Original Post
I've posted a few times before about projects I work on which involve physical simulation of human tissue (organs & blood vessels). It's not really game-tech but since we target mainstream consumer PCs, real-time engines seem the best option.
Many games have some soft-body stuff, but typically it's simple cloth and used for visual effect rather than realism. Human tissues tend to be non-ideal as well since they are often tough and non-stretchy, and the more stiff a system becomes, the harder it is to simulate stably.
Anyway, I'm interested in any opinions which of the well-known engines (open-source or otherwise) are best suited to such work. We've tried bullet and it just seemed to fall apart on stability as well as performance. PhysX continues to look impressive but has minor issues such as Mac support.
We do target consumer PCs but as it's typically a niche market selling to corporations, we do have scope for being able to say "you need a $150 nVidia GPU" - just not forcing people to get special gaming rigs or whatever. For that reason, systems which have hardware acceleration seem preferable purely since the amount of calculation can be pretty steep.
I suppose we're focusing on the list posted in this board's sticky thread, rather than anything too obscure. It either needs commercial backing or a healthy community, some small OSS project that might suddenly die isn't too much help.
As always, links to other sites discussing this stuff are welcome... but I've yet to find anywhere better than GD for these questions.
Many games have some soft-body stuff, but typically it's simple cloth and used for visual effect rather than realism. Human tissues tend to be non-ideal as well since they are often tough and non-stretchy, and the more stiff a system becomes, the harder it is to simulate stably.
Anyway, I'm interested in any opinions which of the well-known engines (open-source or otherwise) are best suited to such work. We've tried bullet and it just seemed to fall apart on stability as well as performance. PhysX continues to look impressive but has minor issues such as Mac support.
We do target consumer PCs but as it's typically a niche market selling to corporations, we do have scope for being able to say "you need a $150 nVidia GPU" - just not forcing people to get special gaming rigs or whatever. For that reason, systems which have hardware acceleration seem preferable purely since the amount of calculation can be pretty steep.
I suppose we're focusing on the list posted in this board's sticky thread, rather than anything too obscure. It either needs commercial backing or a healthy community, some small OSS project that might suddenly die isn't too much help.
As always, links to other sites discussing this stuff are welcome... but I've yet to find anywhere better than GD for these questions.
