• Home
  • Technology
  • Gaming
  • Entertainment
  • World & Business
  • Science
  • Sports
  • AI
HomeTechnologyGamingEntertainmentWorld & BusinessScienceSportsAI
    • HomeTechnologyGamingEntertainmentWorld & BusinessScienceSportsAI
    • Home
    • Technology
    • Gaming
    • Entertainment
    • World & Business
    • Science
    • Sports
    • AI
    AI

    Hyper3D Releases WorldGen for Photo-to-3D Worlds

    Tweet from @grokkedd claims Hyper3D released WorldGen model using CAST tech.

    RS
    GR
    2 Sources, 28d ago, first seen 28d ago

    TLDR

    @grokkedd posted that Hyper3D dropped WorldGen, an AI turning one photo into an interactive 3D world. The tweet states the model comes from Yingmu Technology and rests on CAST, which won SIGGRAPH 2025 Best Paper alongside entries from Google and Meta. It describes feeding a single 2D image to generate the scene. An attached video shows the WorldGen UI with a living room preview on the left and 3D viewpoint on the right.

    Combined views

    13.9K

    2 Sources, first seen 28d ago

    Combined views

    13.9K

    2 Sources, first seen 28d ago

    209 likes
    209 likes
    8 comments
    228 saves
    40 reposts

    Sentiment

    Positive——Negative

    Summary

    Not enough discussion yet.

    No sentiment analysis available yet.

    8 comments
    228 saves
    40 reposts

    Sentiment

    Positive——Negative

    Summary

    Not enough discussion yet.

    No sentiment analysis available yet.

    Today's Rank

    —

    Not ranked yet

    Today's Rank

    —

    Not ranked yet

    2 Sources

    @grokkeddThis is insane: Hyper3D just dropped WorldGen, an AI that turns ONE photo into a full interactive 3D world Yingmu Technology's new model is built on CAST, the tech that won a Best Paper Award at SIGGRAPH 2025, a prize only Google, Meta, and Hyper3D took home that year. Here's what makes it wild: > Feed it a single 2D image, and it picks out every object, fills in the hidden parts, and rebuilds each one as its own separate 3D model > Tables, chairs, small props: all independently movable, swappable, and editable > Physics comes built in. It figures out collisions, mass, friction, and how objects actually sit or lean on each other > You can also manually select just one object to generate And it's not just a demo toy. Assets drop straight into Blender, Unity, and Unreal Engine, ready for robot training simulations, film production, game dev, and XR. The really wild part: pair it with video models like Seedance 2.5, and you get a full pipeline. WorldGen builds the 3D scene and spatial structure first, then the video model adds character performance, materials, lighting, and style on top, turning basic renders into near-cinematic footage. The line between 3D generation, game engines, and AI video is basically gone at this point.
    @ScobleizerRT @grokkedd: This is insane: Hyper3D just dropped WorldGen, an AI that turns ONE photo into a full interactive 3D world Yingmu Technolog…

    2 Sources

    @grokkeddThis is insane: Hyper3D just dropped WorldGen, an AI that turns ONE photo into a full interactive 3D world Yingmu Technology's new model is built on CAST, the tech that won a Best Paper Award at SIGGRAPH 2025, a prize only Google, Meta, and Hyper3D took home that year. Here's what makes it wild: > Feed it a single 2D image, and it picks out every object, fills in the hidden parts, and rebuilds each one as its own separate 3D model > Tables, chairs, small props: all independently movable, swappable, and editable > Physics comes built in. It figures out collisions, mass, friction, and how objects actually sit or lean on each other > You can also manually select just one object to generate And it's not just a demo toy. Assets drop straight into Blender, Unity, and Unreal Engine, ready for robot training simulations, film production, game dev, and XR. The really wild part: pair it with video models like Seedance 2.5, and you get a full pipeline. WorldGen builds the 3D scene and spatial structure first, then the video model adds character performance, materials, lighting, and style on top, turning basic renders into near-cinematic footage. The line between 3D generation, game engines, and AI video is basically gone at this point.
    @ScobleizerRT @grokkedd: This is insane: Hyper3D just dropped WorldGen, an AI that turns ONE photo into a full interactive 3D world Yingmu Technolog…