Methodology
Every tool that earns a score goes through the same 24-point benchmark brief. We publish the input, the settings, the raw output, and the exact rubric — so you can reproduce the test or challenge our verdict.
The 5 Dimensions
Output Quality
Resolution fidelity, temporal consistency, artifacting, lip-sync accuracy, caption correctness, and professional finish.
Speed
Wall-clock time from prompt/upload to downloadable file. Measured on a standard consumer connection (100 Mbps).
Value for Money
Cost per usable output, free-tier generosity, credit expiry policies, and hidden upsells.
Ease of Use
Onboarding friction, UI clarity, documentation quality, and error message helpfulness.
Export Freedom
Watermark status, commercial rights, resolution caps, format options, and API availability.
Evidence Requirements
- Test Run ID with date and tool version
- Complete input prompt or source file (downloadable)
- All settings and parameters used
- Wall-clock render time
- Cost in credits or USD for the test run
- Raw output file (video, audio, or image)
- Screenshot of the tool interface during generation
- At least one "failure" or limitation noted honestly
What “Hands-On Tested” Means
A tool only receives this label when a member of the editorial team has:
- Created an account with their own email and payment method (where required)
- Run the standard brief for the category at least once
- Recorded all 5 dimension scores with written justification
- Uploaded the raw output to permanent storage with a public link
- Written at least 2 specific cons (no tool is perfect)
Limitations & Honesty
We test on English-language inputs by default. Results for Arabic, Persian, or other RTL languages may differ. We test on standard consumer hardware unless stated otherwise. Scores are subjective but reproducible — two reviewers following the same brief should arrive within ±0.5 points on each dimension.