Text-to-video workflow

Sora2 and Sora3 text to video generator

Create Sora2 text-to-video clips now and prepare Sora3 prompts with duration, aspect ratio, and production workflow guidance.

Last updated: 2026-06-16

Write prompts as production notes

The generator works best when the prompt reads like a shot note instead of a keyword list. State the subject first, then add movement, camera direction, lighting, and pacing.

  • Subject: who or what appears in the shot.
  • Motion: what changes during the clip.
  • Camera: pan, push-in, orbit, handheld, tripod, or macro framing.
  • Style: realistic, cinematic, product demo, social ad, or editorial.

Choose duration before adding detail

Short 4-second clips should use one clean motion. Eight-second clips can carry a reveal or camera move. Twelve-second clips need a clearer beginning, middle, and end so the scene does not drift.

Keep iteration measurable

Save one baseline prompt, then change one variable at a time. This makes it easier to compare camera language, lighting instructions, and duration choices across Sora2 tests and Sora3-ready prompts.

Related Sora guides

FAQ

What should a Sora2 or Sora3 text-to-video prompt include?

A good prompt includes the subject, setting, motion, camera direction, lighting, visual style, and duration. Keep it specific enough to guide the shot without adding conflicting instructions.

Is text-to-video better than image-to-video?

Text-to-video is better when you need fast concept exploration. Image-to-video is better when composition, character design, or product framing must stay close to a reference image.