The knowledge library

Methodology: evidence and reproducible checks

The difference between documentary comparisons, worked calculations and actual measured editing tests.

Creator Edit Lab distinguishes three kinds of evidence: documented behavior, worked calculations and measured tests. They answer different questions. A documented feature does not demonstrate speed, and a calculator does not establish the quality of an encoded video.

On this page
  1. Documentary comparisons
  2. Calculator verification
  3. A future measured test must include
  4. Limits of the launch library

Documentary comparisons

For software comparisons, identify the relevant operating system, application release and workflow. Use the vendor’s current manual for behavior. Separate import support, decoding, timeline processing and export support. Avoid a single “supports H.265” label when bit depth, chroma, hardware or edition changes the answer.

Calculator verification

Calculators expose their inputs and formulas and run in the browser. Reference checks cover bitrate-to-size inversion, multicamera storage and reserve math, fractional frame rates, crop geometry, PCM size and invalid inputs. Results use stated decimal or binary units. Export-time ranges are user-assumed variation, not confidence intervals.

A future measured test must include

  • Software and operating-system versions, hardware, drivers and power settings.
  • Source media details, timeline structure, effects, export settings and cache state.
  • The measured sample, repeated runs where appropriate, and the time or quality measurement method.
  • Known limitations, failed runs and enough information for another reader to repeat the procedure.

Limits of the launch library

No original physical hardware benchmark is included at launch. Worked examples and proposed test procedures are not presented as completed product tests. Follow the editorial policy and report a questionable claim through Corrections.