88Z AI, 26 May — Xiaomi EV has released a new framework that deeply couples 3D reconstruction with video generation. The reconstruction module uses sparse 3D anchors, processing 10-second videos in just 10 seconds. The generation module achieves efficient output of 0.19 seconds per frame with 4-step denoising via two-stage training and ODE distillation, supporting videos up to one minute long.
At a glance
- Reconstruction module uses sparse 3D anchors, processing 10-second videos in 10 seconds
- Generation module achieves 4-step denoising and 0.19s/frame efficiency, supporting 1-minute videos
The update
Xiaomi EV has officially launched the new Xiaomi Auto World Model framework, which for the first time deeply integrates 3D reconstruction technology with video generation capabilities. In the reconstruction phase, the system replaces traditional dense Gaussian methods with sparse 3D anchor technology, significantly improving efficiency. This allows the reconstruction of a 10-second video to be completed in just 10 seconds. Regarding video generation, the model utilizes two-stage training combined with ODE distillation to reduce the denoising process to just four steps. This optimization delivers high-speed performance at 0.19 seconds per frame and supports the generation of videos up to one minute in length, demonstrating strong capabilities in dynamic scene construction.
Sources & editorial note
This article is an AI-assisted synthesis by 88Z AI based on the public sources below. Product claims and availability are attributed to their publishers. It does not represent independent testing or first-hand reporting.
- 小米技术Source published
Source dates above refer to the original announcements; the date under the headline is this article’s publication on 88Z AI.