Beyond Pixels: From Video Priors to 4D Worlds

TL;DR: Beyond Pixels provides a unified latent-to-4D interface that lifts terminal video DiT latents into explicit dynamic 3D scenes across compatible generation tasks.

• [2026/08/12] 🔥 Beyond Pixels currently ranks 1st on Hugging Face Daily Papers! Thank you all for your support and love! 🤗
• [2026/08/12] 🤗 Many thanks to taesiri for submitting our paper to Hugging Face Daily Papers!
• [2026/08/11] 📄 Our paper is now available on arXiv!
• [2026/08/11] The official repository and project page are available. Code and model release are under preparation.

• Project Page
• Inference Code
• Training Code
• Pretrained Weights

@misc{liu2026pixelsvideopriors4d,
      title={Beyond Pixels: From Video Priors to 4D Worlds}, 
      author={Zihao Liu and Xiaolong Shen and Zhenglin Zhou and Ruijie Quan and Yi Yang},
      year={2026},
      eprint={2608.10744},
      archivePrefix={arXiv},
      primaryClass={cs.CV},
      url={https://arxiv.org/abs/2608.10744}, 
}