Sora 2 AI video launch (scarystories.live)

OpenAI has launched Sora 2 as a standalone app across the US and Canada, moving text-to-video from research demo to a production-ready tool. Built on GPT-5, Sora 2 accepts loose treatments, scene beats, reference clips or animatics and preserves narrative logic while resolving camera direction automatically. Key upgrades include multimodal inputs (voiceovers, rough animatics), native audio generation (dialogue, ambience, scoring delivered in a single render), improved motion/physics to eliminate “jelly” artifacts in dolly and handheld moves, and longer single-shot outputs (~60 seconds). Team-focused pipeline features—saved workspaces, batch render queues, and preset sharing—make it practical for studios and agencies to iterate at scale. For creators, especially in horror, Sora 2 accelerates mood boards, animatics, rapid previs and pitch decks by producing minute-long atmospheric sequences that sell tone before full production. OpenAI adds clearer prompt feedback and custom safeguards, plus an opt-out for copyright holders—sparking expected legal and union debates around training data, consent for scanned performers, usage logs, and potential watermarking rules. Practically: use Sora 2 to lock camera blocking, pacing and ambience, but pair it with specialized generators when you need beat-perfect jump scares or voice/narrator tuning optimized for dread.