Jgeom

I don't know how beneficial this is to the jME community, but I was reading about the new jgeom project at Java.net and thought perhaps at least the idea might be beneficial.  The web site is https://jgeom.dev.java.net/ but what I took away from it was the idea that maybe a separation layer between the geometry aspects of jME and the application of that geometry might be beneficial so that perhaps even Xith and other projects outside the camp can get some benefit and maybe contribute.  It seems counter-productive to have two projects that share so much in common having concurrent development occurring.  I definitely believe that jME is a much better structured API, but there are some smart guys that are not using jME and if we can provide some way for them to indirectly contribute then I believe it would be beneficial to us all.



As of this point in time I think the biggest struggle we're all facing is getting Java accepted as an acceptable means of game development and not a joke as it has classically been.  Anything we can do to help this endeavor I think we should consider.



darkfrog

I've looked through their code, and honestly, there wouldn't be a lot for us to gain from it. We already do everything that it does, with exception to NURBs.

What about the idea of perhaps splitting those standard features out into a jme-non-specific project?



darkfrog

Well, in all honesty… I don't much care to support other scenegraphs with our work. That being said, basically, geometry is nothing more than arrays of data (vertices, texture coordinates, normals, etc). There's not a whole lot to split out that would be worth a whole project. Loading that data is seperate enough that someone could say, make their own renderer that displays a jME TriMesh.



However, I may be misunderstanding what you are saying. What exactly about geometry do you think would benefit multiple projects if it was seperated out?

You're probably right.  I was just thinking about the idea of complex shapes, perhaps even model loading could fall under this umbrella too.  I keep on thinking more about model loading for gaming and I see just how much there really is that needs to be maintained (such as converters from tons of different formats and features) and it seems like too big an undertaking for one group that already has a focus towards the graphics.  It just seems to me if we could create a non-jme-specific standard that could be loaded into jME we could get some additional help from people not necessarily interested in jME that would still be interested in contributing to it.  Further, it provides a good abstraction layer in my opinion between the mathmatical vectors and the implementation (drawing of them to the screen - or any other representation).



Just a thought…



darkfrog

Model loading is the reason that we have the .jme format. The idea is jME only worries about loading/rendering/animating .jme. It doesn't care about anything else (MD2/MD3/MD5, Milkshape, 3DS, etc, etc, etc). Then as they are needed, exporters can be written to convert a format to .jme. This has proven to work pretty well (some documenting and cleaning up is needed still). But people have been able to write a converter fairly quickly (a lot faster than writing a loader).



If there was some third party library that handled this, they would basically need to do the same thing. There will always be new model formats (as people create a format that meets their needs at the time). Our .jme format and associated tools can be used outside, Xith et al would just have to write a single .jme loader :).