Jump to content

Chris Paulson

  • Content Count

  • Joined

  • Last visited

Community Reputation

3 Neutral

About Chris Paulson

  • Rank
    Advanced Member
  • Birthday 02/24/1970

Profile Information

  • Gender
  • Location
    Belper, UK
  • Interests
    Rallying, 3d games

Recent Profile Visitors

The recent visitors block is disabled and is not being shown to other users.

  1. I haven't done it yet but I believe you do this:- Create body/physic shapes to match bones. ie. cylinders for arms legs etc sphere for head. Create joints between physic bodies to imitate joints of real body. At rag doll time match all joint positions to current animation and let them fall. As they fall match all joint rotation with rotations to the bones in the skeleton of the body.
  2. Indeed - you can have no LW command issued in a thread while another LW is occurring in the main thread. You handle it this way:- mainLoop:- { gameLogic(with non LW commands) mutex(LW) { updateworld networkstuff renderworld } render none LW stuff flip } thread:- loop:- { non LW commands mutex(LW) { LW network commands - which cannot block or wait } sleep }
  3. LW is not thread safe so you need mutex's around sections of LW commands.
  4. Have you managed to get SDL and leadwerks to render to the same window? If so I'd like to see example code (sorry to hijack the thread). Threading with mutexs around LW commands would be best way to handle networking using LW commands. The boost stuff is good for threads, I've got example code if you need it, although not with network stuff.
  5. I would stay away from neural nets, they are for academics not real world games. Check out: - http://gamma.cs.unc.edu/RVO2/ http://digestingduck.blogspot.com/2010/10/rvo-sample-pattern.html Also:- http://aigamedev.com/insider/discussion/motion-planning-thoughts/ and http://aigamedev.com/open/articles/simulating-crowd-flow-dynamics/ but you'll have to register to see these, but if you're doing a game with real game play you should already be registered...
  6. It's still ongoing and it's getting closer. I've redone my navmesh code to get rid of opensteer and use the new RVO/steering in recast as it handles crowds better. My character can walk around and place it's feet on the terrain with 80% success. I just need to handle extreme slopes and stairs better. The blending of idle/walk/runs are now fairly smooth with minimal foot sliding. There's still plenty of work to be done and it all comes down to how much time I can find. I crashed my rally car this weekend so I need to spend time on repairing that
  7. I base my animation speed on the distance travelled by the entity so as to prevent foot sliding. Gives a better effect than "guessing" frame frig factors.
  8. Couple of vids you might like (both using recast): - My link and http://digestingduck.blogspot.com/2010/08/some-more-movement-progress.html
  9. How are you doing this, by setting the matrix? I'd be interested in seeing the code for this.
  10. SetEntityMatrix: - Perhaps, if you could maths to work it out but the matrix is in global space is it not? Also the Vec16 structure is undocumented.
  11. There's a get but no set missing in the API. There should be: - SetQuat( TVec4, global/local )
  12. For me I find one limiting factors is the amount of artwork available especially decent animated characters/models available with a workable amount of animations. I search the internet for decent stuff but I only ever find models that have about 10 animations which is about 90 short of what you really need.
  13. Thanks... but.... where is the code?
  14. This is an untested idea which is complicated but might work. 1) Place camera high (some height depending on grid resolution needed) in sky pointing at ground. Place camera in orthogonal view mode? 2) Render scene to texture/buffer 3) Process depth buffer - this giving you a height map grid (not sure how this is done - this is the tricky bit) 4) Set rain emitters to only fall distance from camera height to height in height map grid
  • Create New...