DreamX-World 1.0: A General-Purpose Interactive World Model

DreamX Team ,

Yancheng Bai ,

Rui Chen ,

Xiangxiang Chu ,

Rujing Dang ,

Hao Dou ,

Bingjie Gao ,

Qiwen Gu ,

Siyu Hong ,

Jiachen Lei ,

Geng Li ,

Jifan Li ,

Ruimin Lin ,

Qingfeng Shi ,

Bingze Song ,

Lei Sun ,

Jing Tang ,

Ruitian Tian ,

Jun Wang ,

Jiahong Wu ,

Pengfei Zhang ,

Shen Zhang ,

Jiashu Zhu

AK's Picks Embodied AI and Robotics World Models GenAI TDIG SCCE

2026年06月15日

DreamX-World 1.0 是一款面向可控长时序生成的通用型交互式“文本/图像→视频”世界模型。它支持相机视角导航、对先前已观测区域的回访，以及在照片级真实感、游戏风格和艺术化等多种视觉域中按提示触发事件。我们的数据引擎融合了具备精确相机建模能力的 Unreal Engine 渲染结果、富含动作交互的游戏实录视频，以及通过几何重建恢复相机参数的真实世界视频。在相机控制方面，我们提出了 E-PRoPE——一种轻量化的投影式位置编码变体：它继承了 PRoPE 对投影相机几何结构的建模能力，同时将相机感知注意力机制应用于空间维度压缩后的令牌（tokens）。我们通过因果强制（causal forcing）、类 DMD（Dynamic Mode Decomposition）风格的知识蒸馏，以及长序列 rollout 训练，将一个双向视频生成器改造为仅需数步即可完成自回归推理的世界模型。在自生成的长时序上下文上进行训练，使模型能够接触自身此前生成的历史内容，从而显著缓解因分段式自回归生成所累积的风格与色彩漂移问题。基于内存条件的场景持久性（Memory-Conditioned Scene Persistence）通过基于相机几何关系的检索机制，从历史帧中召回早期视角；而残差复用（residual recycling）则降低了条件路径对记忆隐变量（memory latents）精度不足的敏感性。事件指令微调（Event Instruction Tuning）赋予模型可组合的事件控制能力；强化学习对齐（reinforcement learning alignment）则在知识蒸馏后有效恢复相机控制精度与整体视觉质量。借助混合精度 DiT（Diffusion Transformer）推理、残差复用、75% 稀疏化的 VAE 解码，以及异步流水线并行技术，DreamX-World 1.0 在八块 RTX 5090 GPU 上最高可达 16 帧/秒（FPS）的推理速度。在我们设定的 5 秒基础评测任务中，DreamX-World 1.0 的相机控制得分达 73.75，综合得分达 84.76，综合得分优于 HY-WorldPlay 1.5（80.79）与 LingBot-World（80.45）。

DreamX-World 1.0 is a general-purpose interactive text/image-to-video world model for controllable long-horizon generation. It supports camera navigation, revisits to previously observed regions, and promptable events across photorealistic, game-style, and stylized domains. Our data engine combines camera-accurate Unreal Engine rendering, action-rich gameplay recordings, and real-world videos with recovered camera geometry. For camera control, we introduce E-PRoPE, a lightweight variant of projective positional encoding that retains PRoPE's projective camera geometry while applying camera-aware attention to spatially reduced tokens. We convert a bidirectional video generator into a few-step autoregressive world model using causal forcing, DMD-style distillation, and long-rollout training. Training on self-generated long-horizon contexts exposes the model to its own generated history and reduces the style and color drift that accumulates across autoregressive chunks. Memory-Conditioned Scene Persistence retrieves earlier views through camera-geometry-based retrieval, while residual recycling makes the conditioning path less sensitive to imperfect memory latents. Event Instruction Tuning adds composable event control, and reinforcement learning alignment recovers camera control and visual quality after distillation. With mixed-precision DiT execution, residual reuse, 75\%-pruned VAE decoding, and asynchronous pipeline parallelism, DreamX-World 1.0 reaches up to 16\,FPS on eight RTX\,5090 GPUs. On our 5-second basic evaluation, DreamX-World 1.0 achieves a camera-control score of 73.75 and an overall score of 84.76, outperforming HY-WorldPlay 1.5 and LingBot-World in overall score, which achieve 80.79 and 80.45, respectively.

许愿