To ensure the highest quality playback experience, you should always begin with the highest quality content possible. This is a standing rule for video transcoding and applies to the entire workflow from capture to delivery.
However, there is a trade-off:
While better source will always create better output, there is a point at which the additional value of working with a higher quality source diminishes and is simply a waste of storage, transit, and time. Assuming your entire production workflow is of the highest quality, the source to use for transcoding can be as small as twice the bitrate of your highest possible deliverable derivative. For example, if you plan to target a maximum bitrate of 8 Mbps, 1080p derivative, you should aim to produce your source content at 16-25 Mbps using H.264. You should consider a higher bitrate if using less efficient video codecs. Transcoding from a source that is higher quality than this is unlikely to provide any noticeable quality improvement.
While working with a source of higher quality is certainly acceptable, it requires additional storage and processing time, and might be difficult to move from location to location. For example, every hour of footage encoded at 10 Mbps consumes roughly 4 GB of storage. A two-hour feature film at 100 Mbps consumes roughly 90 GB of storage and might not produce much visible improvement over the same film encoded at 20 or 30 Mbps when viewed on a desktop or mobile device.
Source content can be in any number of different container or codec combinations, but you should always try to work with source content that is lightly compressed in an industry-standard format to ensure compatibility and quality. Using MPEG containers with the H.264 codecs is highly recommended.