GB2536032A

GB2536032A - Method for processing video clip for live-broadcast replay

Info

Publication number: GB2536032A
Application number: GB1503712.0A
Authority: GB
Inventors: Pan Chongguang; Li Bingyu
Original assignee: Carnegie Tech Invest Ltd
Current assignee: Carnegie Tech Invest Ltd
Priority date: 2015-03-05
Filing date: 2015-03-05
Publication date: 2016-09-07
Also published as: GB201503712D0

Abstract

A method for processing video clips for live-broadcast replay involves establishing OUT-point setting rules, viewing the captured game video and setting an IN point (mark-in / edit point) when a critical motion/event occurs to indicate the beginning of a clip. An OUT-point setting rule is automatically chosen and applied, from which a corresponding OUT point (mark-out) is calculated, so as to form a clip record to be loaded into a playlist for use in a later replay. During replay, the required clip records are sought from the playlist, and contents stored in the clip record are obtained, thus providing the corresponding video, start position and end position required to access to defined clip. The playlist is arranged such that multiple clips of the same critical motion/event (captured by different cameras) are played successively to provide seamless playback of the action/event from different viewpoints/angles. During live broadcast, the start position and the end position of a clip can be tuned forward or backward according to needs. The simplified operation significantly reduces manual intervention and reduces the risk of human error.

Description

METHOD FOR PROCESSING VIDEO CLIP FOR LIVE-BROADCAST

REPLAY

BACKGROUND OF THE INVENTION

1. Technical Field

The present invention relates to a method for processing video clips for live-broadcast replay.

2. Description of Related Art

Presently, for live broadcast of a sports game, multiple video cameras are deployed to capture live videos from different viewpoints, and these video io sources have to be in-site processed for making clips for immediate replay. In the process, a producer monitoring the captured video sources sets an IN point at the beginning of a critical motion, and sets an OUT point at the end of the very critical motions, so as to define and store a video clip. Then the producer writes the recording site and name of the clip to form a retrievable index keyword, is which allows retrieval of the video clip when a corresponding replay is required during the live broadcast. During a session of live broadcast, it is common to collect different video clips of one or more critical motions for serial replay. For doing so, the prerequisite is to find the recorded video clip that meet particular requirements. At the end of one clip, the start point of the next clip has to be found for activating the next clip, and the playback tends to be less smooth at the end of a previous clip to the start of a subsequent clip. For solving this problem, the traditional approach involves downloading the clips, editing the clips in an off-line environment into a new video, and uploading the new video for the live broadcast system to call on for replay. In addition, one machine can only record clips from a single input video. Thus, for a sports game having different videos recorded from multiple viewpoints, multiple machines have to be used in order to make a complete collection of video clips as required by the live broadcast.

SUMMARY OF THE INVENTION

One objective of the present invention is to provide a method for processing video clips for live-broadcast replay, wherein the method allows forward or backward adjustment of a start position or an end position of a clip according to practical needs, so as to significantly reduce labor costs for processing live broadcast videos taken from multiple viewpoints, and effectively reduce the risk of human disoperation.

Another objective of the present invention is to provide a method for processing video clips for live-broadcast replay, wherein the method allows simultaneous recording of plural clips, so as to significantly reduce labor costs to and material costs for performing live broadcast of multi-view videos of a sports game, and effectively reduce the risk of human disoperation.

Still another objective of the present invention is to provide a method for processing video clips for live-broadcast replay, wherein seamless, continuous playback of plural clips can be achieved without downloading the clips and 15 editing the clips in an off-line environment into a new video.

The disclosed method comprises the following steps: Step 1: establishing OUT-point setting rules according to actual situations of a certain game, and setting different durations or a single duration for the OUT-point setting rules; Step 2: capturing at least one stream of a game video; Step 3: viewing the captured game video, and setting an IN point when a critical motion occurs; Step 4: automatically choosing and applying one said OUT-point setting rule, and obtaining corresponding duration information related to the chosen zs rule; Step 5: according to the duration information and a position of the IN point, calculating a position of the OUT point, so as to form a clip record, which at least includes a name of the game video, and information about time-based positions of the IN point and the OUT point in the game video; Step 6: setting a file name for the clip record, and loading the clip record into a playlist so that the clip record is allowed to be later called on for the replay; Step 7: during the replay, searching for the required clip record from the playlist, and obtaining contents stored in the record, thereby finding out the corresponding game video and a start position and an end position of the clip, wherein the start position and the end position of the clip can be tuned forward or backward during live broadcast according to real-time needs.

Steps 2 through 6 can be realized as below: 1(:) Step 2: for an identical game scene, capturing N game videos from multiple viewpoints synchronously, wherein the game videos are named as A1, A2, ..., AN, respectively, and choosing one said game video for recording the clips; Step 3: viewing the chosen game video, and setting the IN point according to a critical motion; Step 4: automatically choosing and applying one said OUT-point setting rule, and obtaining corresponding duration information related to the chosen rule; Step 5: according to the duration information and a position of the IN point, calculating a position of the OUT point, so as to form a clip record, which at 20 least includes a name of the game video, and information about time-based positions of the IN point and the OUT point in the game video: Step 6: setting a file name for the clip record, and loading the clip record into a playlist so that the clip record is allowed to be later called on for the replay; with reference to the generated clip record, generating simultaneously N-1 clip records, wherein the N clip records have the same time-based positions of the IN point and the OUT point in the game video, but have different game video names, namely A1, A2, ..., AN; respectively; setting file names for the N-1 clip records, and loading the clip records into the playlist.

where in Step 6, the automatically set file name of the clip record has to be 30 connected to a game information system in Step 1 and a file-name generating strategy has to be set, in Step 6, the game information system, according to progress of the game, provides information about players and motions, according to the information and the present file-name generating strategy to automatically generate the file name of the clip record.

The Step 7 can also be realized as below: Step A: in the playlist, according to the file names of the clip records to search N said clip records to be successively played, extracting information from the clip records, and forming an index including N play nodes arranged in a playback order, namely an index including: Play Node 1, Play Node 2, ..., Play Node N, wherein the contents of each said play node includes the game video name, and information about time-based positions of the IN point and the OUT point in the game video; and Step B: during the replay, first according to the contents of Play Node 1, namely the game video name 1, and the information 1 about the time-based positions of the IN point and the OUT point in game video, finding out the corresponding game video and the start position and the end position of the clip in the game video, outputting and playing the clip corresponding to Play Node 1, when the clip is played to S seconds before its end, according to the contents of Play Node 2, finding out the corresponding game video and the start position zo and the end position of the clip in the game video, starting to pre-read video data of the clip corresponding to Play Node 2, and storing the pre-read data in a memory, when the clip corresponding to Play Node 1 is played to its end, directly reading and playing the pre-read data from the memory, and so on, until the clip corresponding to Play Node N is played to its end, thereby achieving zs seamless playback.

The Step B involves: opening three video-playing windows, wherein the first video-playing window and the second video-playing window are used, according to the contents of the play node, to find out the corresponding game video and the start 30 position and the end position of the clip in the game video, to play the video clip, and to output video data to the third video-playing window according to a setting, and the third video-playing window serves to play video signals of the live broadcast; during the replay, to direct the first video-playing window according to the contents of the play node currently played, to find out the corresponding game video and the start position and the end position of the clip in the game video, playing the video data from X seconds before the start position, directing the third window to show the clip being played by the first video-playing window when the first video-playing window plays to the start position, when the first video-playing window plays to X seconds before the end position, directing the second video-playing window according to the contents of the next play node to be played, finding out the corresponding game video and the start position and the end position of the clip in the game video, playing the video data from X seconds before the start position, when the second video-playing window plays to the start position, directing the third video-playing window to show the clip being played by the second video-playing window, when the second video-playing window plays to X seconds before the end position, directing the first video-playing window according to the contents of the next play node to be played, finding out the corresponding game video and the start position and the end position of the clip in the game video, playing the video data from X seconds before the start position, directing the third window to show the clip being played by the first video-playing window when the first video-playing window plays to the start position, and so on, until all the play nodes are completely played.

The video-playing window is a thread.

With the foregoing technical scheme, the present invention provides the clips for replay by directly recording the start time and the end time of a clip in the game video, but not capturing and storing separated video data from the game video. Therein, the end time can be automatically generated according to the preset rule for setting an OUT point. This significantly reduces the operational steps required for making video clips, thereby effectively save labor costs for live broadcast. Since the present invention stores video signals continuously during live broadcast, when it is necessary to replay a clip, the method does not use the actually stored video data, but captures the name of the game video and the information about the start time and the end time of the clip in the game video. Thus, during live broadcast, the start position and the end position of the clip can be tuned forward or backward according to needs. In addition, it is possible to determine in site when to switch to other input signals, and the risk of black screen due to delay caused by switching can be eliminated. This also effectively reduces the risk of human disoperation. In the event that the io traditional method where video clips are stored separately, there would be no video data beyond the end point, and if the operator fails to switch to other input signals at the end point of the recorded video clip, broadcast errors such as black screen could happen. Furthermore, since the traditionally recorded clip can only have the live broadcast contents be adjust backward after the start position of the clip or forward before the end position of the clip. Therefore, the operator has to highly focus on setting the start position and the end position, and once the operator makes any mistake, the contents to be played cannot be adjusted according to practical needs. The traditional method is thus relatively inflexible. Besides, it is difficult for one operator to setting IN points and OUT points of recoded clips for multiple game videos simultaneously. To sum up, the traditional method is a work requiring considerable labor costs.

On the other hand, the disclosed method involves capturing N game videos for one game scene from multiple viewpoints synchronously, and choosing one said game video for recording the clips, and choosing IN points and OUT points for critical motions. In addition, N-1 game videos synchronously copy the foregoing information about the time-based positions of the IN point and the OUT point in the game video to form their respective clip records, thereby significantly reduce labor costs and material costs for live broadcast of multi-viewpoint videos of a sports game. The present invention is based on the theory that across the game videos of the same sports game taken from multiple viewpoints synchronously, a critical motion must occur at the same time point. Therefore, by referring to the time points for the clip of one of the game videos, the time points for the corresponding clip in another game video can be identified.

Moreover, instead of editing the clips' data into a new video file, the present invention simply calls on the records of the clips to be successively played, and makes an index file according to the order of playback. Thereby, the workload is significantly reduced, and the efficiency of live broadcast can be improved. Additionally, the present invention implements a memory that has a io read rate typically higher than that of a magnetic disk, and when pre-reading the video data of the next play node, directly stores it in the memory, thereby further ensuring the success of seamless playback.

DETAILED DESCRIPTION OF THE INVENTION

According to the present invention, a method for processing video clips for is live-broadcast replay comprises the following steps: Step 1: establishing OUT-point setting rules, according to, for example, actual situations of a certain game, and setting different durations or a single duration for the OUT-point setting rules; Step 2: capturing at least one stream of a game video; Step 3: monitoring the captured game video, and setting an IN point when a critical motion occurs: Step 4: automatically choosing and applying one said OUT-point setting rule, and obtaining corresponding duration information related to the chosen rule; Step 5: according to the duration information and a position of the IN point, calculating a position of the OUT point, so as to form a clip record, which at least includes a name of the game video, and information about time-based positions of the IN point and the OUT point in the game video; Step 6: manually or automatically setting a file name for the clip record, and loading the clip record into a playlist so that the clip record is allowed to be later called on for the replay; wherein in the step of automatically setting the file name of the clip record s includes connecting a game information system in Step 1, in which the game information system provides information about players and motions according to progress of the game, and automatically generating the file name of the clip record according to the information combined with a preset file-name generating strategy; io Step 7: during the replay, in the playlist, according to the file name of the clip record, searching out the required clip record, and obtaining the contents stored in the record, thereby finding out the corresponding game video and a start position and an end position of the clip, wherein the start position and the end position of the clip can be tuned forward or backward during live broadcast is according to real-time needs, wherein for convenient switching of videos during live broadcast, a countdown clock may be provided at the end position of the clip; According to the present invention, Step 7 may be realized by: Step A: in the playlist, according to the file names of the clip records to search N said clip records to be successively played, extracting information from the clip records, forming an index including N play nodes arranged in a playback order, namely an index including: Play Node 1, Play Node 2, ..., Play Node N, wherein each said play node has contents including the game video name, and the time-based position information about the IN point and the OUT point in the game video; Step B: during the replay, first according to the contents of Play Node 1, namely the game video name 1, and the information 1 about the time-based positions of the IN point and the OUT point in game video, finding out the corresponding game video and the start position and the end position of the clip in the game video, outputting and playing the clip corresponding to Play Node 1, when the clip is played to S seconds before its end, according to the contents of Play Node 2, finding out the corresponding game video and the start position and the end position of the clip in the game video, starting to pre-read video data of the clip corresponding to Play Node 2, and storing the pre-read data in a memory, when the clip corresponding to Play Node 1 is played to its end, directly reading and playing the pre-read data from the memory, and so on, until the clip corresponding to Play Node N is played to its end, thereby achieving seamless playback.

Step B can be also realized by: opening three video-playing windows or threads. The video-playing window 1 and the video-playing window 2 are used, according to the contents of the play node, to find out the corresponding game video and the start position and the end position of the clip in the game video, to play the video clip, and to output video data to the video-playing window 3 according to a setting. The video-playing window 3 serves to play video signals of the live broadcast. During the replay, the video-playing window 1 is directed, according to the contents of the currently played play node, to find out the corresponding game video and the start position and the end position of the clip in the game video and to play the video data from X seconds before the start position. When the video-playing window 1 plays to the start position, the video-playing window 3 is directed to show the clip being played by the video-playing window 1. When the video-playing window 1 plays to X seconds before the end position, the video-playing window 2 is directed, according to the contents of the next play node to be played, to find out the corresponding game video and the start position and the end position of the clip in the game video, and to play the video data from X seconds before the start position. When the video-playing window 2plays to the start position, the video-playing window 3 is directed to show the clip being played by the video-playing window 2. When the video-playing window 2 plays to X seconds before the end position, the video-playing window 1 is directed, according to the contents of the next play node to be played, to find out the corresponding game video and the start position and the end position of the clip in the game video, and to play the video data from X seconds before the start position. When the video-playing window 1 plays to the start position, the video-playing window 3 is directed to show the clip being played by the video-playing window 1, and so on, until all the play nodes are completely played. In such a way, seamless, continuous playback of multiple clips can be achieved without using any editing software to rearranging them into a new video file.

Steps 2 through 6 may also be realized as below: Step 2: for an identical game scene, capturing N game videos from multiple 1(:) viewpoints synchronously, wherein the game videos are named as A1, A,, AN, and choosing one said game video for recording the clips; Step 3: viewing the chosen game video, and setting the IN point according to a critical motion; Step 4: automatically choosing and applying one said OUT-point setting 15 rule, and obtaining corresponding duration information related to the chosen rule; Step 5: according to the duration information and a position of the IN point, calculating a position of the OUT point, so as to form a clip record, which at least includes a name of the game video, and information about time-based 20 positions of the IN point and the OUT point in the game video; and Step 6: setting a file name for the clip record, and loading the clip record into a playlist so that the clip record is allowed to be later called on for the replay; with reference to the generated clip record, generating simultaneously N-1 clip records, wherein the N clip records have the same time-based positions of the IN point and the OUT point in the game video, but have different game video names, with the game video names of A1, A1, ..., AN, respectively; and setting file names for the N-1 clip records, and loading the clip records into the playlist.

Moreover, instead of editing the clips' data into a new video file, the 30 present invention simply calls on the records of the clips to be successively played, and makes an index file according to the order of playback. Thereby, the workload is significantly reduced, and the efficiency of live broadcast can be improved. The index file only records the game video's file name, and the clip's information about time-based positions of the IN point and the OUT point in the s game video, so it is possible to tune the clip's start and end positions forward or backward during live broadcast according to need. By comparison, the traditional method is relatively inflexible because once the video file has been made, fine tune is not possible. In other words, the traditional method requires many operators' high concentration, which means great labor costs and material 10 costs.

Based on the technical presupposition of the present invention that replay is realized using the start and end time of a clip directly recorded from a game video instead of capturing and storing video data from the game video, the present invention first call on the record of clips to be played successively, so as is to form an index that includes plural play nodes. During the replay, according to the contents of the first play node, the corresponding game video and the clips' start positions and the end positions are found. Then the video data are output and played. When the video data are played to S seconds before the OUT point of the first play node, according to the contents of the second play node, the corresponding game video and the clips' start positions and the end positions are found. Afterward, pre-reading the corresponding second play node's video data is performed and the pre-read data are stored in the memory. When the first play node's contents are played to the end, the pre-read data is directly read out from the memory and played. The same process is repeated until all the play nodes are completely played, thereby achieving seamless playback. Besides, the present invention implements a memory that has a read rate typically higher than that of a magnetic disk, and when pre-reading the video data of the next play node, directly stores it in the memory, thereby further ensuring the success of seamless playback. This significantly reduces the operational steps required for making video clips, thereby effectively save labor costs for live broadcast.

Particularly, the end time can be generated automatically according to the preset rule for setting an OUT point.

In the present invention, a game scene is captured as N game videos from multiple viewpoints synchronously, and one of the game videos is chosen for recording clips. The disclosed method then involves choosing the IN point and the OUT point for a critical motion, and allowing N-1 game videos to synchronously copy the foregoing information about the time-based positions of the IN point and the OUT point in the game video, so as to form respective clip records. This significantly save labor costs and material costs that would otherwise be required for processing multi-view videos of a sports game for live broadcast. The present invention is based on the theory that across the game videos of the same sports game taken from multiple viewpoints synchronously, a critical motion must occur at the same time point. Therefore, by referring to the time points for the clip of one of the game videos, the time points for the corresponding clip in another game video can be identified. The present invention provides the clips for replay by directly recording the start time and the end time of a clip in the game video, but not capturing and storing separated video data from the game video, and the end time can be generated automatically according to the preset rule for setting an OUT point, making it possible to allow N-I clip records to be produced synchronously by means of copying.

The present invention has been described with reference to the preferred embodiments and it is understood that the embodiments are not intended to limit the scope of the present invention. Moreover, as the contents disclosed herein should be readily understood and can be implemented by a person skilled in the art, all equivalent changes or modifications which do not depart from the concept of the present invention should be encompassed by the appended claims.

Claims

What is claimed is: 1. A method for processing video clips for live-broadcast replay, the method comprising the following steps: Step 1: establishing OUT-point setting rules according to actual situations of a certain game, and setting different durations or a single duration for the OUT-point setting rules; Step 2: capturing at least one stream of a game video; Step 3: viewing the captured game video, and setting an IN point when a critical motion occurs; lc) Step 4: automatically choosing and applying one said OUT-point setting rule, and obtaining corresponding duration information related to the chosen rule; Step 5: according to the duration information and a position of the IN point, calculating a position of the OUT point, so as to form a clip record, which at 15 least includes a name of the game video, and information about time-based positions of the IN point and the OUT point in the game video; Step 6: setting a file name for the clip record, and loading the clip record into a playlist so that the clip record is allowed to be later called on for the replay; and Step 7: during the replay, searching for the required clip record from the playlist, and obtaining contents stored in the record, thereby finding out the corresponding game video and a start position and an end position of the clip, wherein the start position and the end position of the clip can be tuned forward or backward during live broadcast according to real-time needs.
2. The method of claim 1, wherein the Steps 2 through 6 are realized as below: Step 2: for an identical game scene, capturing N game videos from multiple viewpoints synchronously with names of AI, AN, respectively, and choosing one said game video for recording the clips; Step 3: viewing the chosen game video, and setting the IN point according to a critical motion; Step 4: automatically choosing and applying one said OUT-point setting rule, and obtaining corresponding duration information related to the chosen s rule; Step 5: according to the duration information and the position of the IN point, calculating the position of the OUT point, so as to form the clip record, which at least includes the name of the game video, and the information about time-based positions of the IN point and the OUT point in the game video; Step 6: setting the file name for the clip record, and loading the clip record into the playlist so that the clip record is allowed to be later called on for the replay; with reference to the generated clip record, generating simultaneously N-1 clip records, wherein the N clip records have the same time-based positions of the IN point and the OUT point in the game video, but have different game video names, namely Ai, A2, ..., AN, respectively; setting file names for the N-1 clip records, and loading the clip records into the playlist.
3. The method of claim 1 or 2, wherein if in Step 6, the automatically set file name of the clip record has to be connected to a game information system in Step 1 and a file-name generating strategy has to be set, in Step 6, the game information system, according to progress of the game, provides information about players and motions, according to the information and the present tile-name generating strategy to automatically generate the tile name of the clip record.
4. The method of claim 1, wherein Step 7 is realized as below: Step A: in the playlist, according to the file names of the clip records to search N said clip records to be successively played, extracting information from the clip records, and forming an index including N play nodes arranged in a playback order, namely an index including: Play Node 1, Play Node 2, ... , Play Node N, wherein each said play node has contents including the game video name, and the time-based position information about the IN point and the OUT point in the game video; and Step B: during the replay, first according to the contents of Play Node 1, namely the game video name 1, and the information 1 about the time-based positions of the IN point and the OUT point in game video, finding out the corresponding game video and the start position and the end position of the clip in the game video, outputting and playing the clip corresponding to Play Node 1, when the clip is played to S seconds before its end, according to the contents of Play Node 2, finding out the corresponding game video and the start position and the end position of the clip in the game video, starting to pre-read video data of the clip corresponding to Play Node 2, and storing the pre-read data in a memory, when the clip corresponding to Play Node 1 is played to its end, directly reading and playing the pre-read data from the memory, and so on, until the clip corresponding to Play Node N is played to its end, thereby achieving seamless playback.
5. The method of claim 4, wherein the Step B involves: opening three video-playing windows, wherein the first video-playing window and the second video-playing window are used, according to the contents of the play node, to find out the corresponding game video and the start position and the end position of the clip in the game video, to play the video clip, and to output video data to the third video-playing window according to a setting, and the third video-playing window serves to play video signals of the live broadcast; during the replay, to direct the first video-playing window according to the contents of the play node currently played, to find out the corresponding game video and the start position and the end position of the clip in the game video, playing the video data from X seconds before the start position, directing the third window to show the clip being played by the first video-playing window when the first video-playing window plays to the start position, when the first video-playing window plays to X seconds before the end position, directing the second video-playing window according to the contents of the next play node to be played, finding out the corresponding game video and the start position and the end position of the clip in the game video, playing the video data from X seconds before the start position, when the second video-playing window plays to the start position, directing the third video-playing window to show the clip being played by the second video-playing window, when the second video-playing window plays to X seconds before the end position, directing the first video-playing window according to the contents of the next play node to be played, finding out the corresponding game video and the start position and the end position of the clip in the game video, playing the video data from X seconds before the start position, directing the third window to show the clip being played by the first video-playing window when the first video-playing window plays to the start position, and so on, until all the play nodes are completely played.
6. The method of claim 5, wherein the video-playing window is a thread.