An orchestration system and method are described for at least in part orchestrating a multidevice video session in which a plurality of devices each transmit a video, for example of a user, via a network and in which a composition system receives said transmitted videos and composites, for a respective device from the plurality of devices, videos of other devices into one or more video streams and transmits the one or more video streams via the network to the device. A grouping of the users/devices may be determined to allow the video stream(s) to be generated for respective devices in a manner which may be dependent on the group to which they are deemed to belong. In particular, the video stream(s) may be generated by the composition system such that the videos of devices within a same group are shown in a better quality than the videos of devices from another group, as a user of a device may prefer to see the users from the same group in a better quality than the users from another group.