D3WM · 项目原始材料
核对原文摘录 · 5 段，支持 6 项资料
原阅读方向 / 输出什么
it produces only a single frame at each query.一句话认识
We leverage this insight by jointly regressing both the deterministic next-state prediction and the spatial uncertainty map, then using the uncertainty map to guide a conditional diffusion输入什么
World models of visual information predict future image frames conditioned on given action sequences.动作是否影响预测
conditioned on given action sequences.材料说明的限制
Our formulation also assumes scene changes are driven entirely by the input action sequence, and does not account for predictable agent-independent dynamics.