Video Transcoding System with Motion Prediction Reuse
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Video delivery services face challenges in providing location-specific advertisements, as current methods disrupt the viewing experience and are costly, and advertisers struggle to target relevant audiences, leading to reduced engagement and revenue for regional advertisers.
Innovation Solution
A transcoding system that analyzes video content and user locations to dynamically insert location-based advertisements, using edge servers to decode, replace, and re-encode video segments in real-time, minimizing disruption and optimizing computation by reusing motion prediction information.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If full-screen ad breaks are inserted into video content, then advertisement delivery is achieved, but the viewing experience is disrupted and production costs increase
Solution Approach 1:
The patent segments the video content into manageable units (video segments with identified insertion points) and inserts advertisements at specific temporal locations rather than using full-screen breaks. This allows ads to be integrated seamlessly into the video flow, maintaining viewing experience while achieving advertisement delivery.
Solution Approach 2:
The patent applies different treatment to different parts of the video content by identifying specific insertion points within video segments where advertisements can be placed. Rather than uniformly disrupting the entire video with full-screen ads, the system locally integrates ads at appropriate moments, preserving the overall viewing experience while delivering advertisements.
2Ease of manufacture
If the same advertisement is sent to all users, then advertisement delivery is simplified, but regional advertisers cannot target relevant audiences
Solution Approach 1:
The patent implements a dynamic advertisement selection system that determines which advertisement to insert based on real-time factors including user location, device characteristics, and advertisement availability. The system dynamically selects from multiple candidate advertisements (including local, national, and filler ads) to serve different user segments, enabling regional targeting while maintaining a streamlined delivery process.
Solution Approach 2:
The system incorporates feedback mechanisms by monitoring user location data, device capabilities, and advertisement performance metrics to continuously optimize advertisement selection. This feedback loop allows the system to adaptively serve relevant regional advertisements to appropriate users while maintaining efficient automated delivery.
3Adaptability or versatility
If video content is decoded and re-encoded in real-time for advertisement insertion, then location-specific ads can be delivered, but computational overhead increases
Solution Approach 1:
The patent performs preliminary analysis of video content to identify insertion points and prepare video segments before actual advertisement insertion. By pre-processing the video to mark suitable locations for ad insertion and organizing content into segments, the system reduces the computational burden during real-time delivery, as the heavy analysis work is completed in advance.
Solution Approach 2:
The system extracts and reuses motion prediction information from the original video encoding that is not affected by advertisement insertion. By identifying portions of the video that remain unchanged and reusing their motion data, the system avoids redundant computation during re-encoding, significantly reducing energy consumption while maintaining the ability to deliver location-specific advertisements.
Data Source
AI summary
In some embodiments, a method receives an encoded video and decodes blocks in a frame of the encoded video in which replacement content will be inserted. The blocks are analyzed to determine first blocks of the frame that are affected by insertion of the replacement content and second blocks in the frame that are not affected by the insertion. Information from the decoding of the block is saved for the second blocks. The frame is encoded where the replacement content is inserted to replace the first blocks in the encoding of the frame, and wherein the information from the decoding of the second blocks is reused when performing the encoding of the second blocks.