Video Switching requires all endpoints to connect at same speed and video algorithms. Those that do not match the conference parameters will connect audio only, no video. The video cards are not used for the processing of the video streams.
Transcoding/Continuous Presence allows each endpoint connecting to be a different speed or video algorithm. The video stream is generated by the video cards and transcoded to whatever format each endpoint is using. Transcoding screen layout is one site only whereas Continuous Presence can have all the different screen layouts available.
If the sites don't want split screen layouts and are all using the same type/model of codecs at the same rate, the Video Switching requires less processing by the MCU (video in - video out) and may provide better video quality. However, if you have a myriad of codec types/models and the sites have different network capabilities (1 BRI, 3 BRI's etc) then Transcoding or Continuous Presence will help ensure the full connection of all sites and a successful conference.