How ReelStack scores tools
Current status: every score on the site today is a preliminary editorial score assigned by Studio Hall, pending test renders. Benchmark-based scores replace them as tests are published.
Weights (score out of 100, recomputed weekly)
| Factor | Weight | How it is measured |
|---|---|---|
| Output quality | 35% | Fixed benchmark of 10 prompts per category, rendered and scored on a rubric (motion coherence, prompt adherence, artifacts, faces/hands); 10% human spot-check by Studio Hall; all outputs published. |
| Workflow fit | 15% | API availability, MCP/agent support, export formats, commercial license. |
| Value | 15% | Cost per usable second/minute from the public pricing page (diff-tracked). |
| Momentum | 15% | Changelog cadence, public search/traffic trend, X mentions, GitHub stars where relevant. |
| Reliability | 10% | Render failure rate and queue time measured during tests. |
| Community | 10% | Deduplicated sentiment from Reddit and X. |
Rules
- Affiliate status never affects the score.
- Every score change needs evidence (a link or a render).
- Changes over ±10 points are reviewed by a human.
- Pricing and features are shown only when verified, with source and date; everything else is marked unverified.