search

xurxodiz / MarioLevels

My entry for the Mario AI Championship 2012, implementing adaptive procedural level generation. Holds 1st place at the all-time points and percentage tally.
MIT License
2 stars 1 forks source link

Design of the tension wave #14

Closed xurxodiz closed 12 years ago

xurxodiz commented 12 years ago

The tension wave will be one of the factors that generate the chances for the different chunks. Issue #8 will take care of its specific weight (probably in relation with the difficulty of each chunk). But this is about how to generate the value for this factor.

Leading theory now is to use a basic harmonic wave w = Asen(bx).

Factor b should make sure that the wave's phase is a divisor of 320, the current stage width, so it both starts slow and ends slow.

Or should it?

This kind of thing should be figured out.

xurxodiz commented 12 years ago

Issue #7 has proposed an FSA/Markov approach. The tension wave then would be completely scrapped (since Markov processes usually take into account only the present state, and probabilities are fixed).

Is this a big problem? Is the tension wave so relevant? Can we implement this idea in the system in another way?

xurxodiz commented 12 years ago

For the moment, it seems like the tension wave will be discarded. The issue will remain opened for a while so it doesn't go out of sight.

xurxodiz commented 12 years ago

In some sense, the tension wave was the birth of the probabilistic transitions and still lives on in them. After high tension chunks, low tension chunks should and do follow. So it is indeed embedded in the current approach, albeit not exactly in the purest form as it was conceived.

I'll close this issue now.