Deliver final video files to clients
like a studio.

Every handoff is two files doing two different jobs. The streaming copy exists so the client can press play anywhere and leave notes; the master exists so they own the finished work. Do not make a client scrub a 40 GB download just to leave a comment, and do not let a player re-encode the cut they are approving. Post the web master where it streams byte for byte, run the notes in a review room, and keep the true master one switch away on the same page.

The macOS app supports R3D, BRAW, N-RAW and ProRes on compatible Macs. Playback depends on the codec, hardware, storage and file availability; test a short clip on the recipient’s Mac. For browser review, use a supported H.264, HEVC or AV1 export.

Cloud drives feel free because the storage is already paid for, but the player is where a delivery quietly degrades. Dropbox’s own engineering blog describes previews as transcodes: the source is decoded, re-encoded to H.264, and served as a ladder of quality layers (their 2014 architecture post, still the on-record description). The help pages add the plan math. A Basic owner’s shared link previews at 1080p at most, the player adjusts resolution to the viewer’s connection and screen, and a viewer can stream only the first 30 minutes before a forced download.

Google Drive is blunter: maximum playback resolution is 1920 x 1080 on every plan. A file stored at 5 TB still plays at 1080p, and fresh uploads sit in background processing with no stated time bound; the documented remedy is to try again later. Box caps video previews at 480p on standard Business accounts. None of this touches the stored file. All of it decides what your client sees when they press play on your color work.

Transfer links fail on a different axis. The bytes arrive intact, and WeTransfer states plainly that files are sent exactly as uploaded, but the free tier caps a transfer at 3 GB and the link is designed to expire: a free Dropbox Transfer lasts 7 days and paid ones default to 30. The client who returns months later for the master finds a dead link, and you are back to digging through archive drives.

Bitrate is the number the resolution ceilings hide. The same 200 Mbps export, streamed to a client on each platform:

Approval has to be an action on a specific cut, not a warm sentence in an email thread. In the review room a client approves or requests changes per version, and the mark stays stamped on that exact cut: green on v4 means v4, with the timecoded notes and drawings that led there attached to their frames. That record is what lets you invoice cleanly and treat the next request as a new, scoped round instead of an unpaid revision. One rule protects it: bump the version for every export the client sees, and never overwrite a version that already carries comments, because a note and the frame it points to must stay in sync.

A vertical reframe is a separate cut, not a crop of one already signed off. Deciding what survives when two thirds of the width goes away is a creative call, so a sign-off on the horizontal does not cover it. Put it up as its own film and let it take its own approval.

A delivery closes with three sentences in the final email or the contract, each backed by something on the page: which version was approved and when, who stores the master and for how long, and what counts as a new round. For work that is embargoed before launch, gate the room the way you would gate a screener: password or unlisted access, expiring links, session watermarks, and forensic marking for the leak-sensitive cases. Secure screener sharing covers that tier of caution. Most commercial jobs need none of it, just an unlisted room and a master that stays behind the switch until the invoice clears.