My best advice would be to engage a Cisco certified partner to help you in the overall design.
Some quick tips: the majority of Cisco Telepresence endpoints (i.e. C20, EX90/EX60, E20, Profile 42", MX200/MX300, etc, etc) can work as stand alone systems (e.g. they support H.323 and SIP) and don't need UCM/VCS to call another video endpoints. However, you should try to position UCM in this type of designs because UCM 8.6+ has native support of all Telepresence endpoints plus you would be leaving the main component for deploying unified communications in the near future
The Telepresence endpoints should be chosen carefully to meet the business requirements to deliver the best user experience, and Cisco has a broad range of Telepresence endpoints to address those requirements. Based on my experience, I can say that Cisco Telepresence endpoints can deliver an acceptable video experience with low call rates (even over public network that has no QoS guarantee). However, bandwidth must be carefully engineered, and much more, if it will be shared with other traffic (where QoS design will be extremely important).
Finally, if you want to deploy a couple of video enpoints for tests purposes, I would use Telepresence SX20 (CTS-SX20-PHD4X-K9 or CTS-SX20-PHD12X-K9) instead of C20. You can always check part numbers from datasheets: