When did you first see LightWave?

When it was version 4 on the Amiga. I had a friend who had coded a little 3D engine and was using it to build objects for real-time use. He showed me the possibilities of LightWave. I was quickly seduced! In fact, I found it thanks to the demo party scene on the Amiga at the time. These were huge meetings (very popular in Scandinavian countries) where you could find thousands of people with the same passion: making demos in real-time 3D. LightWave was a widely-used tool and Amiga users swore by it: I didn't really have a lot of choice!

I really started using it seriously when version 5 was available on the PC in 1998. It was really only for modelling and texturing real-time objects. I learnt to optimise the number of polygons and to make them as "tight" as possible.

What do you particularly like about LightWave?

Above all, LightWave is a well thought-out application. I greatly admire its authors for having such an ordered "spirit". It's very clear and intuitive if you stick to its philosophy and there aren't any dead-end menus where you can easily lose yourself.

For you, which is the best tool in LightWave?

Modeler! Subpatch organic modelling is an absolute blessing and it allows you to make different versions for polygonal LOD. The addition of weightmaps in Subpatch mode makes this tool perfect.

There's also the render quality with HDRI (High Dynamic Range Imagery) and radiosity. LightWave has always had a superb renderer, even in plain Raytracing.

What could be done to improve it?

I'd really like to see the notion of scenes within scenes incorporated into LightWave. It would be really interesting to link a scene to a Null Object, for example, or to some particles and to consider it a completely separate group with its own parameters to edit and so on.

In the next version, I'd also like to see a tool that would make setting up a character easier, like Lucasz Pazera's Auto Character Setup.

VIPER should support radiosity and collisions between particles and objects should be much more precise.

It would equally be nice to circumvent the standard rendering pipeline so you could splice in a different one for a video game or another render engine.

OpenGL rendering in real-time should be improved. You should be able to see at least a double texture mapping with real transparency — that's more important than lens flares! :)