to have damage models that accurately reflect the strengths and weaknesses of the individual aircraft.
    Share this post

  2. #2
    BAsed on the 1946 2nd DVD demonstrations and video files they would be HOWEVER the limitation in the game would always be the POLYGON LOAD. The # of polygons would always determine the amount of accuracy of the DM (damage modeling) due to the use of BOUNDING BOXES for COLLISION DETECTION. I do not know if Oleg would be using Adaptive subdivision for collision detection purposes so if not then, it is always a matter of polygon load restrictions because there's just so much polygons a video card can handle in real time and still deal with LOD (level of details) SWAPPING issues based on distance to the camera.

    As you probbaly know already there's LOD involved based on the distance either to the in-game camera or based on the plane geometrys NULL OBJECT/origin reference object. The LOD changes depending on the distance and geometry get swapped out this way. The Bounding boxes also changes based on the LOD.

    BTW these are all based on my knowledge of 3D applications mainly 3DS Max and Maya so how it is exactly implemented in the game depends on the game engine which only Oleg would know but the principle remains the same that it is impossible to have an Ultra accurate DM with 30 planes in a server w/o the accuracy or playability being compromised. A real accurate DM would probably be restricted to 8 planes or less only per server.
    Share this post

  3. #3
    Agamemnon22's Avatar Senior Member
    Join Date
    Dec 2003
    Location
    Montreal
    Posts
    570
    Well, based on WIP shots from a while ago, the damage objects (not bounding boxes)are pretty detailed. Hydraulic lines, fuel, etc, all major components and major structural members appear set up for DM.

    So it should be pretty damn good.
    Share this post

  4. #4
    I don't think you understand, unless things have drastically changed, bounding boxes (which are invisible) are always used for collision detection. There's just no way Oleg can place TONS of bounding boxes for every detailed geometry that is SHOWN. This is plainly impossible due to the fact that this would increase the polygon load since bounding boxes themselves are made from polygons. What is SHOWN VISUALLY is totally different from how those things are accounted for in DM. This is why I wonder if Adaptive Subdivision is being used because if this is used per polygon face the polygonload can be decreased while increasing the collison detection capability within limits.

    There are faster ways to go about this however which is to use RAYCASTING (and other Monte-carlo based occlusion check methods) but this needs to be done via the GPU to make it real fast. Cg (nVidia) and RenderMonkey (ATI) are capable of this and have this capability in the GPU (video card hardware).
    Share this post

  5. #5
    SaQSoN's Avatar Senior Member
    Join Date
    Apr 2004
    Posts
    557
    the limitation in the game would always be the POLYGON LOAD. The # of polygons would always determine the amount of accuracy of the DM (damage modeling)
    The ammount of polygons in the visual model does not have any relation to the physical DM.

    due to the use of BOUNDING BOXES for COLLISION DETECTION.
    Even in the IL-2 bounding boxes weren't used for collision detection. Instead, collision was calculated using special collision objects, which had shape close to respective aircraft parts and internal structures.

    I do not know if Oleg would be using Adaptive subdivision for collision detection purposes so if not then, it is always a matter of polygon load restrictions because there's just so much polygons a video card can handle in real time and still deal with LOD (level of details) SWAPPING issues based on distance to the camera.

    As you probbaly know already there's LOD involved based on the distance either to the in-game camera or based on the plane geometrys NULL OBJECT/origin reference object. The LOD changes depending on the distance and geometry get swapped out this way. The Bounding boxes also changes based on the LOD.
    Sorry to say that, but the text in the quote above is a complete nonsence.

    The fidelity of the collision detection does not depend on distance to the model, or far LOD shapes.

    Certainly, the DM calculation principles will in generally remain the same in BoB, as in the IL-2, but there will be an additional physical factors taken in account. Besides, the number of collision objects per aircraft will be greately increased. Which means, the situation with the Spitfire wings will be possible to reproduce in the game.

    Edit:
    Agamemnon22 is absolutely correct.
    BTW, Andrey, how things going with that object on the background? Haven't heard of you in a while.
    Share this post

  6. #6
    Originally posted by SaQSoN:
    <BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">the limitation in the game would always be the POLYGON LOAD. The # of polygons would always determine the amount of accuracy of the DM (damage modeling)
    The ammount of polygons in the visual model does not have any relation to the physical DM.

    due to the use of BOUNDING BOXES for COLLISION DETECTION.
    Even in the IL-2 bounding boxes weren't used for collision detection. Instead, collision was calculated using special collision objects, which had shape close to respective aircraft parts and internal structures.

    I do not know if Oleg would be using Adaptive subdivision for collision detection purposes so if not then, it is always a matter of polygon load restrictions because there's just so much polygons a video card can handle in real time and still deal with LOD (level of details) SWAPPING issues based on distance to the camera.

    As you probbaly know already there's LOD involved based on the distance either to the in-game camera or based on the plane geometrys NULL OBJECT/origin reference object. The LOD changes depending on the distance and geometry get swapped out this way. The Bounding boxes also changes based on the LOD.
    Sorry to say that, but the text in the quote above is a complete nonsence.

    The fidelity of the collision detection does not depend on distance to the model, or far LOD shapes.

    Certainly, the DM calculation principles will in generally remain the same in BoB, as in the IL-2, but there will be an additional physical factors taken in account. Besides, the number of collision objects per aircraft will be greately increased. Which means, the situation with the Spitfire wings will be possible to reproduce in the game.

    Edit:
    Agamemnon22 is absolutely correct.
    BTW, Andrey, how things going with that object on the background? Haven't heard of you in a while. </div></BLOCKQUOTE>
    Thanks for the clarification, it sounded like those comments were being pulled out of the a**, specially the DM increasing polygon count (??) and the DM mesh being dependant on the LOD mesh.
    Adaptive subdivision for collision detection
    Share this post