Hyper3D by Deemos has announced the conditional acceptance of its innovative SATO (Strips as Tokens) model to SIGGRAPH 2026, marking a significant advancement in addressing the persistent challenges of topology and UV mapping in 3D Generative AI. The new autoregressive model promises to streamline artist mesh generation with native UV segmentation, a long-standing "headache" for 3D creators.
The company shared the news via a recent tweet, stating, > "🤯Topology & UV — the NO.1 headaches in #3D GenAI. 🔥We just move closer to BOTH at the same time. Introducing SATO: Strips as Tokens, a new autoregressive model for topology & UV, has been conditionally accepted to #SIGGRAPH 2026. Will available at #Hyper3D More details👇"
SATO is described as the first autoregressive framework capable of simultaneously generating mesh geometry and UV chart partitions. This unified approach aims to provide artists with high-quality mesh generation that offers direct surface editing and precise control over connectivity and edge flow, which are crucial for downstream processes like deformation and texturing. The model explicitly encodes UV island boundaries with dedicated tokens, a novel feature in autoregressive generation.
The development comes as the 3D generative AI landscape continues to evolve, with many existing methods struggling to produce artist-friendly, editable 3D assets. Traditional 3D GenAI often results in complex, unoptimized meshes and problematic UV layouts, requiring extensive manual cleanup by artists. SATO's ability to tackle these issues simultaneously is expected to significantly reduce post-generation workload and accelerate content creation workflows.
SIGGRAPH, a premier conference on computer graphics and interactive techniques, is a key venue for showcasing groundbreaking research in the field. The conditional acceptance of SATO underscores its potential impact on the industry. Hyper3D by Deemos plans to make the SATO model available through its platform, Hyper3D, further empowering 3D artists and developers.
