The FFmpeg documentation tells you what a filter does. It does not tell you what it costs. Every filter here was run against the same source file, and each page carries the exact command, the encode time on its own, the output size and what the job was billed.
A filter transforms a stream as it passes through FFmpeg. Video filters are chained with -vf, audio filters with -af, and anything that needs more than one input or produces more than one output needs -filter_complex instead.
Yes. Every figure on every page in this reference came from a job executed against the Rendobar API, not from an estimate. All 72 variants ran against the same source file so the numbers compare, and the whole set cost $0.193082 to measure.
Because the FFmpeg documentation tells you what a filter does and nothing about what it costs. Encode time and output size are what decide whether a filter belongs in a pipeline that runs a thousand times a day.
No. Every command in this reference ran through the Rendobar API against a URL, with no local FFmpeg and no server. The job id beside each measurement is the real job it came from.
The measurements are produced by a script in the repository and committed, so a page never changes silently. Re-running the script refreshes every number at once, and the measurement date is printed on each page.
Every command in this reference ran through the API against a URL.
