Xith vs jME

[phazer]

Exactly. It seems unnecessary to create three transform matrixes just because I create three TriMesh'es. I don't understand this design choice, besides the fact that it saves lazy persons the extra code of adding a transform group to the graph.

[darkprophet]

right, so you have created your triMesh, where is it gonna go? Ummm...near the heavens?

It has to have a place in the coordinate system, if no location/rotation/scale is specified, its at (0,0,0) with the scale being (1,1,1).

Tell me, in Xith3D, if you created a triMesh, how do you place in the scene? and if no transform group is there, on what premises is it placed there?

[cep21]

The reason this happens is because each level of the scene graph has a bounding volume.  It may seem wastefull to have to transform your node all the way down the scene graph but the real speed up is during rendering.  If the middle of the scene graph can be culled with view fustrum culling, then none of the lower leaf nodes have to be checked.  To do this, you need transforms at each level.  It's also very easy and fast O(1) to get the exact world location of any node on the scene graph.

Games published by our own members! Check 'em out!

[kevglass]

You place potentially multiple TriMeshs within the same transform group.

Xith3D also supports bounding volume determination. I'm not sure why you believe you need a transform at every level to achieve this. Java3D did this also incidently.

Kev

[phazer]

It's impractical and inefficient when you don't need the transform, for example if you have one transform group and multiple TriMesh'es as children to this node. Sometimes you have mutiple TriMesh'es which can't be replaced by one:

- You want to be able to add/remove/hide/show individual TriMesh'es.

- You have different materials on the TriMesh'es. One TriMesh can only have one texture.

[phazer]

I agree with kevglass. There is no need for a transformation matrix in each node to perform transformations in advance. Just use the matrix in the parent node if there is none in the node.

[cep21]

I was meaning to say if you have a SceneGraph of depth 10, consider each leaf node as a renderable object.  If I can figure out that the node at depth 5 is not viewable I don't need to check any nodes or leaves below it.  I need the exact position of that node to figure out if it is viewable or not.

[cep21]

I see what you're saying.  I just think everything comes out even (speed wise) with how often jME uses the world location of an object and the nodes in the scene graph.

[phazer]

I know what you mean, but that isn't an argument for having a transform matrix in each graph node. I think Xith does the bounding bounds/view frustum optimization too, but it doesn't keep a matrix in each graph node, only transform groups.

I think jME is a great project (and so is Xith), I'm just a bit surprised by this design decision.

[bozoface]

this is sad, horrible, and embarassing, but true:

I glanced at jME but did not even give it serious consideration
once I found Xith.   Why?  

Just because of the name, "monkey engine".  I'm in an organization
where  convincing people to use java is in itself an issue,
and I feared that getting them to consider using something
called "monkey engine" would be impossible.

As java's reputation in graphics improves in the next couple years
this won't be an issue, but for now...

for your consideration.

Games published by our own members! Check 'em out!

[SpuTTer]

Make sure not to tell the guys at work that you go by "bozoface" Unless of course you work at a circus, in which they'd be familiar with monkeys..

[Matzon]

this is sad, horrible, and embarassing, but true

Not to mention very unprofessional and stupid! Judging a technology by its name alone is really really stupid!
I (and my company) would have no problem whatsoever using RectalProbeMaster2005Ultra as a profiler, if it worked better than its competition...

[cep21]

Just because of the name, "monkey engine".

*nod*.  That's why I use SuSE at home.  I refuse to use FreeBSD because of the satanic mascot.  And those penguins?  Who could take that seriously.  Lizards are way more profesional.

[Bombadil]

Names actually have a big influence on the person/project/etc . That's why names usually have some special meaning. :-)
However, this is nothing new at all. The ancient Romans put it into the famous and so true sentence: Nomen est omen.

Back to topic: Is the jME maintainer still with us Java fans? I'm asking (seriously) because last time I visited his site (some 1+ months ago) he's been busily working with Dotnet and applied to the Xbox headquarter (I think he got the job?)

[blahblahblahh]

this is sad, horrible, and embarassing, but true:

I glanced at jME but did not even give it serious consideration
once I found Xith.   Why?  

Just because of the name, "monkey engine".  I'm in an organization
where  convincing people to use java is in itself an issue,
and I feared that getting them to consider using something
called "monkey engine" would be impossible.

LOL. Start a new poll, asking for new "marketing names" for jME (i.e. three-word names using JME that they could use when trying to appeal to conservative companies).

[Per_T]

Is the jME maintainer still with us Java fans? I'm asking (seriously) because last time I visited his site (some 1+ months ago) he's been busily working with Dotnet and applied to the Xbox headquarter (I think he got the job?)

On avarage mojomonk posts ~4 times per day at the jME forums.

As for the name, it was one of the things that made me choose jME over Xith .

/Per

[Middy]

Well I have my own homemade engine and took a look a jME. it had everything I had and solved every problem I had.

That was a bit frustrating thinking of the time I spend on my own engine.

Xith, would only solve my scenegraph problems. But seemed a bit more advanced.

Perhaps I'll trash my own... I'll see after I publish my first small game

[SpuTTer]

Tell them it's called Java Marketing Engine...

[Mojomonkey]

But Monkeys are funny, who would hate a money? Seriously, if the name affected your choice, *shrug* oh well, that's too bad, but I won't lose sleep over it.

I'm still heavily involved with jME, never really left, just took breaks here and there mostly do to a cross-country move. However, I still also play with .NET + Managed DirectX as well, but that doesn't mean I've abandoned Java or anything. Just ask, I've recently committed a significant amount of code that caused a significant amount of problems.

In fact, I don't post much on these boards anymore because I do spend so much time on the jME board, and work/life limits my board posting time.