🌌 How to use Sora 2 like a pro
Sora 2 is OpenAI's flagship video model — the one that renders realistic water, cloth, fire and crowds better than anything else on the market. Here's how to prompt it right.
What Sora 2 is best at
Sora 2 is the only model that reliably nails complex physics: liquid dynamics, cloth simulation, fire, smoke, crowd behaviour and reflective materials. Use it when the physicality of the shot IS the shot — hero moments, not filler b-roll.
The Sora prompt formula
Sora rewards long, cinematic briefs. Structure your prompt as: scene → subject and action → camera move → lens and framing → lighting and palette → mood reference. Skip any layer and Sora fills it in with average choices.
- Scene — where we are, time of day, weather
- Subject and action — what happens beat by beat
- Camera — orbit, dolly-in, crane-up, handheld tracking
- Lens and framing — 35mm anamorphic wide, 85mm close-up
- Light and palette — golden hour, teal-orange grade
When to use Sora Turbo vs Sora Pro
Sora Turbo is the daily driver — faster and cheaper for iteration. Sora Pro is the hero render for the shot that ships. Lock composition on Turbo, then re-run the winning prompt on Pro at full resolution.
When to pick something else
For synced dialogue and 4K narrative shots, Veo 3.1 is stronger. For animating a photo, Kling 2.5 is faster and cheaper. Reserve Sora for the moments where physics or realism is the point.
Ready to put this into practice?
Open the studio and apply what you just learned in under a minute.
Open Sora 2 in the studio