Digital media system

Abstract

A digital media system is configured to support any one or more of multiple features with respect to virtual-reality content. Examples of such features include directional picture-in-picture (PIP) windows based on viewing direction, directional audio based on viewing direction, user recommendations based on anomalous viewing times of certain visual features in video content, dynamic adjustment of time-to-live (TTL) durations prior to requesting deletion of video files uploaded to a content distribution network, dynamic adjustment of durations of video files to upload based on network capacity, dynamic adjustment of quantities of video files per set to upload based on network capacity, dynamic resizing of top-depicting or bottom-depicting regions within the picture areas of sets of video files, dynamic resizing of the picture areas themselves within sets of video files, or any suitable combination thereof

What is claimed is: 
     
         1 . A method comprising:
 training, by one or more processors of a machine, an artificial intelligence to identify one or more visual features depicted in one or more video files;   detecting, by one or more processors of the machine, a viewing duration of a user that viewed at least part of a first video file, the detected viewing duration transgressing a threshold duration whose transgression indicates an anomalous duration for which the user viewed at least the part of the first video file;   identifying, by one or more processors of the machine, a visual feature depicted by at least the part of the first video file by inputting at least the part viewed by the user for the anomalous duration into the trained artificial intelligence, the detected anomalous duration being correlated with the identified visual feature;   updating, by one or more processors of the machine, a viewing preference profile of the user based on the identified visual feature that is correlated with the anomalous duration; and   generating, by one or more processors of the machine, a recommendation that indicates a second video file based on the updated viewing preference profile of the user.   
     
     
         2 . The method of  claim 1 , wherein:
 the generating of the recommendation is based on the second video file depicting the identified visual feature that is correlated with the anomalous duration for which the user viewed at least the part of the first video file.   
     
     
         3 . The method of  claim 1 , further comprising:
 presenting a message that indicates the generated recommendation, the presented message recommending that the user view at least part of the second video file.   
     
     
         4 . The method of  claim 1 , further comprising:
 presenting a warning that indicates the generated recommendation, the presented warning recommending that the user avoid viewing at least part of the second video file.   
     
     
         5 . The method of  claim 1 , further comprising:
 determining that the updated viewing preference profile of the user indicates viewing preferences of multiple users by analyzing the updated viewing preference profile of the user and detecting multiple different clusters of viewing preferences corelated with multiple different times of viewing.   
     
     
         6 . The method of  claim 1 , wherein:
 the training of the artificial intelligence further trains the artificial intelligence to identify one or more viewing angles at which the one or more visual features are depicted in the one or more video files;   the method further comprises: 
 identifying a viewing angle at which the visual feature depicted by at least the part of the first video file is viewed by the user for the anomalous duration; and wherein: 
 the updating of the viewing preference profile of the user is based on the identified viewing angle at which the visual feature is viewed by the user for the anomalous duration. 
 
   
     
     
         7 . The method of  claim 1 , further comprising:
 modifying the second video file by adding a visual highlight that indicates the identified visual feature depicted by at least part of the second video file.   
     
     
         8 . The method of  claim 7 , further comprising:
 presenting at least the part of the second video file to the user, the presented part of the second video file depicting the added visual highlight that indicates the identified visual feature;   detecting a user input that corresponds to the visual feature indicated by the added visual highlight; and   presenting supplemental information to the user in response to the detected user input that corresponds to the visual feature indicated by the added visual highlight.   
     
     
         9 . A non-transitory machine-readable storage medium comprising instructions that, when executed by one or more processors of a machine, cause the machine to perform operations comprising:
 training an artificial intelligence to identify one or more visual features depicted in one or more video files;   detecting a viewing duration of a user that viewed at least part of a first video file, the detected viewing duration transgressing a threshold duration whose transgression indicates an anomalous duration for which the user viewed at least the part of the first video file;   identifying a visual feature depicted by at least the part of the first video file by inputting at least the part viewed by the user for the anomalous duration into the trained artificial intelligence, the detected anomalous duration being correlated with the identified visual feature;   updating a viewing preference profile of the user based on the identified visual feature that is correlated with the anomalous duration; and   generating a recommendation that indicates a second video file based on the updated viewing preference profile of the user.   
     
     
         10 . The non-transitory machine-readable storage medium of  claim 9 , wherein:
 the generating of the recommendation is based on the second video file depicting the identified visual feature that is correlated with the anomalous duration for which the user viewed at least the part of the first video file.   
     
     
         11 . The non-transitory machine-readable storage medium of  claim 9 , wherein the operations further comprise:
 presenting a message that indicates the generated recommendation, the presented message recommending that the user view at least part of the second video file.   
     
     
         12 . The non-transitory machine-readable storage medium of  claim 9 , wherein the operations further comprise:
 presenting a warning that indicates the generated recommendation, the presented warning recommending that the user avoid viewing at least part of the second video file.   
     
     
         13 . The non-transitory machine-readable storage medium of  claim 9 , wherein the operations further comprise:
 determining that the updated viewing preference profile of the user indicates viewing preferences of multiple users by analyzing the updated viewing preference profile of the user and detecting multiple different clusters of viewing preferences corelated with multiple different times of viewing.   
     
     
         14 . The non-transitory machine-readable storage medium of  claim 9 , wherein:
 the training of the artificial intelligence further trains the artificial intelligence to identify one or more viewing angles at which the one or more visual features are depicted in the one or more video files;   the operations further comprise: 
 identifying a viewing angle at which the visual feature depicted by at least the part of the first video file is viewed by the user for the anomalous duration; and wherein: 
 the updating of the viewing preference profile of the user is based on the identified viewing angle at which the visual feature is viewed by the user for the anomalous duration. 
   
     
     
         15 . The non-transitory machine-readable storage medium of  claim 9 , wherein the operations further comprise:
 modifying the second video file by adding a visual highlight that indicates the identified visual feature depicted by at least part of the second video file.   
     
     
         16 . A system comprising:
 one or more processors; and   a memory storing instructions that, when executed by at least one processor among the one or more processors, cause the system to perform operations comprising: 
 training an artificial intelligence to identify one or more visual features depicted in one or more video files; 
 detecting a viewing duration of a user that viewed at least part of a first video file, the detected viewing duration transgressing a threshold duration whose transgression indicates an anomalous duration for which the user viewed at least the part of the first video file; 
 identifying a visual feature depicted by at least the part of the first video file by inputting at least the part viewed by the user for the anomalous duration into the trained artificial intelligence, the detected anomalous duration being correlated with the identified visual feature; 
 updating a viewing preference profile of the user based on the identified visual feature that is correlated with the anomalous duration; and 
 generating a recommendation that indicates a second video file based on the updated viewing preference profile of the user. 
   
     
     
         17 . The system of  claim 16 , wherein:
 the generating of the recommendation is based on the second video file depicting the identified visual feature that is correlated with the anomalous duration for which the user viewed at least the part of the first video file.   
     
     
         18 . The system of  claim 16 , wherein the operations further comprise:
 presenting a message that indicates the generated recommendation, the presented message recommending that the user view at least part of the second video file.   
     
     
         19 . The system of  claim 16 , wherein the operations further comprise:
 presenting a warning that indicates the generated recommendation, the presented warning recommending that the user avoid viewing at least part of the second video file.   
     
     
         20 . The system of  claim 16 , wherein:
 the training of the artificial intelligence further trains the artificial intelligence to identify one or more viewing angles at which the one or more visual features are depicted in the one or more video files;   the operations further comprise: 
 identifying a viewing angle at which the visual feature depicted by at least the part of the first video file is viewed by the user for the anomalous duration; and wherein: 
 the updating of the viewing preference profile of the user is based on the identified viewing angle at which the visual feature is viewed by the user for the anomalous duration.