A method comprises receiving video streams from a plurality of mobile devices associated with respective distinct users of a crowd-stream video application, evaluating the received video streams relative to an event of interest, selecting particular ones of the video streams for use in generation of crowd-stream video output based at least in part on results of the evaluating, generating the crowd-stream video output utilizing at least portions of the selected video streams, and providing the crowd-stream video output to one or more servers of a content delivery network for delivery to one or more viewer devices. The video streams illustratively comprise live video streams of the event of interest captured by respective ones of the mobile devices. In some embodiments, evaluating the received video streams relative to an event of interest comprises evaluating the received video streams in terms of at least vantage point relative to the event of interest.