FaceSim lets you have mixed reality conversations with GPT-4o as a real mixed reality humanoid character - so that you can have a humanoid-mixed reality best friend for all of your hobbies!
Each FaceSim session begins anew, since we want to GPT-4o to be able to change up its look each time. By default, a TV head, where we wait for GPT-4o to determine its voice and generate its own face visualization prompt, for DallE-3 to create the image, which we send back through the vision endpoint, asking GPT-4o if this image looks like what it thinks it looks like. And then we use the faced.io AI3D AvatarSDK pipeline to turn it into a 3D model
Today OpenAI released GPT-4o - a model that natively takes both text + other modalities (such as image and 3d models converted to images and videos) - immediately, we did a quickie hackathon and now GPT-4o lives and looks like what it wants to look like on Quest Pro~!
Built With
- openai
- presence-platform
- realityscript
- unity
Log in or sign up for Devpost to join the conversation.