Note: full functionality (cameos, Remix, the social feed and some creation flows) is available via the Sora mobile app — currently rolling out on iOS — so plan to use the mobile app for the smoothest experience.

What’s new in Sora 2

• Synchronized audio + visuals — Sora 2 generates aligned audio (dialogue, SFX, ambience) that matches lip- and on-screen events.
• Better physics & materials — improved gravity, cloth, fluid/particle and reflective behavior for realistic product motion (drops, balm ribbons, splashes).
• Remix & Cameo workflows — surgical, timestamped edits (Remix) and user “cameo” insertion are first-class features in the Sora ecosystem. For full social/remix/cameo flows use the Sora app.

Process — 3 simple steps

1. Upload reference images (2–5 min)
   • Upload one clean, high-res hero shot and optionally a label/texture closeup to your Sora project. Describe any people/hands in text (e.g., “adult hand, neutral tone”) instead of uploading photos of recognizable faces unless you have explicit consent.
2. Ideate with ChatGPT (3–8 min)
   • Give ChatGPT: product name, 1-line USP, hero filename (for your own reference), and constraints (e.g., “no faces — describe hands only”).
   • Ask for: a short storyboard and copy-and-paste Sora-ready prompts (camera moves, 1–3 physics micro-directives, a short continuity anchor line, guardrails like “keep label legible”).
3. Paste prompts into Sora, render draft, then Remix (5–30 min per iteration)
   • Render low-res drafts first, then use timestamped Remix instructions to make surgical fixes (e.g., “00:03–00:05 — improve label-region contrast; do not change jar geometry”). Surgical remixes are supported in the Sora workflow.

Quick ChatGPT ideation prompt (copy/paste)

You are my short-form video director & Sora 2 prompt engineer.

Context:
• Product: [Product name] -ATTACH PRODUCT IMAGE INTO PROMPT
• Hero image filename (for my reference): hero.jpg
• One-line USP: [e.g., "Melts makeup, hydrates skin"]
• Shots: 3 shots (each ~10s Sora unit)
• Constraint: 

Deliverable:
1) A concise 3-shot storyboard (title + 10s timestamps + one-line action).
2) For each shot, a Sora-ready prompt with: Short scene/camera instructions (timestamps inside 10s), Action (hand description), 2 physics micro-directives, one continuity anchor line, and 2 guardrails (label legibility, no faces).
Return as three numbered copy-paste prompts.

Example Sora-ready shot prompt

20251007_1852_01k6zs5dy5es5bw4dpd3fbh6bm.mp4

Scene: Medium-close, vanity setting with soft, warm top-left light. Camera starts on hands applying a small scoop of balm to the back of the hand (0–2s), then dollies slightly out and tilts up to show the woman gently massaging the product into her cheekbone/jaw area (2–7s), finish with a soft 1.5s hold on the radiant skin patch and a calm pause for overlay text (7–10s).

Subject: A beautiful woman applying the balm on her face

Action: Using a small stainless spatula or fingertips, scoop a pea-sized amount, warm between fingers, and press into skin with gentle upward motions. No eye contact with camera.

Physics micro-directives:
 • Balm texture: show glossy-to-milk emulsification on skin contact; no grain or aliasing.
 • Skin sheen: subtle, natural specular response (no plastic look).
 • Motion: hand motion ≤3° rotation per second; no jitter.