Latest news with #AIAnimation
Fast Company
21-05-2025
- Entertainment
- Fast Company
Cartwheel uses AI to make 3D animation 100 times faster for creators and studios
After years of AI disrupting industries and streamlining repetitive workflows, the technology is now poised to transform animation. In 2024, director and writer Tom Paton's AiMation Studios released Where the Robots Grow, a fully AI-animated feature film. Everything from animation and voice acting to music was generated using AI, at a cost of just $8,000 per minute—totaling around $700,000 for the 87-minute production. While IMDB reviewers criticized the film as 'soulless and uninspired,' it proved that AI can deliver full-length animated features at a fraction of traditional budgets. But it's not just filmmakers driving this shift. Indie game developers want to prototype characters and worlds in hours, not weeks. TikTok and social media creators are looking to animate original characters without studio resources. Major brands, too, seek emotionally resonant storytelling without monthslong timelines or ballooning 3D animation costs. The challenge: most 3D animation tools are still slow, technical, and expensive. Hoping to remove these barriers, a team of developers from OpenAI, Google, Pixar, and Riot Games launched Cartwheel, an AI-powered 3D animation platform. Cartwheel promises to make high-quality 3D character animation 100 times faster, simpler, and more affordable. Users can record motion with a smartphone, describe a scene with a text prompt, or pull from a library of expressive 3D movements. The platform's AI transforms input into production-ready animations. Artists can refine them in Cartwheel or export into tools like Unity, Unreal Engine, Maya, or Blender—without disrupting their pipeline. The startup was cofounded by Andrew Carr, a former OpenAI scientist who helped develop Codex and ChatGPT's code generation, and Jonathan Jarvis, former creative director at Google Creative Lab and founder of the animation studio Universal Patterns. The two met after OpenAI, intrigued by Jarvis's concept for a generative animation tool, introduced him to Carr, who had just left the company to explore how AI could make animation more accessible. 'I had a unique job, where I used animation to share complex research concepts clearly within Google, and make prototypes that couldn't yet be built by software. Andrew always wanted to animate, and later invented a way to 'talk' to Blender, a popular open-source 3D software, with computer code,' says Jarvis. 'We always wanted to build tools to help others get ideas moving and sensed the potential to animate in new ways using gen AI, that it would be centered around creative control.' After two years in stealth, Cartwheel is gaining traction. The company recently closed a $10 million funding round led by Craft Ventures, with support from WndrCo (Jeffrey Katzenberg), Khosla Ventures, Accel, Runway, and Tirta Ventures (Ben Feder), bringing total funding to $15.6 million. Over 60,000 animators, developers, and storytellers joined Cartwheel's wait-list during stealth. Early adopters from DreamWorks, Duolingo, and Roblox are already using the platform. 'All of our AI models are developed in-house. Behind the scenes, we've employed careful software engineering to ensure that all the pieces of our system work together in a way that can be plugged into existing animation pipelines,' Carr says. 'Ensuring that the generated animation is properly scaled, moves naturally, and remains consistent throughout has been one of our biggest challenges.' A Creator-First AI Animation Tool While the generative AI field is increasingly crowded, Cartwheel positions itself differently: not as a replacement for artists, but as a tool that amplifies their creativity. 'Animators and creatives don't care if motion is generated, done by hand, motion-captured, or drawn from a library. They just want it to move to tell their story, make their game, or get their job done,' Jarvis says. 'Our motion models can generate a lot of useful animation quickly, but they can't do everything. That's why we love a hybrid approach. Computers are great at finding patterns, but it's the artist who brings the soul.' A key differentiator for Cartwheel is its team. Carr and Jarvis are joined by industry veterans with experience in film, games, and interactive design. Catherine 'Cat' Hicks, former Pixar animation director on Coco, Inside Out, and Toy Story 3, serves as head of Animation Innovation. Neil Helm, head of Interactive Animation, worked on crowd systems at Pixar for Turning Red, Lightyear, Up, and Inside Out 2. The platform's design is shaped by Steven Ziadie, former Sony and Riot designer, while production is led by Buthaina Mahmud, who helped define Unity's real-time animation workflows and developed shaders used in the Spider-Verse films. 'We reached out, and some reached out to us. Over time, we realized we all shared the goal to make storytelling faster, easier, and more powerful,' Carr and Jarvis tell Fast Company. 'Culture is being shaped in increasingly dynamic, interactive, and immersive spaces like Fortnite, Minecraft, and Roblox —all animation-driven experiences. We're building tools for where animation is headed, and that's resonating with industry veterans.' User feedback has helped shape Cartwheel's interface. 'We began with a focus on text to animation. In beta, we learned that while that's compelling in many situations, often folks want to browse motions for inspiration, use video reference, or act out the motion themselves—so we've moved to a multimodal interface,' Carr says. What's Next for Cartwheel? High-quality animation data remains scarce, with most data sets proprietary or lacking in diversity and detail. To address this, Cartwheel is using synthetic data—AI-generated animations that mimic real-world motion—to train and refine its models. 'The next generation of AI companies has to find and curate the hard data types, and do the hard work to refine it and make it useful to people in that field. That's where the value is,' Carr says. 'While at OpenAI, I worked on the science of data quality and was able to generate millions of dollars of model improvements with just a few lines of code. We are following the same path at Cartwheel to ensure we produce the styles, qualities, and delightfulness in our motion data that artists need.' With fresh funding, Cartwheel plans to deepen R&D, grow its team, and bring its platform to broader markets. 'Over the next 12 months, we aim to be a catalyst, enabling both large and small animation projects to flourish,' Jarvis says. 'Ensuring ethically sourced data that empowers artists is fundamental to our approach. We are a team of artists building tools for artists.'
The National
11-05-2025
- Entertainment
- The National
'What is Abu Dhabiness?' Exhibition challenges AI's stereotypical view of the UAE at Venice Biennale
At Zayed University, a student's attempt to portray the UAE through AI animation repeatedly hit a wall. The software kept defaulting to deserts and opulent skylines, ignoring the diversity of the country's urban landscape and culture. This frustration served as inspiration for four faculty members at the university. They began building a new image database aimed at offering a richer, more accurate visual language for Emirati architecture and culture, particularly that of Abu Dhabi. 'The student had this really interesting story prepared and wanted to have the culture and architecture represented in it,' multimedia professor Omair Faizullah says. 'But no matter what she did, she could not get the results she wanted. Instead, it kept generating a stereotypical form – like an Instagram picture – which is not exactly the representation of the area and its culture.' Enter The Dis-Orientalist, which is being showcased at the Venice Architecture Biennale. The project was developed as a collaboration between Faizullah and Lina Ahmad, Marco Sosa and Roberto Fabbri, faculty members from the architecture and interior design department. It amasses a new data pool of images of architecture from Abu Dhabi in an AI model that offers a more nuanced perspective of the emirate. The Dis-Orientalist came directly as a result of technology's evident misinterpretation of the UAE, but it has grown to present significant pedagogical and architectural potential. The project's title is a pun, referencing to an exotic and static perception of the region, but also to a state of disorientation caused by AI. 'We are disoriented by this technology and how these technologies are coming into the field of architecture,' Fabbri says. 'These new tools are changing the profession and we are trying to understand how we should interact with that, but it is also changing teaching in schools and universities. That's where we start putting those two meanings together.' AI is only as good as the data it relies its intelligence on. As such, the team behind The Dis-Orientalist set out to collect thousands of images of structures in Abu Dhabi, most of them examples of modernist architecture. The photographs, Ahmad notes, were sourced from Abu Dhabi Streets, the Instagram account run by Silvia and Alex, European expatriates who have lived in Abu Dhabi for close to a decade. 'We went and looked at the lesser-known Abu Dhabi,' Ahmad says. 'The buildings that were mostly constructed in the late 80s and the 90s.' 'We were interested in collecting the data,' Fabbri adds. 'That's because without a data set, there's no project and the data set defines and determines the output of the project. Thanks to Silvia and Alex, we were able to put together 7,000 image of lesser-known pieces of Abu Dhabi.' Through a grant from Zayed University, the four faculty members began developing an AI model that would come to generate images of new structures based on these 'lesser-known' examples. This was no straight-forward task, and required the team to develop a visual lexicon of architectural elements within the images - pinpointing what constitutes a door or a window. 'The way that things are put together, especially in an architectural pattern, it's all based on a canonical structure,' Faizullah says. 'A window, door or facade can be put together in infinite configurations. The training involved feeding the AI all of these images, and asking the algorithm to start to understand what's what. Especially with the architecture of the UAE, a lot of these terminologies are not very defined. We had to create our own method of introducing those kind of that kind of topology into the training.' The technology, Ahmad adds, may also help identify what is 'Abu Dhabiness'. 'We all live in Abu Dhabi,' she says. 'We look at buildings and neighbourhoods, and we see what we call Abu Dhabiness. But what does that mean? I think that the software is one of the things that we try to extract the DNA of Abu Dhabi, and not only that, but also generate infinite examples of an impossible Abu Dhabi that feels so familiar, but yet it doesn't exist in real life.' So what are the implications for a technology such as The Dis-Orientalist? Faizullah says that it has 'tremendous educational potential, whereby students can sort of learn, understand the history of the region, its cultural heritage, its visual language, and then use it to create newer things". The platform, Ahmad adds, offers 'an infinite example of regional architecture, something that's vernacular, something that's from the region where students could keep looking at and feeding into their design". This may spark a resurgence of modernist elements in contemporary designs, but Fabbri adds that the team is not advocating a modernist renaissance, but rather proposing a new educational platform that may have inspire new techniques and trends in contemporary design. 'Perhaps if you have an intervention in the city centre, maybe then you want to harmonise the new intervention with the existing structure,' he proposes. The project, the faculty members note, is still in its early stage and the aim is to make it as accessible as possible – or, as Ahmad puts it, 'to democratise the conceptual process of conceptual design'. 'We're also thinking of, how can we use the tool to open it beyond architecture?' Ahmad adds. 'So we're starting to have this conversation and dialogues with different disciplines and inviting them to contribute or to think of how this can be appropriated.' The exhibit at the Venice Architecture Biennale is a sneak peek of The Dis-Orientalist. The faculty members are planning to offer a more comprehensive look at the project in a more immersive exhibition, this time presented locally. 'This is our first exhibition for The Dis-Orientalist,' Ahmad says. 'It's in the pipeline to showcase this in a local exhibition.'
