Editors often export a piece, reimport it and export again after a change. Each pass through an encoder removes information permanently, and the final settings cannot bring it back.
Encoding is lossy by design
A delivery encoder discards information it judges imperceptible, keeping what matters most at the chosen bitrate. That judgment is irreversible once the file is written.
Reimporting that file gives the editing system a picture that already carries the encoder's decisions, including the artifacts it introduced in difficult areas.
The second export starts from that degraded picture, so it preserves the first pass's artifacts and adds its own on top of them.
The damage is not proportional
Two passes are worse than twice as bad as one, because the second encoder allocates bits to reproducing artifacts that carry no real information.
Compression noise sits in exactly the high-detail areas an encoder finds expensive, so it competes directly with the picture for the available bitrate.
Raising the second export's bitrate reduces new damage but preserves the old damage more faithfully, which is why the result never matches a single clean export.
Where extra generations sneak in
Most unnecessary generations are introduced for convenience rather than by decision.
- Exporting a rough version for review, then editing from that file.
- Downloading a graphic or clip from a platform instead of using the original.
- Running a file through a separate compression tool before uploading.
Each step feels harmless in isolation, and none is visible in the timeline, which is why the accumulated loss usually surprises the person who caused it.
Intermediate codecs solve the workflow problem
Where an intermediate render is genuinely required, an editing codec is the right choice, because it compresses each frame lightly and independently.
Files are large, but the quality cost of a generation is small enough to be inconsequential, which is the entire reason such codecs exist.
Delivery codecs used as intermediates are the failure case: they were designed for the last step in the chain and behave badly anywhere else.
Working from the project instead
The reliable practice is to keep the project and re-export from it whenever a new version is needed, rather than editing an exported file.
This requires that the project's media remain available, which is an argument for consolidating media alongside the project rather than leaving it scattered across drives.
Where a client requests a change to a delivered file, exporting again from the project takes little longer than opening the file and guarantees a single generation.
Archiving the project alongside the export is what makes that possible months later, when the alternative is reopening a finished file and accepting a second generation.