Rayfire Tool @ Cryengine²
polycounter lvl 17
Some of you my know that I'm doing a lot of destructable assets at the moment (beside my trees of course hehe).
After some testing and messing around I tried the "Rayfire Tool" made by Mir Vadim. The tool has a lot of awesome features and mainly made for CG scenes/movies/effects.
CE² is like CG but with less polies in my opinion. So I turned down some values for complexity, set some values here and there (mass, limits etc) and connected all the stuff with joints.
My basemodel is a simple wallset I've made. You can use it in a repetitive manner. The UV tiles without issues and no seams are visible (except my sloppy Uv-layout) after the breaking made by the rayfire tool.
Heres a shot made out of my box:
It could be improved with proper Physics_proxy (its set to box at the moment) and sub-breaking pieces for some f the bigger chunks to make it bit more realistic.
The nice thing is if the wall is hit it doesn not break to full pieces at once but shows some cracks. I think its because I set the proxy to "box" and the proxies are a litte bit stuck into each other and finally fall apart after more force is applied. Nice effect.
This is the wall before and after shooting at it.
My next step will be to set up an apartment block with this tool. I will stick to a modular buildup again so I have to do all this pre-breaking only once.
Some more testing is also required to clean up the meshes. Polycrunsher does a good job here and protects my borders and materials.
To sum it up:
Its not a real physics thingie. Its pre-computed and pre-broken. Gameart uses pre-computed stuff for ages now so its ok for me.
Like all other physics going on in the CE2 its CPU bound but can be reduced with good physic_proxies. The higher polycount can be compensated by occluder and lod meshes.
I havn't tested whats up with drawcalls and batches but the materialcount is simple on this wall and the modula pieces can maybe instanced by the engine. This has to be tested too.
If some of you guys do own testings with this method keep me updated.
Finally a movie:
Bad quali: [ame]
http://www.youtube.com/watch?v=TdhqPsmwe00[/ame]
Better quality: http://michi.be.hosted.dasprids.de/michibe/WIP/rayfire_realtime.avi
After some testing and messing around I tried the "Rayfire Tool" made by Mir Vadim. The tool has a lot of awesome features and mainly made for CG scenes/movies/effects.
CE² is like CG but with less polies in my opinion. So I turned down some values for complexity, set some values here and there (mass, limits etc) and connected all the stuff with joints.
My basemodel is a simple wallset I've made. You can use it in a repetitive manner. The UV tiles without issues and no seams are visible (except my sloppy Uv-layout) after the breaking made by the rayfire tool.
Heres a shot made out of my box:
It could be improved with proper Physics_proxy (its set to box at the moment) and sub-breaking pieces for some f the bigger chunks to make it bit more realistic.
The nice thing is if the wall is hit it doesn not break to full pieces at once but shows some cracks. I think its because I set the proxy to "box" and the proxies are a litte bit stuck into each other and finally fall apart after more force is applied. Nice effect.
This is the wall before and after shooting at it.
My next step will be to set up an apartment block with this tool. I will stick to a modular buildup again so I have to do all this pre-breaking only once.
Some more testing is also required to clean up the meshes. Polycrunsher does a good job here and protects my borders and materials.
To sum it up:
Its not a real physics thingie. Its pre-computed and pre-broken. Gameart uses pre-computed stuff for ages now so its ok for me.
Like all other physics going on in the CE2 its CPU bound but can be reduced with good physic_proxies. The higher polycount can be compensated by occluder and lod meshes.
I havn't tested whats up with drawcalls and batches but the materialcount is simple on this wall and the modula pieces can maybe instanced by the engine. This has to be tested too.
If some of you guys do own testings with this method keep me updated.
Finally a movie:
Bad quali: [ame]
http://www.youtube.com/watch?v=TdhqPsmwe00[/ame]Better quality: http://michi.be.hosted.dasprids.de/michibe/WIP/rayfire_realtime.avi
Replies