Creating a 3D animation from scratch has traditionally been a complex, time-consuming, and costly endeavor, demanding the use of advanced software and motion capture tools. However, Cartwheel aims to simplify this process by leveraging AI to make basic animations as easy as describing them, enabling creators to concentrate on more intricate tasks.
“There hasn’t been a lot of innovation in animation since I started my career,” shared Jonathan Jarvis, co-founder and CEO of Cartwheel, and a seasoned animator. “The industry faces a ‘blank screen’ problem, where animators are overwhelmed by countless buttons and options. It’s not uncommon to spend hours working before even seeing preliminary results.”
Cartwheel addresses this issue by automating the initial steps of animation. It converts text descriptions into basic movements, allowing animators to bypass elementary motions such as walking, swatting a fly, or sitting down.
“We deliver the kind of motion capture results you’d typically get from extensive setups or manual keyframing, but much faster,” Jarvis explained. “The value lies in quickly translating an idea into movement, which can then be refined further.”
The platform features a straightforward interface: a character and a text box. Users input descriptions, and within minutes, a basic yet fluid animation is generated, ready for export to any standard 3D editing software.
For those interested, live 3D examples are showcased on their website, including animations like a boxing figure or a solo waltzing character.
“Our model is completely original,” stated co-founder and chief scientist Andrew Carr. “We utilize ethically sourced data and have our own team labeling these motions. Motion is encoded in a matrix incorporating poses, time, and velocity. This motion matrix is then paired with text descriptions, enabling standard training on motion-language pairs, akin to image or video generation.”
According to Carr, the animations produced are “about 80% complete on average,” while Jarvis noted the potential for professional-grade results. Although some outcomes vary, the process is significantly faster and simpler compared to traditional animation workflows, especially in collaborative environments where even minor adjustments require extensive coordination.
The efficiency of their models also stands out. “For a video model, predicting high-resolution frames at 60 fps is challenging. Our models, predicting smaller, manageable matrices, run efficiently on CPUs or older GPUs, facilitating quicker training times,” Carr remarked.
Jarvis even hinted at future capabilities like real-time animation rendering, a breakthrough for game interactivity where character motions are often limited to pre-set actions. Additionally, intuitive camera movements and non-human character animations are in development, although the current focus is on human animations given their widespread demand.
This approach is reminiscent of tools like Wonder Dynamics, which streamline 3D character insertion. The industry increasingly adopts AI-driven tools to save time, allowing creatives to focus on innovative work. For instance, Autodesk’s recent acquisition of Wonder Dynamics underscores this trend.
Speculating on Cartwheel’s future, it could either be acquired as a valuable asset or grow into a platform-agnostic tool offering a suite of services to professional animators. The startup recently secured $5.6 million in seed funding led by Accel, with investments from Khosla Ventures, Human Ventures, Heretic VC, MVP Ventures, Correlation Ventures, and Pelion VC, along with several angel investors.
Currently, Cartwheel positions itself as a prosumer application that animators can rely on. Beta access is available for those interested in testing the platform.
“There’s a misconception that AI will replace creative work,” Jarvis argued. “As a creative professional, I see AI enabling more animation, more motion, and more productivity for individual creators. This empowerment will cascade up to larger studios, leading to unprecedented innovations. But the real creative work happens across various intermediate levels, not just at big studios or among casual users on their phones.”
