InfinityStar: Unified Spacetime AutoRegressive Modeling for Visual Generation

2025-11-06

AI Summary

Existing generative frameworks lack a unified, efficient modeling paradigm for high-resolution image and long-duration video synthesis. Method: This paper introduces the first purely discrete, spatiotemporal autoregressive generation framework, enabling text-to-image, text-to-video, image-to-video, and interactive long-video synthesis within a single architecture—without task-specific design or additional optimization. Its core innovations include: (i) the first end-to-end discrete spatiotemporal autoregressive video generation at industrial-grade 720p resolution; and (ii) vision-language-coordinated discretization to jointly model spatial and temporal dependencies. Results: The method achieves 83.74 on VBench, substantially outperforming prior autoregressive approaches. It generates 5-second 720p videos approximately 10× faster than mainstream diffusion models, while also surpassing state-of-the-art diffusion-based methods such as HunyuanVideo in quality.

Technology Category

• Computer Vision: Diffusion Models for Vision
• Natural Language Processing: Generation
• Machine Learning: Deep Generative Models & Autoencoders

Application Category

• Search and Retrieval-Augmented AI: Large language models for search
• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Social Networks and Social Media: Generative AI / large language models and their impact on social systems

Abstract

We introduce InfinityStar, a unified spacetime autoregressive framework for high-resolution image and dynamic video synthesis. Building on the recent success of autoregressive modeling in both vision and language, our purely discrete approach jointly captures spatial and temporal dependencies within a single architecture. This unified design naturally supports a variety of generation tasks such as text-to-image, text-to-video, image-to-video, and long interactive video synthesis via straightforward temporal autoregression. Extensive experiments demonstrate that InfinityStar scores 83.74 on VBench, outperforming all autoregressive models by large margins, even surpassing some diffusion competitors like HunyuanVideo. Without extra optimizations, our model generates a 5s, 720p video approximately 10x faster than leading diffusion-based methods. To our knowledge, InfinityStar is the first discrete autoregressive video generator capable of producing industrial level 720p videos. We release all code and models to foster further research in efficient, high-quality video generation.

Problem

• Unified framework for high-resolution image and video synthesis
• Jointly captures spatial and temporal dependencies in single architecture
• Enables efficient generation of industrial-level 720p videos

Innovation

• Unified spacetime autoregressive modeling for visual generation
• Jointly captures spatial and temporal dependencies
• Generates 720p videos faster than diffusion methods