88Z AI, 5 August — JD.com has open-sourced its self-developed real-time streaming video editing model, JoyAI-Video-Edit. The model supports editing while playing, achieving 30 frames per second inference at 720P resolution and handling videos of any length. It outperforms similar streaming models like SANA Streaming and LiveEdit in OpenVE-Bench evaluations, with editing quality matching that of offline models.
At a glance
- JoyAI-Video-Edit supports real-time streaming editing at 30 fps in 720P resolution
- The model leads competitors in OpenVE-Bench evaluations with quality comparable to offline models
The update
JD.com has officially open-sourced its self-developed real-time streaming video editing model, JoyAI-Video-Edit. This model features a unique capability to edit videos while they are playing, maintaining an efficient inference speed of 30 frames per second at 720P resolution. It is not limited by video duration, allowing for flexible processing of content of any length. In the authoritative OpenVE-Bench benchmark, JoyAI-Video-Edit comprehensively outperformed existing similar streaming editing models such as SANA Streaming and LiveEdit. Notably, despite using a streaming processing architecture, its final editing quality matches the standards of traditional offline models, achieving a dual breakthrough in both speed and effect.
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.
- 京东黑板报Source published
Source dates above refer to the original announcements; the date under the headline is this article’s publication on 88Z AI.