Maya=FAIL

I've been working in the 3D visual effects industry for over 10 years now. In the beginning, the industry had forbiddingly high startup costs. SGI workstations were the norm (Octanes, a few Indigo2s, and an O2). Frames were transferred from D-Beta with a networked hardware frame buffer called an Acom. Most studios used a mix of proprietary and commercial software, usually Alias Poweranimator or Softimage.

Not us, though. We used Side Effects software's Houdini, the successor to the lovingly obfuscated Action animation tools and ICE compositor.

I spent those halcyon days writing plugins for fur, dynamics, feathers, and lots of other studio tools.

For all the flack Houdini has taken -- complicated, artist-unfriendly, horrible UI -- it was remarkably powerful. We enjoyed a very close relationship with the small development team at Side Effects, and I became used to the idea that my input mattered, and that software should be cleanly written above all, and that forward progress -- not backwards-compatibility -- should be the ultimate goal.

To be fair, most of the shows we worked on were small in scope, and generally a single artist/TD (there was little distinction at this point) crafted a shot from plate to final comp. You could do pretty much anything in Houdini, yet still be able to backtrack and find out what was going on by looking through the object and SOP networks.

Houdini slowly progressed, improving on both the inside and the outside. APIs were refactored, or even thrown away completely (e.g. COPs vs. COPS2). UI became progressively cleaner.

Suddenly, my bliss was interruped by a momentous event. SoftImage had foundered while Microsoft tried to make it work under Windows, and Alias|Wavefront's Maya had slipped into the gap. And they had DROPPED THE PRICE. Maya Complete, formerly priced at $7,500 would now sell for $1,999 and Maya Unlimited, which formerly sold for $16,000 would be priced at $6,999. This was significantly less than Houdini's $18,000 price tag.

The economic climate at the time was difficult. Strikes and 9/11 had reduced the amount of available work, and the high margins formerly associated with VFX work were thinning. Maya's new price made it a very attractive option for the low-budget cartoon we had managed to snag. So, the studio made the leap. We put down our trusty Houdini 5, picked up Maya 4.5, and went to work.

In many ways, Maya turned out to be a breath of fresh air. Artists loved its clever UI, and since most visible parts were written in MEL (its home-grown scripting language) it was customizable beyond belief. We wrote a rudimentary asset-management system, hooked it into our proprietary distributed rendering system, added custom animation tools to seamlessly import lipsync data -- it felt like the sky was the limit.

But the bugs. Oh my, the bugs. Daily crashes, corrupt scenes, inexplicable errors and failures...

6 years has now passed. I'm still working in this industry, and Maya is still the industry standard. The bugs haven't gone away. They've multiplied and festered. New features seem to solve problems, until up close one realises they do little more than fulfill a marketing check-list. Maya is still wildly popular, partially due to its use by large studios (who have rewritten large swathes of it to work around its inherent problems) and partially due to Autodesk's relentless marketing and sales force.

After 4 years of using it, I can say without a doubt that I hate Maya. I hate it with a passion. I hate its lack of encapsulation, I hate its mickey-mouse toy scripting language, and I hate Autodesk's creeping influence. Most of all, I hate how bloated and buggy and slow it has become.

This blog is a way for me to let off steam and a diary of the bugs and general stupidities I uncover in my day-to-day work. It's about how Maya=FAIL.