Wednesday, October 23, 2013

Workflow, Workflow, Workflow. It's all about the workflow.

I've been spending a lot of time at work with a certain software vendor trying to help me do some rigging using their tools.  They have been a big help in getting me the result that I need, but not necessarily the way that I want.  During this time, we had a lot of discussion about rigging workflow and one thing that he said to me was very telling.  He said, "It seems like rigging is an antiquated method that started bad and has never been truly solved."  We talked about almost every package, and the pros and cons of each, and how there aren't any solutions out there (other than proprietary methods) that really make rigging a natural process.  We tried using almost every method of skinning.  None were giving us a good result to start from.  Some were better than others.  But each was going to need hours of corrective shaping just to get reasonably good deformation that didn't look like rubber hoses and candy wrappers.

When I was a Disney Interactive, we had written an in-house tool that used a heat-diffusion algorithm and did an amazing job getting us 80-90% there.  It didn't bleed from one finger to the other, or from the inside of one thigh to the other.  It understood topology, and although it wasn't perfect, it removed the need to do most of our weight painting, and it was a push button solution.

At Pixar, we didnt' have any automatic weighting tools, so we had to populate weights by hand (before I left, we were developing some projection methods that would transfer weighting from a lo-res to a high-res mesh and did an amazing job).  But weights were split into different axes, so you could weight XYZ, or each axis individually, and they blended very nicely.  Then you could do corrective sculpting that worked amazingly well, although a bit tedious (I'm sure a year now after I have left, they have those tools dialed in really tight!).  But the workflow was pretty straight forward, and quite artist-friendly.  The workflow wasn't perfect, but it was getting there.

Way to look like you're working, Lou ;) (Lou is a rigging, modeling, animating, and modeling GENIUS.)

Now to today.  Man, I wish I was back in Presto!  But alas, I am not, and have been resigned to off-the-shelf solutions. :(

What's so interesting to me in software is that everyone talks about feature sets.  They all talk about what they can do.  They talk about the depth of their tools.  They talk about how powerful their toolset is.  Yet they never mention one thing...HOW to do it.  I think this is for a very good reason...they know it's not a selling point.  Go into Maya and try to create a corrective blend shape for a joint.  Go ahead.  Try it.  Vanilla Maya.  Yes, you CAN do it.  But how much pain and anguish was it to create it?  Okay, get a plugin, and try it now:

It may have a nice new window.  But it still lacks simplicity, and is still hampered by the inherent limitations of the software.  I still have to duplicate a mesh, pose it, create an offset mesh, sculpt that mesh, and then hook it back up as a blend shape that is set with a driven key.  This is how most 3D software works.  It really is quite pitiful.  No fault of the developers of the plugin.  They have to work within the bounds that they were given, and most software boxes are pitifully limited, or give such tight constraints that coming up with a more elegant solution usually means scripting your way through the tough parts (not bypassing them altogether).

Disney seems to finally have come up with a good, artist-friendly solution that is fast and easy to use (I'll have to verify that with friends that work there).  But it is a volumetric solution that is fast and apparently automatic (I'm sure it's not quite THAT easy. But when they say that their goal is to make it super fast AND physically accurate, I'm sure ease of use is definitely a goal of theirs).

This is a perfect example of writing function into tools that otherwise would not allow for this approach. This is a proprietary addition to Maya, but it sure would be nice to have a system like this native.  Things like this need to be standardized in rigging workflows.

I'm a big Modo user.  I'm not a fan boy - I actually feel like it gives me better results faster, and is growing its tech base very quickly.  Their rigging tools are coming along nicely, and I hope that soon, there will be a rigging "round 2" for a future release of Modo.  I can see huge potential in their rigging system.  My good friend, Rich Hurrey (a Pixar rigging TD), has done an extensive rigging training series and is working on the next iteration.  Rich is a genius when it comes to rigging, and knowing that he's using Modo gives me confidence that I'm not the only one who thinks it has amazing potential.  Can't wait to see it!

In the end, I think I'll just continue looking and hoping that some software vendor "gets it", and realizes that feature sets are one thing, but the ability for an artist to create what they want with ease will sell more copies of their software than any other factor there is.  Make a powerful, feature-rich system with a slick and easy to use workflow, and you will own the industry.