Re: [tuxracer-devel] Roadmap
Status: Beta
Brought to you by:
jfpatry
From: Steve B. <sjb...@ai...> - 2000-03-18 17:11:02
|
Jasmin Patry wrote: > > I've sketched out a roadmap (list of objectives) for Version 1.0. I'd like > some feedback on these. I think it's important to define a clear vision for > 1.0 so that the development has some direction. Here's what I propose: We talked a while back about using my PLIB libraries (especially SSG) in TuxRacer, if you did that then many of your objectives would be easy to do: > * View frustrum culling (performance enhancement, will help support larger > levels, especially wide ones) SSG does this automatically as it renders. > * Better meshing of the terrain (you may have noticed that some sloped > surfaces are smooth while others are rough -- this will fix that) This is a general problem with regularly gridded terrain (which is one reason I try to avoid using it). Essentially, you'd like to skin the terrain with quads - but they have to be planar. When you split the quad into two triangles, it's hard to know which diagonal to choose when you do the split. In fact, you simply can't tell unless your original height data is at higher precision than the resulting triangle grid. I suggest you get people to paint their terrain images at (say) 4x the resolution you intend to skin the terrain. Then your skinner can test which pair of triangles are the 'best fit' to the (say) 16 height points in the original painted image for that quadrilateral. However...when you start splitting some triangles one way and some the other, your triangle mesh generation techniques go to hell...ugh! You also start to need more general collision detection and so on. Better still (IMHO) is to go to irregular triangle meshes and generate more triangles in areas of high curvature, etc. This is a REALLY deep subject though and it gets you into dynamic skinning techniques where you use more triangles close to the eye, etc, etc. > * Sound and music. I haven't thought about this much (and I have no > experience with it). This is REALLY easy with PLIB. You only need a couple of lines of code to set it up - then one line to load a sound effect or an entire music file and one line to start it playing. Getting free sound effects and music off the web is pretty easy. > * Health power-ups -- floating objects (octahedrons, probably) that Tux > must hit to get extra health If you used SSG to add general 3D objects then it would be easy to make these be more attractive shapes. In my Tux game, I have spinning herring of various colours for this (or is that just a red herring?) > * Joystick support Again, PLIB makes this easy. > * Paddling (Tux pushes on the ground to speed up) This is the most important think IMHO. It's very bad when you get 'stuck' in a tiny hollow with no way out at all - and it's almost impossible to design levels where that can't happen. > * Some extra Tux animation during race (rotates head, feet flap in wind, > etc.) Easy in SSG > * Polygonal objects can be placed on the course, basic collision detection > works with these Easy in SSG > * Multiple tree types (different colours in trees.rgb file, or separate > rgb files?) Once you have 3D objects, trees become just another 3D object so this is a no-brainer. > * Built-in level editor I don't think you need this. > Please give me your feedback on this plan. Is there anything that you think > should be delayed until 2.0, or are there things missing from 1.0 that you > would like to see? Now's your chance... * Scheme where you need to reach a certain time in order to 'beat' a level so you can go onto the next one. * 'Ghost' - most racing games can record your last game and replay it as you play again (usually with your last run shown as a translucent version of yourself - hence the term 'ghost'). Thus you play against your previous run. This is VERY useful for improving your play and tuning your run to beat a tough level time. -- Steve Baker http://web2.airmail.net/sjbaker1 sjb...@ai... (home) http://www.woodsoup.org/~sbaker sj...@ht... (work) |