MV-Forcing: Long Multi-View Video Generation via 4D-Grounded Spatio-Temporal Self-Forcing

  • 类型:arxiv
  • 标识:2607.05376
  • 链接:https://arxiv.org/abs/2607.05376
  • 主分类:multimodal
  • 形态:method
  • 被引:0
  • 被引来源:Semantic Scholar + OpenAlex
  • S2被引:0
  • OpenAlex被引:0
  • 影响力被引:0
  • TLDR:MV-Forcing, a framework that composes temporal and view-wise autoregression within a single diffusion model by introducing a 4D geometric bridge between sequentially generated views, and closes the train-inference exposure bias gap for both temporal and view-sequential autoregression.
  • OpenAlex ID:W7167582037
  • OpenAlex DOI:10.48550/arxiv.2607.05376
  • DOI:10.48550/arxiv.2607.05376
  • DOI来源:OpenAlex
  • 开放获取:green
  • 开放获取链接:https://doi.org/10.48550/arxiv.2607.05376
  • OpenAlex更新:2026-07-19
  • 待LLM分类:否
  • 标题中文:MV-Forcing:通过 4D 几何基础的时空自强迫实现长多视角视频生成
  • TLDR中文:提出 MV-Forcing 框架,通过在顺序生成视角之间引入 4D 几何桥梁,在单一扩散模型中组合时间与视角自回归,并弥合时间与视角序列自回归中训练-推理的曝光偏差差距。
  • 来源文件
  • /inbox/tom/_candidates/2026-07-07-agent-rag-longcontext-candidates.json
  • [S2 enrich]
  • [OpenAlex backfill]