Efficient-LVSM: Faster, Cheaper, and Better Large View Synthesis Model via Decoupled Co-Refinement Attention

Documenti analoghi

• Resolving Representation Ambiguity in Feedforward Novel View Synthesis Transformer via Semantic-Spatial Decoupling
  di: Wu, Yihang, et al.
  Pubblicazione: (2026)
• Attention Itself Could Retrieve.RetrieveVGGT: Training-Free Long Context Streaming 3D Reconstruction via Query-Key Similarity Retrieval
  di: Zou, Zichen, et al.
  Pubblicazione: (2026)
• LaGen: Towards Autoregressive LiDAR Scene Generation
  di: Zhou, Sizhuo, et al.
  Pubblicazione: (2025)
• DriveTransformer: Unified Transformer for Scalable End-to-End Autonomous Driving
  di: Jia, Xiaosong, et al.
  Pubblicazione: (2025)
• LVSM: A Large View Synthesis Model with Minimal 3D Inductive Bias
  di: Jin, Haian, et al.
  Pubblicazione: (2024)