LightWave 9 review

'LightWave reborn' is how NewTek describes the ninth incarnation of its high-end 3D application. And while the program will look familiar to seasoned users, there are more than enough changes across the board to justify this claim.

First, what hasn't changed: LightWave is still two apps - Layout and Modeler - connected by the Hub, a small program that mediates the transfer of meshes between the two halves. Rumours had it that version 9 would finally see the long-awaited integration, although some modelling tools have made the migration across into Layout.

In Modeler, the first big change is the adoption of the Catmull-Clark algorithm for subdivision surfaces (SubDs). This means that the polygon cages from which SubDs are derived can now contain polygons with more than four sides, placing far fewer constraints on the artist when they're creating meshes. There's a slight calculation overhead for this, but it's unlikely it will trouble anyone. You can now also have Adaptive Subdivision, giving you high polygon counts in areas of high detail or movement and lower counts for more static areas of a model. This is controlled in many ways - for example, by the application of a weight map to the mesh, by expressions or by a gradient. Using a distance-to-camera gradient, the subdivision level can actually vary - coarse when the object is in the distance and fine when closer to the camera.

Also making their first appearance in Modeler are Edges. Well, they've always been there, but you can now select them directly for performing modelling operations, in addition to the usual Points, Polygons and Volumes. Edges can also be weighted by the application of weight maps. The OpenGL performance in Modeler is still pretty clunky, but we hope this is addressed soon.

LightWave's Core Renderer has been rewritten and while it still retains the trademark gorgeous rendering quality, it has received a speed boost of around 30-40% if you're using a PowerPC Mac. Users of Intel-based Macs will have to kick their heels a while longer waiting for the universal binary version. Discussions on the NewTek forum indicate this to be imminent, but until then, you can expect to halve your rendering speed and experience a fair few random crashes running under Rosetta. Indeed, as we went to press, an Open Beta of 9.2 was posted, which was still not Universal.

The rendering speed (or lack of it) can be gauged with the new, more informative Render window, which now includes a progress bar to keep you informed of the expected rendering time. Something that hasn't been addressed is that the renderer still takes its time rendering polygons that are hidden in the final scene - and there's still no marquee/area render tool.

Rendering takes place through cameras, and these have been supplemented by the new Advanced camera object. This gives you an unprecedented level of control over the image seen through its lens. You can specify standard Perspective views, but you can also choose from Orthographic, Fisheye and even a 360 panorama from a single camera. Other variations include the ability to deform the camera plane, to generate UV map projections, create non-existent or imaginary lens types and even morph one lens type into another.

As well as rewriting the renderer, LightWave 9 incorporates a completely new shading model, and a new way to design the surface characteristics of an object. Yes, at last, LightWave gets Nodes. Using the new Node-based editor, you can build surface characteristics by stringing together chains of Nodes - in other words, by building a Shader Tree. The Nodes can have different properties - they can be gradients, procedural textures or whatever - but every Node has an input and an output. The output from one Node can be linked to the input of another Node, and so on. Nodes can control every aspect of a material (colour, bump, transparency, luminosity) and the variety of ways that they can be chained together is, essentially, infinite. There's really only one word to describe LightWave's new Node-based shading system: deep.