AI Startup Viggle Raises $19 Million to Create Controllable 3DCharacters for Memes and Visualizing Ideas
San Francisco, CA – Viggle, a Canadian AI startup specializing in creating controllable 3D characters, has secured $19 million in Series A funding led by Andreessen Horowitz, withparticipation from Two Small Fish. The company plans to use the investment to scale its operations, accelerate product development, and expand its team.
Viggle’s technology allows users to create custom animations by specifying desired actions for their characters. This contrasts with other AI video models that often generate unrealistic movements. Viggle’s model, dubbed JST-1, is trained to understand physics, ensuring morenatural and controllable animations.
We’re essentially building a new type of graphics engine, but purely using neural networks, said Hang Chu, CEO of Viggle. The model itself is very different from existing video generators, whichare mostly pixel-based and don’t really understand the physical structure and properties. Our model is designed to understand these, which is why it’s much better in terms of controllability and generation efficiency.
Users can create animations in three ways:
- Uploading a video and a character image: This allows users to replace a character in an existing video, such as the viral meme where Lil Yachty was replaced by Joaquin Phoenix’s Joker.
- Uploading a character image and text prompts: Users can provide instructions on how they want the character to move.
- Creating animationsfrom scratch using text prompts: This allows users to bring their ideas to life without any pre-existing footage.
While Viggle’s technology is gaining traction for creating memes, its potential extends beyond humorous content. Chu envisions its application in various creative fields, including filmmaking, animation, and video game development.
We’re seeing a lot of adoption from creative professionals who are using our model as a visualization tool, Chu said. It’s a powerful way for them to bring their ideas to life quickly and easily.
Training Data Raises Concerns
However, Viggle’s reliance on YouTube videos for trainingits AI model has raised concerns about potential copyright infringement. In an interview with TechCrunch, Chu initially confirmed that Viggle uses YouTube data for training. However, a spokesperson for the company later retracted the statement, claiming that Chu had spoken prematurely and that Viggle could not disclose details about its training data.
Thisbacktracking comes after YouTube CEO Neal Mohan stated in April that using YouTube videos to train AI text-to-video generators would be a clear violation of the platform’s terms of service. While Google, the parent company of YouTube, may have contracts with some creators for using their videos in training sets for projectslike Google DeepMind Gemini, unauthorized use of YouTube videos is strictly prohibited.
Viggle’s use of YouTube data is not an isolated incident. Numerous AI model developers, including OpenAI, Nvidia, Apple, and Anthropic, have been reported to use YouTube video transcripts or clips for training. This practice highlights agrowing ethical dilemma in the AI industry, where the lines between public data and copyrighted content are increasingly blurred.
Future Plans
Despite the controversy surrounding its training data, Viggle remains optimistic about its future. The company is currently offering a free, limited version of its AI model on Discord and its web application.It also offers a $9.99 subscription for increased capacity and provides special access to creators through its creator program.
Viggle is also exploring licensing opportunities with film and video game studios. Chu believes that the company’s technology has the potential to revolutionize the way creative content is produced, empowering both independent creatorsand large studios.
As AI technology continues to evolve, the ethical implications of using publicly available data for training models will remain a critical issue. Viggle’s case serves as a reminder of the need for transparency and responsible data practices in the development of AI-powered tools.
Views: 0