The Foundry announced today that Digital Domain has purchased a site license for KATANA, making them the first customer of the product since The Foundry took over development in 2009. We caught up with Bill Collis for an update on this news as well as more info about open source and support of the recent deep compositing spec that's part of OpenEXR 2.0.
The first sale of a product is always significant and the fact that the first customer is a facility with a stature as Digital Domain. Digital Domain is currently using a pre-release version of KATANA, since the product isn't shipping at this point in time. While DD has been testing the software since the very first alpha releases, Bill Collis, CEO of The Foundry, says that "a number of other large post houses are now working with KATANA and are quite interested in the software."
Work on KATANA by The Foundry began back in November 2009, after acquiring it from Sony Pictures Imageworks (SPI) in a deal which included a site license of NUKE. Rob Bredow, CTO of SPI, referred to KATANA as "a pretty fully featured suite of tools that handles almost all the backend of our pipeline – everything after the animation is done and any processes are simulated - really KATANA takes over from there. It gets everything ready for the renderer and then does all the compositing”.
fxguidetv episode #111 includes a primer on KATANA with Jeremy Selan of SPI and The Foundry product manager Andy Lomas. The episode also includes a discussion regarding Open Source software with Weta digital's CTO Sebastian Sylwan. At FMX2011, Sylwan announced that they were contributing their deep opacity tech to the upcoming OpenEXR 2.0 specification. In addition, they are licensing their deep opacity compositing tools to The Foundry for inclusion in NUKE.
"We've been following the deep opacity tools since Weta had been using them in Shake," says Collis, "and they were interested in us putting them in NUKE but it didn't make sense at the time. Peter Hillman then implemented them as NUKE plugins."
However, The Foundry wanted to wait until file format specifications were nailed down as it didn't make sense for them to develop support for several different types of files. "We had implemented some very basic deep opacity plugins....the basic operators...but Weta had pushed the technology much further," says Collis. "They had put a huge amount of effort into the tech, so it made sense for us to improve and build upon what we had done in house."
The relationship that The Foundry has with Weta in developing Mari has certainly helped facilitate this process, but is even simply the fact that they are also a NUKE customer. "The idea of The Foundry just staying really close with its customers and listening and watching and being part of what they're doing. And then when it makes sense to try and bring that to the mainstream NUKE or some other mainstream product."
It's these two separate items -- the open source OpenEXR 2.0 file format/deep opacity spec and the licensing of tools -- that have served to bring deep composting to the wider masses. Tools will be available in NUKE 6.3 which is due to ship before the end of June. This is actually before the final OpenEXR spec is released, so there will be an additional release shortly afterwards that supports the open source format.
At the FMX2011 open source discussions, one thing that was mentioned time and time again is that the best open source contributions are developed in use at facilities on productions. Effectively, battle tested. Bredow and others at the talks said they feel it is incredibly unlikely to have open source software adopted by facilities that hasn't been used in production. "We believe the best technology is made in the head of a film," says Collis. "It's where the best ideas come up. And if we can take those ideas that have been created when time is pressing and you're just getting the job done...and then later spend the time to commercialize them...that just works brilliantly."
Another common thread of the FMX panels was that it was clear what works best as open source and what doesn't work so well. Standards, file formats, and specific tools work; large scale applications don't. This is why we're seeing open source offerings such as OpenEXR, Ptex, and Alembic, and others.
For applications, there are several issues that serve as roadblocks. First, there are legal issues: can a company ensure that every single line of code is 100% their IP. This isn't to say that the code isn't theirs...it's simply that the time and cost of vetting the code is prohibitive.
Second, it takes time to clean up and comment code to the level that makes it useful for other programmers to contribute to it. Code at facilities is written in production and under tight pressure....this means structure and commenting sometimes isn't up to the level needed and when there are comments maybe they're not the best things. Cleaning up this code takes a tremendous amount of time as well...it's one reason it takes time to get the tech out of facilities and into open source.
In addition, managing large scale applications and the interaction between various modules is a very complex process. In addition, one generally needs a strong personality leading development of a software product -- the key product manager. This is much more difficult to have happen when working with open source software.
As far as developing and maintaining applications in house, the industry has trended towards moving continued app development out of house and to a third party such as The Foundry. You see this reality with Nuke, Mari, and Katana. Collis says all of these facilities are "trying to get away from the idea of massive in-house engineering teams. And I'm sure if you ended up with an open source compositor you'd end up with a very large in-house engineering team. We have source code licenses of NUKE out there and no one touches it. What people actually want is a supported, developed, maintained, documented bit of software with people they can shout at when it doesn't work to have them help make it work."
We've been a free service since 1999 and now rely on the generous contributions of readers like you. If you'd like to help support our work, please join the hundreds of others and become an fxinsider member.