UK based PhD studentship Architecture into Virtual Environments

Hello all, firstly apologies for posting an advertisement for a PhD post here, but we are looking for as wide a range of candidates as possible.

I work for a small architecture practice (Slider Studio) which also has a small software development arm. We have been using java3d for a project of ours called YouCanPlan (, which I led development on. We are constantly looking into new ways to improve the links between architectural CAD models and virtual environments. With respect to this we are offering a PhD studentship to research this subject, which will involve working with the company, researching current gaming technology and improving performance of high polygon 3d CAD models. For more information please check for more information and how to apply.

Thanks for reading and apologies for going a bit off topic, I'll definitely be considering jME for some of our future projects. Having used java3d and experienced first hand some of its limitations, jME certainly seems a winner, keep up the good work!

Kind Regards,

Chris McDonald

Senior Software Engineer, Slider Studio Ltd.

Neat project! I wish you guys the best.

What are the limitations Java3d that make you consider jMe?

I have worked with Java3D. I think the worst limitation of Java3d is the huge memory footprint it takes. That is one reason why I have been learning jME.

Thanks for replying, sorry for the delay in getting back. Yes, by far the biggest limitation of java3d is the memory usage. We have attempted many optimisations to reduce memory usage. The biggest hit for us is the texture management. We have developed a commercial result from the project, where the platform is being used for public consultation on new housing developments (, for this project we took photos of many of the houses in the area. The model we produced is also very large, as a result we've had to set a heap size of nearly 1 GB!

Apart from the memory issues there appears to be better ways of handling level of detail management. I particularly like the sound of dynamic LOD. The scenegraph in jME appears to be better structured, with nodes inheriting what we would call nodecomponents from their parents.

Also generally with jME being a game engine it's designed for exactly the sort of problems we've been encountering with j3d. I'm sure we will encounter the same or similar issues using jME but they should be easier to solve with the correct tools. One other thing I prefer about jME is the game loop idea, I know that in j3d you can also have a similar setup by using immediate-mode rendering rather than retained-mode, but I haven't tried this out yet. By default j3d uses retained mode rendering so whenever I change anything in the scenegraph j3d will try and render it quickly. Often this results in inconsistant frame rates as you move your view from an are of low geometry into an area with complex geometry. It also results in billboards and other behaviors that have short delays before being activated, and generally less control over your scene.

Also with jME reputed to be faster helps too!

The best for then was they are offering a PhD studentship to research this subject.Thanks for it.