Two-person AI avatar dialogue — What makes a city feel like home?

Completed production jobs matched to credit consumption. The full video and audio decoded without errors; frames, speaker framing, turn-taking where applicable and complete narration transcripts were reviewed. Original file bytes and SHA-256 are recorded. First-party editorial evidence, not an independent review.

The original generated MP4 is unchanged, with no splicing, face repair, replacement audio or added visible watermark. A poster was extracted at 0.1 seconds. An optional English caption sidecar was transcribed locally and checked against the supplied narration; its timing is approximate.

One curated fictional example, not a guarantee of perfect lip-sync, facial motion or identical quality on other inputs. Elapsed time includes preparation and queueing; not a latency benchmark. This credit-funded example does not test free fallback eligibility, its 180-minute wait or a 10-minute generation. No real person was impersonated. The public editorial copy is retained separately from the private 72-hour library.