Snapshot Capture for a Communication Session

Techniques for snapshot capture for a communication session are described. According to various embodiments, a communication session is established that includes a video feed that is streamed between devices involved in the communication session. The video feed is sampled to extract still images (e.g., frames) from the video feed that include images of a particular participant in the communication session. A snapshot is then selected from the images based on one or more selection criteria, such as an image score for the snapshot. Should a quality of the video feed fall below an acceptable video quality threshold during the communication session, the snapshot is used in place of the video feed for a visual representation of the participant during the communication session.