I was confused about why we use inverse mass instead of mass. Turn out that it goes back to the fact that division is expensive
ninjab3381
I read about a variant of velocity based Verlet integration called the leapfrog algorithm where the velocity and position are at updated at different times but it turns out to give the exact same values as velocity based and it seems like the same algorithm so ws just curious if anyone had any thoughts on this: https://en.wikipedia.org/wiki/Verlet_integration
zeddybot
An interesting thing to note here is that since we are projecting the points pi onto the constraints Cj at every timestep, the choice of a good solver is essential to making sure that this procedure is fast. Another intesting thing to note is that the velocity always has to updated after adjusting the predictions pi to align with the constraints, otherwise we would be introducing second-order inconsistency into the system.
I was confused about why we use inverse mass instead of mass. Turn out that it goes back to the fact that division is expensive
I read about a variant of velocity based Verlet integration called the leapfrog algorithm where the velocity and position are at updated at different times but it turns out to give the exact same values as velocity based and it seems like the same algorithm so ws just curious if anyone had any thoughts on this: https://en.wikipedia.org/wiki/Verlet_integration
An interesting thing to note here is that since we are projecting the points pi onto the constraints Cj at every timestep, the choice of a good solver is essential to making sure that this procedure is fast. Another intesting thing to note is that the velocity always has to updated after adjusting the predictions pi to align with the constraints, otherwise we would be introducing second-order inconsistency into the system.