Slope Implementation and Research...

[datahead8888]

This is a task for research & implementation for slopes.

We need to decide:

• Do we want to try to use a library like Bullet, if it does collision handling, for this? (I don't think 3rd party libraries are the solution to everything, but if this is really hard, we should give it consideration.)
• Can we write this ourselves?

I've heard from multiple sources that implementing slopes is hard, so this requires planning regardless of what we do.

Also, if anyone can think up a way to use existing sprites to represent slopes (manipulated via code?), it would save us from having to create a massive number of new sprites to represent slopes.

Additional Information:

• To do coding list: http://www.secretmaryo.org/wiki/index.php?title=Todo_Code
  • Says, "[After the switch to Bullet] Collision detection slope support ( line start and end y position declared in image settings )"
• Forum discussion - http://www.secretmaryo.org/phpBB3/viewtopic.php?f=4&t=898&hilit=slopes
  • Contains some detailed discussion on how to implement them
• Forum discussion: http://www.secretmaryo.org/phpBB3/viewtopic.php?f=4&t=4190&p=26680&hilit=slopes#p26680
  • Cites Bullet2D as a solution for slopes
• This document mentions a slope issue with Super Tux: http://supertux.lethargik.org/wiki/Milestone_2_Design_Document/Styleguide
• I finally found the document that directly talks about slope issues in Super Tux - it's the general collision detection article - http://supertux.lethargik.org/wiki/Collision_detection

[Luiji]

Slopes aren't that complicated to implement. I think that if we wanted to add real physics simulation (e.g. through Bullet, Box2D, or Chipmunk) it would be advisable to do so and then implement slopes since the custom slope code would likely be deprecated during such a transition.

[datahead8888]

I think that if we wanted to add real physics simulation

The physics in SMC aren't necessarily "unreal", though they probably aren't going to be as accurate as what's in Bullet.

it would be advisable to do so and then implement slopes since the custom slope code would likely be deprecated during such a transition

I was wondering if it's still possible to have some geometry / physics algorithms we code ourselves and others implemented in Bullet 2D. The reality, though, is that we will probably never have someone on the SMC project who is as skilled at this kind of work as the makers of Bullet - they were publishing tutorials at the premier graphics convention - SIGGRAPH. I gave up trying to pursue physics simulations in grad school myself because the serious stuff is extremely difficult.
It does make sense to decide how slopes will be implemented long term before spending time on them so as to avoid rework.

Slopes aren't that complicated to implement

As I said, I'm not on the Bullet for everything bandwagon. I guess I just think there's something to be said for implementing some algorithms ourselves given that that's what game development gives you a chance to do; with enough work some customization could be possible. On another note, I think Super Tux Kart switched to Bullet.