arXiv:2611.23417v1 Announce Type: new Abstract: We present the Autoregressive View Model (AVM), which generates a scene as a stream of views without constructing an explicit 3D representation. Through a unified in-context interface, AVM conditions on images or videos to generate short clips along a requested camera path. Visual context grounds reconstruction and interpolation, while a learned prior enables extrapolation into unseen regions. AVM retrieves relevant views from inputs and generated history to form a bounded active context. We fine-tune a pretrained bidirectional video diffusion transformer with flow matching in an N-in-1-out formulation. We then tune a causal model on its own rollouts to improve photorealism, reduce denoising steps, and limit long-horizon drift. We scale training to approximately 8M scenes and video clips. We evaluate generation capacity and scene persistence across benchmarks for camera control, text control, visual quality, dynamics, sparse and 4D reconstruction, and long-horizon consistency. The results demonstrate that AVM adapts to diverse input and output settings through one in-context interface.
Autoregressive View Model
A team of researchers posted arXiv paper 2611.23417v1 describing the Autoregressive View Model (AVM), which generates a scene as a stream of views along a requested camera path without building an explicit 3D representation. AVM conditions on images or videos through a unified in-context interface, retrieves relevant views from inputs and generated history to form a bounded active context, and was trained on approximately 8M scenes and video clips by fine-tuning a pretrained bidirectional video diffusion transformer with flow matching in an N-in-1-out formulation, then tuning a causal model on its own rollouts to improve photorealism, reduce denoising steps, and limit long-horizon drift. The authors report the model adapts to diverse input and output settings through one in-context interface across benchmarks for camera control, text control, visual quality, dynamics, sparse and 4D reconstruction, and long-horizon consistency.
AVM generates camera-path video clips without an explicit 3D representation by fine-tuning a bidirectional video diffusion transformer with flow matching in an N-in-1-out formulation, then tuning a causal model on its own rollouts to cut denoising steps and long-horizon drift, trained on roughly 8M
Run your AI side-project on zahid.host
EU-based hosting, git-push deploys, automatic HTTPS, no cold starts. Free tier with a custom domain — perfect for shipping the agent you just read about.