Why Video Gets Blurrier: Generational Loss
A video passes through several encoders on its way from camera to platform: the editor exports once, a converter runs once, the platform transcodes once. Every lossy pass discards information — generational loss, the number-one reason footage grows soft over time.
Where the loss comes from
Lossy encoding is "discard detail the eye will not miss": the encoder blocks the frame, transforms it, then quantizes — and quantization merges near-identical values. The information is truly gone and cannot be recovered. A single high-quality pass is visually invisible, but its output becomes the next encoder's input: the faint block edges left by the first quantization get treated as real detail — and preserved or amplified.
The compounding is uneven: the lowest-bitrate pass does the most damage. A high-bitrate master squeezed through one low-bitrate intermediate loses detail forever — every later high-bitrate encode just repackages the leftovers.
How to avoid it
First rule: stream copy when you can, re-encode only when you must. Container moves like MKV to MP4 with compatible codecs should take the stream-copy path — a new wrapper with untouched content, zero loss, finished in seconds.
When re-encoding is unavoidable, keep intermediates high quality: pick the high-quality side of the compression range (low CRF values) to leave headroom for later passes. Export your edit once with final settings instead of bouncing between drafts. And always keep the highest-quality master — derive every distribution version from it once, rather than stacking derivative on derivative.