Based in Vancouver, BC, Image Engine (http://image-engine.com) is a visual effects studio that’s known for its work on Neill Blomkamp’s District 9. The company, which also has offices via partner Cinesite in Montreal and London, has contributed to
Game of Thrones, Jurassic World, Straight Outta Compton and
Chappie too. Here, research & development lead Andrew Kaufman provides insight into the studio’s open-source visual effects tools.
Can you tell us a little about the tools Image Engine produces, and what they are designed to do?
“Character and creature work, both the creatures themselves and the worlds they interact with, are at the heart of Image Engine’s strength as a visual effects studio. Our tools are designed to drive and enable creativity from our artists with these strengths in mind, while tying our third-party, open-source, and proprietary software together into a unified studio workflow.
“We initially developed Cortex, a suite of open source libraries providing datatypes and algorithms applicable to a broad spectrum of visual effects, along with support libraries for third-party applications like Maya, Nuke, Houdini, RenderMan, and Arnold. Our more recent open-source framework, Gaffer, is a general purpose node-based application framework, which provides basic tools for procedural scene generation, shader authoring, rendering and image compositing. Gaffer provides a multithreaded deferred evaluation engine, enabling users to work with large 3D and 2D datasets, and develop flexible procedural workflows.”
What would you say is the philosophy that underpins tool development at Image Engine? What do you want to achieve with the development of these tools?
“Our primary philosophy is to take the common visual effects applications and open-source toolsets as a base, and build the glue and plug-ins and added-value tech on top, to enable our artists to work more efficiently — get more iterations on the art without getting bogged down in the semantics. The goal is towards artistic flexibility within an otherwise automatable workflow.”
How many resources are dedicated to tool development at Image Engine?
“The combined R&D and pipeline team today is 11 people strong, and we have a healthy proportion of pipeline TDs embedded into the production departments as well. Our core tools are in constant development by various members of the team, while other, more specific tools might see off and on development over the years based on the needs of production.”
What was the initial thought process behind the main tools that led to their creation and what problems were these tools created to address?
“When we started developing Cortex, there wasn’t much in the way of visual effects specific open source software. The goal was to develop a common codebase which could be both leveraged and enhanced by multiple studios or individuals.
“With Gaffer, we’ve tried to leverage all the hard work that has gone into Cortex, along with the other open source software that has become available in recent years, taking the best of all of it to build a procedural node graph user experience which enables flexible, templatable workflows.”
How do these tools make Image Engine more efficient? Can you give any examples from production?
“The best example has to be Jabuka, our proprietary asset management system. By building on strong foundations with Cortex and Gaffer, we’ve created a flexible database driven system that controls the user experience for data authoring, publishing, quality control and delivery across all departments in the studio. Users are presented with a consistent user interface within any of our main DCC applications, and in many cases, are able to templatize and automatically distribute their artistic decisions across entire sequences. This automation was critical to our work on Kingsglaive: Final Fantasy XV, enabling us to artistically manage multi-shot lighting for the full CG feature.”
How long have you been working on the various tools at Image Engine, and what has their development history looked like?
“We started developing Cortex in 2007 and it has been in simultaneous development and production use since the beginning. The Gaffer frameworks were initiated independently by John Haddon, also in 2007, and have been used and extended in production at Image Engine since they were open-sourced in 2011. Both projects have become vital to the entire Image Engine pipeline, as have many of the proprietary tools built on top of them, like Jabuka.”
Do you believe that all studios should have a robust and dedicated R&D department dedicated to the production of in-house tools?
“Not necessarily, but it can certainly help drive studio success in many circumstances. Dedicated in-house development has been a part of Image Engine’s strategy from the moment we transitioned to a film oriented visual effects studio. A big part of the reason Image Engine was able to scale up to blockbuster level film work while retaining a mid-sized studio culture is due to the stability of the pipeline we’ve built, and that just wouldn’t be possible without in-house tool development.”
What are the company’s plans for these tools in the future? Will we see more tools?
“We’re always making new tools and improving old ones while trying to build on the foundations we’ve created with Cortex and Gaffer. We’ll continue to evolve those foundations, enriching them with the latest and greatest open source tech from the visual effects community, and providing added value on top, both as open source and proprietary toolsets and workflows.”