Take a look at the features of the Unreal Engine3
It would be great to be paid even a part of what it cost to get the UT3 engine developed… 
I second that 
You know sheet, check the price for a commercial license of the Unreal 3 Engine, after that check the price and the features of jMonkeyEngine again. I am quite sure you will find jME to be a way better bargain.
I think one of the coolest things about UE3 is the editor – almost the entire game can be created through it it seems.
Go MonkeyWorld3D team!
Well, i agree with all of you… I checked already the price before posting the first message
One important feature of AAA engines is the art pipeline. How easy is it to get art into the engine and how easy is it to manipulate it (editors etc). At least that's my understanding of the art pipeline as a non-artist!
well UE3 looks good but, I have been sinking my teeth into FarCry recently and I have to say "damn" CryteK engine kicks ass, now living in a tropical environment all my life it's quite natural that I would have a bias for such, in fact been using mw3d to put a jungle together long before I ever considered buying farcry
I've said it before and I'll say it again. Get the inside scoop from Tim Sweeney himself. http://www.pqhp.com/cmp/gdctv/
Game Technology and Content Creation for the Next Generation
Tim Sweeney
Length - 50:28
This talk will discuss the tools and development practices that Epic Games, a relatively small East-coast development studio, uses to develop good games effectively in this environment. The content-creation pipeline, showing off bits of two upcoming next-generation games powered by Unreal Engine 3, will be generated, and the major techical, artistic, and game development issues related to this will be discussed.
It is heavily influencing the small amount of work that I'm doing in MW3D.
How do they implement such accurate collision system, can we do collision with the level (not bounding) in jme.
From what little I understand about it; the engine editor lets the designer customize the bounding by using collision primitives for models. So, a single model doesn't just have one, or a group of, engine optimized box and/or spere bounds but also allows for the manual tweaking of those bound primitives per model.
They also use bounding primitives inside meshes to occlude the objects on the opposite side; in jME terms a quad mesh placed inside the underside of a mesh terrain hill. During its processing of the frustrum the renderer sees the blocking primitive and ignores everything lower than it in the queue. It would be like an anti-node in jME that doesn't let the lower nodes get rendered
physics-wise you can achieve alot of what can be achieved in unreal with jme-physics…i've done full triangle-triangle collision with 30 player-models and a 100k poly world on the xbox with ODE.
what would be nice though, is a wrapper to opcode(can also be used in ODE instead of their own collision detection), for fast collision detection and fetching polys affected by a collision…this would be much needed in alot of situations not only connected to physics, like collecting polys close to a character for projected lights, shadows and so on, without having to go through every poly in all meshes…
MrCoder,
Sounds like you've got a lot of experience with using ODE. Can you explain how you dealt with rolling friction in your games?
darkfrog
isn't the physics forum moving away from ode? I disagree with it but never said anything ode is really powerful.
darkfrog:
just like they usually says in the ode forums…just damping the angular velocity and/or the velocity when in collision… usually though, the only time i use spheres are for player-objects, and they need to be easily controlled programatically so i end up using only the most basic collision response by ode in that case…
wakebordin:
what do you mean by the physics forum moving away from ode?
To be completely honest I don't really like ODE. It seems to be lacking a lot. However, ODE is unfortunately the only physics option for Java at the moment (at least that I'm aware of). I might be helping someone build a wrapper for Novodex but that will probably be a ways in the future before that's available for any real use.
darkfrog
what do you think is lacking compared to other engines?
It seems to be pretty error-prone, lack of rolling friction support, aerodynamic drag, and some more advanced capabilities that newer engines support now. Also, it doesn't seem like ODE is being maintained much anymore.
darkfrog
yean you are right, i agree… would be great to see a wrapper for some other engine like novodex(physx)… but actually, i'm really found of the newton game sdk, i love the fact that its so stable… the worst nightmare when developing a game with full physics is exploding objects, keeping threasholds and integration times in their boundarys and so on…