The file your viewers watch is not the file you uploaded. On almost any video service, your upload is converted into several versions at different quality levels, and the player picks between them while the video plays. Knowing why explains a few things about uploading and sharing video.
Progressive MP4, the simple way
A progressive MP4 is one file at one quality. Put it on a web server, point a player at it, and the player downloads it from the start and plays as the data arrives. Seeking ahead works because the server can hand out any byte range the player asks for.
It's simple and it works almost everywhere, which is why MP4 with H.264 video and AAC audio is the default when people say "just give me a video file." Phones, laptops, TVs, and browsers play it.
The limit shows up on a weak connection. If the file was encoded at a bitrate higher than the viewer's connection can deliver, the player runs out of data and stops to buffer. The viewer sees a spinner. The only fix is a smaller file at lower quality, and with progressive MP4 you have to choose that file before anyone presses play. One file, one quality, for everyone, on every network. The 1080p file that looks great on home Wi-Fi is the same file that freezes on a bus.
HLS, the adaptive way
HLS stands for HTTP Live Streaming. It takes a different approach: instead of one file, the video is split into short segments of a few seconds each, and the whole thing is encoded several times at different quality levels, called renditions. A small text playlist lists the renditions and the segments in each.
The player reads the playlist, starts with a segment it can fetch quickly, and measures how fast that segment arrived. If segments are arriving faster than they're being played, it steps up to a better rendition for the next one. If they're arriving slower, it steps down. The switch happens between segments, so the picture changes sharpness but playback doesn't stop. This is adaptive bitrate streaming, and it's what large video platforms do.
Because the segments are ordinary files fetched over HTTP, they can be cached by the same infrastructure that serves images and web pages, which is a big part of why this approach won.
Why one video becomes several
The conversion step, often called transcoding, is where the uploaded file becomes those renditions. It's the part that takes a while after an upload finishes.
Each rendition is a full re-encode of the whole video at a target resolution and bitrate. A 360p rendition needs a fraction of the bandwidth of a 1080p one. Having all of them on hand is what lets the player make its choice in real time rather than guessing in advance.
On FileNext, every uploaded video is converted to 360p, 480p, 720p and 1080p renditions, and playback adapts to the viewer's connection. Nobody has to pick a quality, and the same link works for a viewer on fiber and a viewer on a weak mobile signal. How shared videos play for free viewers and Premium members is on the pricing page; the short version is that free viewers of a shared video get SD, below 720p, and Premium adds HD streaming and original-quality downloads.
What it means for viewers on slow connections
The practical difference is in the first ten seconds and in what happens when the network wobbles.
Startup is faster, because the player can begin with the lowest rendition instead of waiting for enough of a big file to arrive. The video starts soft and gets sharper as the player confirms it has bandwidth to spare.
Freezes become quality drops. When a mobile connection dips, a progressive MP4 stops; an adaptive stream steps down a level and keeps going. For a lecture, a tutorial, or a match, "a bit softer for a minute" is a far better outcome than "stuck at 12:40."
Data use matches the connection. A viewer on a slow link never downloads 1080p segments they can't play in time. A viewer with a fast link isn't stuck at a quality chosen for the worst case.
Lower renditions are visibly softer; that's the trade. If the connection can't carry 720p, the viewer gets something below it. The alternative isn't 720p; it's a spinner.
Why you keep the original
Renditions are derived copies. Each one is a lossy re-encode, and you can't turn a 1080p rendition back into the original, let alone a 480p one. The original is the master, and three things depend on it.
Editing. If you ever need to cut a different version, fix the audio, or crop for a different aspect ratio, you want the source, not a re-encode of a re-encode.
Future formats. Encoders improve. AV1 is newer and more efficient than H.264, and hardware support for it is still uneven, which is why services don't lean on it everywhere yet. When that changes, new renditions get made from originals. Anything that only has the old renditions is stuck with them.
Downloads. When someone needs the file rather than a stream, the original is the file to give them.
FileNext keeps your original alongside the renditions, and you always get your own originals back; the how it works page follows a file from upload to share link. Converting for streaming is for your viewers; it never replaces what you uploaded. The same goes for the cover frame the drive picks for each video: it's a preview, and you can swap it for your own image via "Change cover" without touching the file.
A note on formats before you upload
Because the service converts whatever you upload, container choice matters less than it used to. An MKV that many phone players won't open becomes the same set of renditions as an MP4. HEVC (H.265) files are smaller than H.264 at the same quality and less universally supported, but once converted the viewer never meets the HEVC file directly.
Where format still matters is in the original you keep. MP4 with H.264 and AAC is the safest choice for a master you'll want to open on any device for the next decade. If your camera records in something else, keep that file too; it's the true original.
When progressive MP4 is still fine
Adaptive streaming isn't always the right tool. A short clip embedded on a page you control, a video on a local network, a file you're handing to one person to play from disk: a single MP4 is simpler and there's nothing to convert. The cost of HLS is a conversion step and several copies on disk; for a 15-second clip, that's overhead for nothing.
For anything you're sharing with people whose devices and connections you don't control, especially on phones, adaptive streaming is the difference between a video that plays and a video that buffers. That's the reason the conversion happens, and the reason the original stays where you can get it.