Graham Clark Interview On Xsibase

[peliosis]

I've just read this : Graham Clark interview

It's xsi pipeline ang Graham says about creating custom tools for everything :

custom hair system with dynamics and shaders, custom particle system(!), custom pipeline software.

And ofcourse more proceduralism and nonlinerity is on the hot list:)

I wonder if houdini shops have to produce so many tools either?

I thought they base more on otl's than hard coding? Apart from SESI support that create solutions to later like in CORE.

Worth reading after all.

[old school]

When you need to support numerous artists of varying abilities in different departments moving gigs of different types of data through different departments and keep things straight and true, yes.

Looking from the outside in (like I can) every shop in feature film and VFX are constantly building, refining and redesigning their pipelines to meet different and evolving requirements. Fur/Hair is one example for studios as they are always improving from previous jobs.

As for the article, I just love the part where Graham is asked "Which areas should be improved?" (in XSI):

- particles too basic/black box and difficult to navigate

- op stack is far to linear and can't change inputs / can't accomodate changes

- mid level integrated compositing to support lighting at comp time

How about looking at Houdini and Nuke?

It is also troubling that there are so few really knowledgible CG artists available in general in the industry. I really don't think it is limited to any software. Graham's comment >>> "...still, a shortage of schools with *production* experienced teachers. "

If the education landscape has changed in the rest of North America like it has in Toronto over the past few years, then the shortage of good technical CG artists will continue (houdini schools excluded of course).

[Jason]

I agree with old school here... except that I'll mention that many custom tools are mostly built on the fly - and then kept around and refined for later projects. In other words, its often not a big sit-down meeting with in-house coders where a spec is typed up and money is spent; more often it's smaller case-by-case tools where a generalized component of which can be rolled into HDAs and have application further down the line in other areas. Sure, there are a couple of bigger projects that each facility spend some time on (like DD's "Storm" renderer and COREs hair tools which have been playing a role in their companies for years). These simpler tools are built by artists or TDs on fly and herein lies the power of HDAs: they're accessible to every single artist.

These tools serve mostly to enhance efficiency and store techniques and this is how they pay for themselves. These days its not so much a case of "this tool will allow us to do something nobody else can do" but more a "we can do it quicker and hopefully better due to this tool".

[edward]

Note that those shortcomings Graham listed is harkening back to the days when he used Houdini.