Delivering a wedding film means solving three problems at once: a file too big to email, playback devices you don’t control, and a master that has to survive for decades. Here is how a working studio handles bitrates, codecs, streaming, and archiving — with real numbers.
Start with the arithmetic, because every delivery decision follows from it. Bitrate times runtime is file size: a 4K H.264 export at 60 Mbps is 7.5 megabytes every second, which works out to about 27 GB per hour. A typical delivery set — feature edit, highlight film, full ceremony, speeches — lands between 30 and 60 GB per couple. That is not a number email, chat apps, or most transfer services are built around, which is why “how do I get this to them” becomes a real workflow question rather than an afterthought.
The master is another order of magnitude. A ProRes 422 HQ export at UHD runs roughly 700 Mbps — around 320 GB per hour. That file is not for the couple; it is your archival and re-export source. Keeping the two roles separate — a mezzanine master you keep, a delivery encode couples receive — is the single most useful mental model in this whole topic. Almost every delivery disaster we have seen comes from blurring them: handing couples a file they cannot play, or archiving only the file they can.
| Export | Typical bitrate | Size per hour | Plays for couples? |
|---|---|---|---|
| ProRes 422 HQ (UHD) | ~700 Mbps | ~320 GB | No — edit/archive codec |
| H.264 High, 4K | 50–80 Mbps | 22–36 GB | Yes — the universal one |
| H.265/HEVC, 4K | 35–50 Mbps | 16–22 GB | Mostly — older Windows and TVs struggle |
| H.264 High, 1080p | 16–24 Mbps | 7–11 GB | Yes |
The delivery file has one job: play flawlessly on hardware you have never seen, operated by someone who will not troubleshoot. That rules the choice completely. H.264 High profile, 4:2:0 chroma, 8-bit, in an .mp4 container is the answer, and it has been for a decade — every phone, laptop, tablet, and TV made since roughly 2010 has a hardware decoder for it. It is not the most efficient codec anymore. It is the one that never generates a support email.
H.265/HEVC is genuinely better compression — the same quality at roughly two-thirds the bitrate — and if your delivery pipeline is Apple-heavy it will mostly just work. But “mostly” is the problem: older Windows machines need a codec extension the couple’s dad will not install, and plenty of older smart TVs and Android devices choke on it. The other traps: ProRes is not a delivery codec, ever — most couples cannot open it. And 10-bit H.264 breaks hardware decoders that handle 8-bit fine; if you grade in 10-bit, that depth belongs in your master, not the file you hand over.
A download is ownership. A stream is the actual watching experience. Treating these as competing options is the mistake — they solve different problems and a proper delivery includes both.
Download-only delivery fails at the moment of watching. A 30 GB file does not fit comfortably on a phone, will not AirPlay gracefully from a Downloads folder, and asks the couple’s grandparents to manage file transfers before they can see the ceremony. In practice the film gets watched once, on the laptop it was downloaded to, and then it is an icon in a folder. You spent months on color and sound design for a premiere that happens in a file manager.
Streaming fixes the watching. Proper streaming means the platform transcodes your upload into an adaptive ladder — multiple renditions from a few hundred pixels tall up to 4K — and the player switches between them second by second based on the viewer’s bandwidth. That is what lets a guest on hotel wifi and a parent on a fiber connection tap the same link and both just watch, no codec questions, no buffering wheel. But streaming-only delivery fails at ownership: platforms change terms, links get paywalled, companies disappear. The couple should always end up holding a real file. Stream for watching, download for keeping — deliver both and neither failure mode can reach you.
Every lossy encode discards information, and the loss compounds — encode a file that was already encoded and the artifacts of the first pass become the input noise of the second. This is generation loss, and it is why a film can leave your timeline looking pristine and reach the couple looking like a cable broadcast from 2009. The scenes that die first are exactly the ones weddings are made of: dark dance floors, confetti, rain, fog, film grain. High-entropy frames need bits, and each recompression starves them further — blacks turn to macroblocks, confetti smears into noise.
The rule is one lossy generation between your timeline and the viewer, two at the absolute most. In practice: export your master from the timeline (ProRes, or a very high bitrate H.264 at 80+ Mbps), and make every downstream file from that master — never from another delivery encode. When you upload to a streaming platform, remember the platform will always re-encode; that is unavoidable and fine, provided you feed it your fattest file so its transcode is the first real compression, not the third. Uploading an 8 Mbps “web version” to a service that re-encodes it is how you get mush with your name in the credits.
This is also the argument against re-exporting the delivery file from a previous delivery file when a couple asks for a different format two years later. If you kept the master, a fresh encode takes an evening of render time. If you only kept the 1080p delivery MP4, there is no path back — you cannot un-discard information.
Decide what the master actually is, then protect it. For us it is three things per wedding: the graded mezzanine export (ProRes or high-bitrate H.264), the project file, and the delivery encodes as shipped. Raw camera originals are a judgment call — they are enormous, and honestly, once the film is finished and approved we rarely go back to them. The mezzanine master is the file that lets you re-deliver in any future format; that one is non-negotiable.
The framework is 3-2-1: three copies, two different kinds of storage, one off-site. A realistic small-studio version is a working copy on your edit storage, a second copy on rotating external drives, and a third in cold cloud storage — deep-archive tiers cost on the order of a few dollars per terabyte per month and are perfect for files you hope to never touch. Bare hard drives are cheap but they die quietly on shelves: spin every archive drive up at least once a year, and generate checksums when you archive so “the file copied” and “the file is intact” are provable rather than assumed. Studios with serious volume graduate to LTO tape; most of us never need to.
One warning from experience: do not let a streaming platform become the only home of your master. A host’s copy of your film is a playback asset that exists at their pleasure, in their format. Whatever platform you deliver on, you should be able to get your original file back out — if you can’t, your archive has a hole in it shaped like your best work.
Knotflix is built by Red Bank Studio — we shoot weddings and deliver them on this platform every week, which is why the pipeline looks the way it does. The export from the timeline goes up through the Mac app, which does resumable multi-gigabyte uploads that survive a restart mid-transfer — relevant when the file is 40 GB and the upload runs overnight. The platform transcodes it into a 4K adaptive streaming ladder, and the original file we uploaded stays downloadable, untouched: the streaming copy is for watching, the master remains the master.
The couple gets a private gallery link — no account to create, nothing to install. They stream in 4K on whatever they own, download real files to keep, and their photo gallery lives alongside the films with favorites and full-res ZIPs. Every gallery includes 90 days of hosting; when couples extend it, half of every extension is paid out to the studio, so the long tail of “can you keep it online for our anniversary” became a small revenue line instead of a hosting bill we ate. None of this changes the physics above — you still need the right export, and you still need your own 3-2-1 archive. It just removes the part of delivery that was never the craft.
For the couple-facing file, 50–80 Mbps H.264 High profile is the sweet spot — enough headroom for dark receptions without absurd file sizes. For the master you archive and upload to a streaming host, go higher: ProRes, or H.264 at 80 Mbps and up. Let the platform make the small versions from your big one.
No. Most couples cannot play ProRes, and a 300+ GB file is a burden, not a gift. Give them a clean H.264 MP4 they can open on anything, and keep the ProRes in your own archive so you can re-deliver in any future format they ask for.
Not as the only file. HEVC is excellent compression, but older Windows machines, TVs, and Android devices still fail on it, and the person hitting play will not troubleshoot codecs. If you want HEVC’s efficiency, deliver it alongside an H.264 version — never instead of one.
Every streaming platform re-encodes your upload — that is how adaptive playback works. The quality you keep depends on what you feed it: upload your highest-bitrate master so the platform’s transcode is the first compression generation, not the second or third, and keep the original downloadable as the true copy.
Knotflix is free for 30 days — the full product, $0 today, cancel anytime.