Real affiliate clones, not stock avatars
Public QA transparency: stress tests that map one winning seed to a creator's face and voice. These prove capability — they are not enterprise-ready guarantees for every output.
What these benchmarks prove
Each row is a capability stress test: product fidelity, emotion/cadence, or full-body motion under difficult shots. Use them to evaluate fit for sample-constrained activation after a winning format — not as a promise that every brand SKU will look enterprise-ready.
- Winning seed video
- Capability under stress
- Not a guarantee
Full-body cadence (Jeremy Fragrance)
Benchmark 01- Actor replacement
- Full-body motion
Problem
Face-only swaps leave stiff bodies and kill the movement that makes UGC convert.
YPH
Jeremy Fragrance seed: skeletal motion, weight, and kinetic energy map to the new identity, not just the face.
Object occlusion (SNUGSG)
Benchmark 02- Actor replacement
- Product occlusion stress
Problem
AI wrappers glitch when a product or hands cross the face.
YPH
Spatial mapping isolates product layers. SNUGSG asset: identity and body swap behind the SKU; textures, shadows, and edges stay untouched.
Emotion & vocal cadence (Superordinary)
Benchmark 03- Actor replacement
- Cadence preservation
Problem
Generic AI faces look dead on shoppable feeds. No micro-expressions or impulse-buying energy.
YPH
Original emotion, finger tracking, and cadence kept; target face and voice re-rendered. Capability stress test — not an enterprise-ready guarantee for every SKU.
