Disney patents video compression using generative models to cut file sizes
Disney has filed a patent for a video compression system that uses generative models to reduce the amount of data needed to store or stream video. Instead of sending every single frame, the system sends a few full images (keyframes) and instructions on how objects move between them, allowing the decoder to reconstruct the missing frames.
In the filing’s own words
A system includes a video decoder, and a video encoder including a processor configured to execute video compression software code to partition video frames into video keyframes and video inter-frames, encode the video keyframes to generate encoded video keyframes, process, using an optical flow model, the video inter-frames to produce forward flows from the video inter-frames to a first video keyframe and backward flows from the video inter-frames to a second video keyframe, generate, using the forward and backward flows, bidirectional flow maps corresponding respectively to the video inter-frames, encode the bidirectional flow maps to provide video decoding metadata, and generate a bitstream including the encoded video keyframes and the video decoding metadata. The video decoder receives the bitstream and produces reconstructed video frames corresponding to the video frames including the video keyframes and the video inter-frames using a generative video model.
What exactly they claim (claim 1)
1. A video encoder comprising: a hardware processor; and a memory storing a video compression software code and an optical flow model; the hardware processor configured to execute the video compression software code to: partition a plurality of video frames into a plurality of video keyframes and a plurality of video inter-frames; encode the plurality of video keyframes to generate a plurality of encoded video keyframes; process, using the optical flow model, the plurality of video inter-frames to produce a plurality of forward flows from the plurality of video inter-frames to a first video keyframe of the plurality of video keyframes and a plurality of backward flows from the plurality of video inter-frames to a second video keyframe of the plurality of video keyframes; generate, using the plurality of forward flows and the plurality of backward flows, a plurality of bidirectional flow maps corresponding respectively to the plurality of video inter-frames; encode the plurality of bidirectional flow maps to provide video decoding metadata for the plurality of video inter-frames; and generate a bitstream corresponding to the plurality of video frames, the bitstream including the plurality of encoded video keyframes and the video decoding metadata.
Facts
- Company
- Disney (Disney Enterprises, Inc.)
- Official title
- Low Bit Rate Video Compression With Generative Video Models
- Publication number
- US20260261686A1
- Application number
- 19446235
- Filed
- Jan 12, 2026
- Published
- Sep 3, 2026
- Inventors
- 5
- Claims
- 20
Classified under
H04N 19/159 Electricity & electronicsH04N 19/119 Electricity & electronicsH04N 19/172 Electricity & electronicsMore from Disney in AI & software
- Disney patents AI character that creates art from your shared memories Sep 8, 2026
- Disney patents AI characters that create art from shared memories Oct 22, 2026
- Disney patents tech to sync translated speech with body gestures Sep 17, 2026
- Disney patents AI characters that pause and react when you interrupt them Sep 24, 2026
- Disney patents video system to detect when guests need help Oct 15, 2026
Summaries are simplified from the filing’s own abstract and first claim. A patent filing is not a product announcement.
Patent data and drawings: U.S. Patent and Trademark Office (USPTO). Nebula is not affiliated with or endorsed by the USPTO.