http://www.fxguide.com/article570.html
Posted on Nov 01, 2009 by Mike Seymour The Foundry have acquired the technology of Sony Pictures Imagework’s (SPI) Katana 3D lighting & compositing package. Katana’s hardcore lighting node based technology will be integrated into an upcoming version of the Foundry’s Nuke 2D package. We spoke to both companies about why SPI would part with such a great in house tool and why the Foundry needs another compositing lighting system, in this online exclusive.
fxguide sat down with Rob Bredow CTO of Sony Pictures Imageworks, Bill Collis Managing Director of The Foundry Visionmongers and its Chief Scientist Simon Robinson to discuss this major strategic move.
Katana as it looks inside SPI currentlySony Pictures Imageworks, (SPI) developed Katana as their primary internal 3d lighting and compositing package. SPI has been using it constantly since Spiderman 3 and on almost every picture since then. It is a very solid and comprehensive package "it is 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" explained Rob Bredow CTO of SPI. "It gets everything ready for the renderer and then does all the compositing".
On the lighting side the package, while not writing shaders, does allow nodal based manipulation of the shaders and speaks to third party renderers such as Pixar’s Renderman and Arnold. It started life as a 3D lighting package some five to six years ago, it takes assets (geometry) "assigns materials, textures, – gets the lighting rigs set up – it is the primary 3D lighting interface that all of our lighting people use" Bredow explains. The system is node based, like Nuke but covering lighting in addition to compositing.
On the compositing side it has all the main compositing tools, color correction, roto, paint, keying, camera mapping and stereo compositing & manipulation etc. "Our previous compositor was Bonsai, and that was developed over a number of years, it had very sophisticated keying, roto tools, etc.. Katana took all those tools in as a starting point and then we built on top of it, so Katana has pretty sophisticated fully featured tools in there". adds Bredow.
A traditional workflow that requires scripting to prebuild comps from a 3D render is that once you have set up the lighting, one of the advantages Katana has is that the comp is already pre-built. You can script more complex pipelines, but you can also "just do it all at once" as Bredow jokes. But it is wrong to think of Katana as just a 3D lighting package with 3D compositor tools. Katana handles live action very well, for example it was Katana that did the hyper plates for G-Force, the SPI team had very extensive mapping of live action photography into the background plates for those scenes to create a moving live action virtual environment. Currently, SPI is gearing up to run full 4K composites through Katana for an upcoming project.
Katana is a remarkably polished package for an in-house development, so the user experience is remarkably good for an in-house package. Katana is internally a floating point, 64 bit application as is almost all the Foundry’s Tool set, which begs the question why would SPI want to sell and why would the Foundry need another package that so overlaps with Nuke?
The answer can be found in the fact that Katana will never be sold by the Foundry as a stand alone product, and SPI are moving to a Nuke pipeline. Given the overlap with Nuke, Katana makes no sense for the Foundry as another product, but merged with Nuke, they take a massive shortcut on the road to where the Foundry wants to go.
While the detailed financials are not known, the deal involves no company ownership either way, nor any loss of jobs or reduction of any R&D staff. This deal was not driven by the current economic conditions for either company. SPI aims to have someone else expand and manage the development of Katana, "we dont want to be a commercial software company" jokes Bredow. The Foundry simultaneously expands its customer base inside SPI, (you just know SPI got a killer NUKE site license out of this deal ) – they get a huge technology leg up on the competition, and they can now expand with a second development team beside the Nuke team handling 3D.
As the products overlap so much, it will not require major re-writes of Nuke to integrate Katana. Bill Collis commented that he would be "really disappointed" if they were not showing the first Katana tools inside Nuke in some form by Siggraph 2010". When a company buys technology it can slow current development while the new team tries to work out the code and see how to integrate it, but in this deal the SPI staff are not being let go, so that team will aid the Foundry in integrating the technology (while not leaving SPI). From SPI’s point of view the deal is understood to fully protect SPI’s patents and IP. If at some point in the future the two companies have a falling out, a "corporate divorce" – SPI is free to use some or all the technology in-house again, and the deal allows for the Foundry to continue on fully independently of SPI.
In this world of freelancers, there is one other major advantage of the deal for SPI, it will be much easier to staff up for any project if they require Nuke compositors and if their in house tools are now widely known outside the company. Training up freelancers unfamiliar with any companies in house tools can be a costly time delay in getting short term staff productive.
While this move brings The Foundry closer to the major SPI studio just as they have separated themselves from Digital Domain, Collis was quick to point out that SPI will get the same level of access to Nuke as any of their other customers. SPI will not get special or early releases of Nuke, and there is no corporate special arrangements that are not fully open to any of their other major customers worldwide. For current Nuke users Collis points out that this will not take resources away from Nuke, as a separate team is being established beside the Nuke team and this will not require any major re-writes of Nuke.
From SPI’s point of view Bredow explains "this is not motivated by short term financial gain, this is most definitely a long term play for us, – we looked at what the Foundry is doing and we wanted to put a bet on them, – they have a proven track record of taking production technology and making it a good commercial industry package, and when we were looking to the future, with say one of our products like Katana, we looked at a range of options". Furthermore he observes "there is an industry wide move towards choosing standards, and streamlining the process, – because you know every studio has some form of nodal lighting pipelines – and we think we have one of the best – so if we can get it into the hands of someone like the Foundry, that is great".
Under the management of the recently promoted Bredow, SPI has shifted its technology playbook, to include offering some code free to the open source community. This experiment was to gauge if the community would respond and contribute to complex visual effects code if presented as open source. Bredow commented to fxguide that the response has been overwhelmingly a huge success. "I was hoping that we’d at least get a couple of peoples attention and it might get adopted at a couple of places, but I just had no idea how much interest there would be. Our Field 3D plugin we have literally people talking to us from all industries, including people who put rockets into space – who are interested to see if this format is a good fit for them, and our OSL (Open Shading Language) – I cant think of a studio we have not spoken to about that, – that code is just about ready to release, and we are hoping to get that out in the wild by the end of the year… it has been very exciting". Bredrow feels If SPI can help the industry in any way, he thinks that would be great, "we want the industry to be healthy and collaborative".
In the future Bredow is open to exploring this Katana style of deal with other in-house tools, but for now he comments "it is in my best interest to see wide adoption of this technology in the industry, and internally we have a lot of people that like Nuke, so now we can allow them to use it". Collis himself hinted that they were hoping to do similar deals with other studios in the near future. As a side note, one has to wonder if this does not lead to the Foundry hooking into more industry standard renders in the future, such as say Renderman – rather than just Nuke’s own internal renderer, – a point we could not draw out of the Foundry, other than to say they will be considering it down the track.
