Today we’re announcing two new AI products: Unity Muse, an expansive platform for AI-driven assistance during creation, and Unity Sentis, which enables you to embed neural networks in your builds, enabling previously unimaginable real-time experiences.
Whether you’re just getting started or already a Unity expert, Unity Muse and Unity Sentis are here to help you open up new avenues of creativity and innovation. We’re committed to working closely with you to explore, validate, and define our AI offerings, which is why both will be available via closed beta at this time.
I’m just wondering though since its going to be a closed beta, will there be issues if I use the tools for the game I’m currently developing when muse AI starts out.
This should be very interesting. I look forward to seeing more. What makes this special isn’t really the AI, but the Unity ecosystem working with it to include it in workflows. Hopefully this will improve productivity!
Since the beginning of the year I was thinking of using something similar to GPT for my open world game, but the good solutions ran in the cloud and there were API costs, not good thing for immersive games. Unity Sentis is exactly what I was waiting for!
I hope it has some good and easy training, something like Character.AI.
I’m all signed up! I’m in the beginning prototype phase for a casual family-friendly VR multiplayer game I’m making by myself and launching in 2 years with many more years of updates.
Can we find somewhere on what data Muse was trained? Or at least get the guarantee it’s trained on legally unproblematic data? With how the AI landscape is rapidly evolving, and laws will likely follow with some delay, it would be good to know any generated data is always going to be in the clear if we end up using it in our applications.
I’m assuming that the alien looking character from the Unity Sentis demo is an AI 2D animated model… similar to D-ID generated content, is that right or?
To create Unity Muse Chat, we licensed third party LLMs and integrated them with first-party Unity technical documentation (including manuals, release notes, tutorials, scripting guides, etc.) so creators can get the most relevant and up-to-date information.
I came to hear about lots of concerns regarding the contents generated by AI being used in game development, do we need to provide addtional layer of licensing or anything of that sort if we are using sentis or muse for generating in game contents like materials or characters.
I would assume that applies to the chat-based app/game development instructions. Does this extend to the other functionality you show such as character animation as well as the sprite and texture generation? Is that all based on first-party Unity content or for-purpose source assets from other partners?
Hi. What’s the difference between Barracuda and Sentis? Is Sentis just a rebrand of Barracuda? Barracuda is already an on-device AI model inference solution.
My best guess is if using copyrighted works in a data set without explicit permission is deemed illegal, most AI tools will disappear (including Unity’s), which is something to think about before basing your game and workflow around those tools.
Exciting times. I’ve been using Unity since it was Mac-only, and this development is one of the most exciting ones I have seen in a long time. I’m particularly thrilled that there are capabilities for using AI in the executables, and that there seems to be a commitment for webGL support. Like others asking above, I am very interested in hearing about which training sets have been used to train the LLMs, since I also believe this is crucial for legal reasons.
I have also applied for the beta, and hope to get in.
