Key Frame Conversion for Low-Latency Video Streaming
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional video streaming techniques suffer from latency and computational intensity, particularly when starting playback at a point other than the beginning of a video, due to the need to decode preceding content and the progressive loss of video quality during transcoding.
Innovation Solution
The method identifies a target key frame preceding the desired start time, converts it to a compatible frame type, and streams only the target frame header, key frame, and subsequent video bitstream, allowing the client device to decode the video efficiently without modifying the original reference list.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of time
If conventional video streaming techniques are used to start playback at a point other than the beginning, then the video can be streamed from the desired start time, but latency increases due to the need to decode preceding content
Solution Approach 1:
The patent converts the target key frame to a first frame type (such as IDR frame) in advance, so that when random access is needed, the client device can decode directly from that frame without needing to decode preceding content. This preliminary conversion eliminates the latency caused by decoding previous frames while maintaining the ability to start playback at the desired time point
2Ease of operation
If transcoding is performed to enable random access at arbitrary points, then playback can start at the desired time, but computational intensity increases
Solution Approach 1:
Instead of transcoding the entire video stream, the patent applies frame type conversion only to the target key frame that is nearest to the desired start time. This localized approach enables random access capability at the specific access point while minimizing computational intensity by avoiding full-stream transcoding
Solution Approach 2:
The patent changes the frame type parameter of the target key frame from its original type to a first frame type (such as converting to IDR frame). This parameter change enables the frame to serve as a random access point without requiring full transcoding, thus reducing computational intensity while maintaining random access capability
3Loss of time
If key frames are converted to enable direct playback, then latency is reduced, but video quality may deteriorate due to progressive loss during transcoding
Solution Approach 1:
The patent converts the target key frame to a first frame type that preserves the original video data without progressive loss. By using frame type conversion rather than full transcoding, the original frame data is copied and reclassified, maintaining video quality while enabling direct playback and reducing decoding delay
4Quantity of substance
If only the target frame header and key frame are sent to the client device, then bandwidth usage is reduced, but the client device must efficiently decode without the original reference list
Solution Approach 1:
The patent makes the target key frame self-sufficient by converting it to a first frame type (such as IDR frame) that contains all necessary decoding information within itself. This allows the client device to decode the frame independently without requiring the original reference list, thus reducing data transmission volume while keeping decoding complexity manageable through self-contained frame information
Data Source
AI summary
In a computing device for streaming media content, information relating to a video bitstream is transmitted to a client device for displaying a user interface. An input signal is received from the client device indicating a desired start time in the video bitstream. Based on the start time, a target key frame is identified within the video, the target key frame comprising a nearest key frame in the video bitstream preceding the desired start time. A target frame header comprising a nearest header frame in the video preceding the target key frame is identified within the video bitstream. If necessary, the target key frame is converted to a first frame type. Only the target frame header, the target key frame, and video bitstream that follows the converted target key frame is sent to the client device, where the original frame referencing encoded in the video bitstream is utilized.