Long live Janus. ;-)

 

I’ve been doing some custom development for an architectural visualisation company in Australia called 3DVIZ recently. During Janus’s commercial phase,  3DVIZ bought a licence. It is usually the case that I don’t know what Janus user actually do with Janus technically-speaking. Few users ask for feature requests, and even fewer ones explain their workflow and how to better improve it through Janus. That’s mainly why Janus developed the way it was: my own personal production needs, and my curiosity in proceduralism.

It is the proceduralism that seemed to draw 3DVIZ into Janus. When I wrote the FOR loop constructs in Janus, it was mostly out of admiration of Houdini and the curiosity of what proceduralism could look like in Janus. No had asked for it, I only had an inkling that I may, just possibly, need it if ever the chance I get into a large-scale project in LightWave. But, if I’m honest, I’ve never actually used FOR loops for any of my commercial projects; none of them were ever big enough to warrant to advantages of using the feature.

When 3DVIZ contacted me for support, I realised that they were using it in a far more advanced way than I personally used it as a TD in commercial work. It was gratifying to see that someone actually had some need to proceduralise their scene management and rendering to that level, and that Janus’s FOR loops actually made all the difference.

3DVIZ approached me again with a permutation problem. From the little I know about it so far, their asset hierarchy is well-organised (ie rigid). And this is the case because they need to be able to render more variants upon variants; from a house, they render parts of the house in varying configurations, each with their own types of materials, and so on so forth.

Part of 3DVIZ’s own request, since they know their problem more than I do, is to enable them to automate scene loading and Janus operations from a more ‘global’ script; as FOR loops handle the breakouts for any given scene, now they want to expand that capability across scenes. The concept is similar to LightWave’s own Render-Q script, where a script resides ‘persistently’ and orders LightWave to do tasks.

The most obvious idea to automate Janus is to allow it to accept macro commands. A ‘controller’ script handles the scene loading, then writes a macro to a file, which contains commands to breakout so-and-so render pass; then signals Janus to receive the macro; when Janus completes the macro, the ‘controller’ loads up another scene in its queue, then writes another macro, and repeats the procedure.

Thanks to the years put into Janus, the implementation of macros was clean, and with some limited testing, the concept works as well as imagined.

However, my main obstacle would be their expansive asset hierarchy. The real challenge is to make sense of it in my head, and design a ‘controller’ script that creates sensible macros that solve 3DVIZ’s particular problem of permutations.