AI meets Design
With great design needs, come better solutions!
Everyone grows up with at least one favorite superhero. Mine? Superman. Now, combine that with my love for animated movies, and watching Spider-Man: Into the Spider-Verse, was like watching magic-on-screen.
Produced by Sony Picture Imageworks Inc.(SPI), the film displayed some of the most creative visual effects and dynamic character animations. In fact, it won a Golden Globe for the Best Animated Feature owing to its rich, detailed and experimental approach that appealed to a wide range of audiences.
SPI has a 26-year, award-winning history of developing software and tools for computer animation and visual effects. But how long does it take to produce an animated feature film? Well, in the words of the writers themselves:
Each second of the 117-minute film took an artist a week to complete.
Let’s find out why…
Script to Screen: The Pipeline
From storyboarding to sound design, the entire production involves 100s of members in each team and tons of expensive software. In fact, if you were to calculate the number of manhours based on the quote above,
You’re looking at 134 years to complete a single animated feature film and that is if you are an expert in every field!
This led me to think: today we have computer-generated animation (CGI), and SPI used very sophisticated animation tools to bring Spider-Man to life.
How long did this process take back in the day, when animation was at its inception and hand-drawn?
Let’s compare the production of Cinderella (1952) to Spider-Man: Into the Spider-Verse (2018). One would hope that the production timeline would shrink owing to all the technology we have at our disposal today but the reality is quite different.
Today, it takes the production team 3X as long from script-to-screen.
There is something very wrong with this picture — pun intended. Truth is that the process of CGI is still very tedious and similar to traditional animation while adhering to many of the same principles.
Even with many sophisticated tools (SPI, Houdini, Maya, Blender etc), the design procedure lacks a seamless and integrated workflow. The tools themselves are complicated to use, require expertise to learn, and have an antiquated “point-and-click” graphical user interphase (GUI).
Clearly, design is broken.
What if artificial intelligence could help design?
After all, AI has immense potential for generating data automatically and reliably. What about AI as a design solution that overcomes the challenge of creating these scenes one by one, using antiquated tools. Such a system would not only optimize the existing workflow but also generate more data in less time and in a cost-effective manner.
Artometa is an AI-driven design solution that builds 3D worlds from natural language descriptions. New emerging technologies relate to and interact with physical systems. Think of how sophisticated and real an experience an animated film is able to provide just by incorporating 3D elements.
We clearly need more data for the real world, in less time and in higher quantities. As such, we are in dire need for a design solution that drives workflow and generates scalable synthetic data for many advanced technologies — not just animation.
Autonomous vehicles training and testing, drones, robotics, mechanical design — all have AI-components and demand constant data. But in reality, existing design tools are not ready to meet this demand. Why?
- Off-the-shelf models and assets lack variance,
- Real-world data leads to over-fitting to specific scenarios, and
- Existing design tools are not optimized for the modern workflow due to their overly complicated GUI.
The scene above, in its dynamic reality, can easily take artists and engineers anywhere from 3-5 weeks to complete. With Artometa, you can create it within minutes!
Real-time synthetic data, with ‘n’ number of iterations, is just a few words away!
