Absolutely, FarseerBMX, Chipmunk and Box2D are different and not ‘competing’ against ^^

Choosing between theses 3 physics engines will be a nice contribution to blitzmax !

I’m wondering if concave bloc could be use as deformable terrain ? Does the polygons definition of objects in Farseer can be dynamically modified ?

Also it’s probably the most CPU consuming type of collision to process. Is the number of maximum ‘vertex’ point is important in term of performance ?

Digging holes in the ground will create a huge amount of these vertex, maybe it’s not the best approach ?

I will ask the same question to Jeff since he probably know the answer

Also keep in mind the ~1-5ms differences are coming when LOTS of objects and activity are going on at the same time (stress tests). During normal use of Farseer BMX there is virtually no performance difference.

I’ve been working on this port for 80+ hours. Still some hours to go as I implement that last few springs/joints/controllers along with demos to test their functionality. Once everything is ported and working I’ll make an initial release for others to play around with, maybe even get it working on mac/linux.

I’ll keep that in mind for the future demos I develop to make a faster time step. So far I haven’t had much issue using a faster time step with the engine.

FarseerBMX in no way is ‘competing’ against Box2D. I hope to release this to the community to give users of Blitzmax more choices when it comes to physics engines. Right now there is Chipmunk, soon to be Box2D, and soon Farseer. Each has its own advantages/disadvantages. Farseer has concave collision support, and Box2D has Continuous Collision Detection. Which ever is more important to the developer they can decide. Who knows, maybe in the future both features will be present in either engine?

I didn’t notice at first that Farseer don’t have any CCD. Hope this won’t be an issue with bullet type element.

Effectively the impression of “slowliness” surely came by the way of blitzMax render things.
Well I suppose a difference of 1~5ms is better than nothing Hope it’s not too CPU consuming.

You working since a long time on this complete revamp ?
I suppose we only be able to play with Farseer when you decide it’s features complete and optimized for the best performance.

Would it be possible to write a tiny example with a faster time step like for a normal game. I like physics engines but always find examples playing too slow and not realistically (maybe to prevent inaccuracy ?)

Maybe both physics engines (box2D and Farseer) will come at the same time so we could compare them and see which one is the more suitable.

Thanks

Box2D looks like a great 2d physics engine. I believe Brucey is working on the 2.0 port (which includes continuous collision detection!).

The port of Farseer to Blitzmax is pretty much a re-implementation into 100% Blitzmax. There are no outside dependencies besides a basic blitzmax install.

What made you have the impression that the examples were running slower on the Blitzmax version? Was there a graphical stuttering or something? My goal is to make it as close as possible performance-wise to the C# version. Visually right now, for higher-end machines there’s no difference. It’s the lower end machines that are taking a bigger hit obviously. There are slight differences in performance right now that I’m trying to optimize as much as possible. There is about 1-5 ms difference between C#/Blitzmax on the more stress-test centric demos. The thing with XNA, is due to it’s higher requirements VS Blitzmax, it tends to work very well on most PC’s that can run XNA at all. I’m going to be doing testing on machines that are a bit older (no pixel-shader support, slower clock speed, etc) so performance is a bit more critical.

Another bottleneck with this that isn’t as much physics related is rendering. So while FPS give a good indication of performance it doesn’t give you the ultimate cause of why things are performing badly. XNA I believe has an advantage here as rendering wise it simply renders the demos faster than Blitzmax.

My original motivation behind making this port is to create a game of my own in Blitzmax. I wouldn’t be doing it if I didn’t think it’d enable me to create dynamic and fun games. So in short, Yes when finished I don’t see why it wouldn’t be capable of creating games similar to the above.

As for the license, I believe this would qualify as a ‘derivative’ work of Farseer and would fall under the Microsoft-Permissive License or something compatible with it. Which to my understanding, is pretty loose and has minimal strings attached. more info here: http://www.codeplex.com/FarseerPhysics/license

I’m very interested in Physics engine and already have my preference with Box2d which is currently wrapped by Brucey in the blitzmax community.

I already play with Farseer in C# and find it very powerful (mainly with concave shape collision).

It’s a very good idea porting it to Blitzmax ! I didn’t even know it was possible to do such with .net project. Did you rewrite the whole source (?) or wrap a precompiled library ?

I was having good FPS running your examples (like 105fps on a 64×2 3.8ghz AMD 2go Geforce 7600se) but my impression was that it was a bit slower than the Farseer examples…

Is this bmx module is fast enough to handle “now days” games physics based (Like locoroco, gish, tower of goo and similar with a indie 2D look) ?

Finally what’s your main idea with this module ?
Open source / Shareware / Freeware / License ?

Thanks