Abstract
PanDWM stitches six camera views into a 360° panoramic band and uses a distilled causal student with a feed-forward 3D Gaussian decoder to generate consistent 4D driving scenes faster and with better FID/FVD than multi-camera baselines.
Driving world models usually generate videos from six pinhole cameras with separate optical centres and small overlaps, and must learn to keep these views consistent. We present PanDWM, which stitches them into one 360^circ panoramic band. Camera boundaries then fall inside one image, so keeping the views consistent means keeping the image continuous, and each pixel still records its camera ray. Conditioned on layouts, a causal student generates the band window by window in four steps; it is distilled from a causal teacher and trained on its own rollouts. A feed-forward decoder lifts each frame along these rays into 3D Gaussians, and static Gaussians from neighbouring frames complete them into 4D scenes. On nuScenes, scored in one stitched 360^circ view, PanDWM has lower FID and FVD than the multi-camera baselines at 8-46times their speed, and keeps its 50-step teacher's FID while lowering FVD by 22-25%.
Community
We present PanDWM, a panoramic 4D driving world model that generates view-consistent 360° scenes from multi-camera layouts. Our approach combines a causal generative model with a feed-forward 3D Gaussian decoder for efficient dynamic scene reconstruction. Experiments on nuScenes show improved generation quality and substantially faster inference than multi-camera baselines. We welcome feedback and discussion.
Get this paper in your agent:
hf papers read 2611.19857 Don't have the latest CLI?
curl -LsSf https://hf.co/cli/install.sh | bash Models citing this paper 0
No model linking this paper
Datasets citing this paper 0
No dataset linking this paper
Spaces citing this paper 0
No Space linking this paper
Collections including this paper 0
No Collection including this paper