Uthana

Today is the most exciting date in Uthana's history as we announce our funding round and bring our product into Open Beta. Anyone can now experience generative AI for 3D character animation by 1. Uploading any rigged character (regardless of skeleton) 2. Prompting with text or video OR searching a library of 10,000 motions 3. Editing in a web application and downloading in FBX or GLB Below is our explainer video that gives a quick overview of what's available. This is a first step toward Uthana's vision to power?real-time, novel motion in response to narration, environment, and character intent. In furtherance of that vision, we're announcing a $4.3M funding round from IA Ventures, a16z speedrun, Acequia Capital, Hustle Fund, MetaVision VC and others. Thank you to Brad Gillespie, Jesse Beyroutey, Andrew Chen, Andrew Lee, Jonathan Lai, Troy Kirwin, Robin Guo, Badrul Farooqi, Todd Hooper, Elizabeth Yin, Leif Danielsen, Jori Pearsall, Rez Khan, Borja Burgos, Eden Chen, Zach Gonzalez, Jo De Ridder, Dan Bunting ?? GDC, Nick Whiting, Mathilde Pignol ??GDC, Lauren E. Jarvis, Curt Bererton, Tejas Kulkarni, Jeremy Jones, J.T. Trauber for the trust you've put in us and the support provided along the way. To all the game developers, animators, and creators out there—dive in, let me know what you think, and join our Discord. Your feedback is what drives us to get even better. Finally, thank you to the team and my brother. Uthana is made up of truly impressive, dedicated, and smart technologists that are pushing the boundaries of what's possible. Ryan Woodard, Dave Savostyanov, Matt Billenstein, Kam Yip, Gina Condotti, Anthony Thibault, Prerit Jaiswal, Brandon Shelley, Henry Long, Stefanie Syman

In Gen AI for 2D video, one of the tell-tale signs of AI is incorrect fingers (and that’s quickly being solved). In Gen AI for 3D character animation, an equivalent could be feet hitting the floor. ?? In the Video: - blue dots show the positions of the feet before adjustments - the number is the frame number (30 frames per second) - red shots show heel and toe adjustment after an inverse kinematic (IK) function to adjust heels and toes - we further adjust the character's pelvis height to improve realness This type of attention moves each footstep just centimeters, but that adjustment makes the difference between jumping over the uncanny valley or falling into it. In future videos I'll share a side by side of the before and after. ?? At Uthana, one of our values is Go Deep to Compete —obsessing over the details to deliver the best animation tools. Reach out if you're interested in being an early tester of this feature! #AnimationTech #GameDevelopment #AIInnovation

At an #LATechWeek event with Jason Yeh this morning I met a founder (Teddy Thoren) who was excited about Uthana and what we were doing with 3D Animation. He mentioned the NFT collection Forgotten Runes that has thousands of 3D assets. Well, I thought, how long would it take to get one of those assets moving in Uthana? The answer: under 2 minutes. When retargeting "just works" and inference is fast, creation is so much easier. PS - who knows the creators of Forgotten Runes? I'd love to get in touch. cc: Alexis Ohanian Sr. Julien Paredes ?? GameVision Studios

"We use physical simulation to teach characters how to move." "Huh" "So, we combine a generative AI process to predict the next frame of movement and do that within a physically simulated environment (i.e. the physics of our world) so the result is both realistic and controllable. No cameras, no keyframes." "Woah" "Here, let me show you an example." #GenAI #3DAnimation #VideoGames

"So do you guys use OpenAI?" This is a common question I get from folks trying to understand what we build at Uthana. The quick answer is 'we don't.' Where ChatGPT predicts the next word in a sentence, we predict the next frame of movement. Our models are all about creating motion from scratch, and making it be so life-like you can't tell it wasn't a real human that created it. It's a wildly fascinating space to be building in and I'm enjoying educating folks on an area of #genAI that's not LLM-based.