Describe subject, a second-by-second action timeline, camera, environment, and audio in one block, with negative constraints, so the shot is controllable instead of a lucky roll.
プロンプト
Generate one continuous 8-second cinematic shot. Subject: [SUBJECT]. Action timeline: 0-2s [ACTION], 2-6s [ACTION], 6-8s [ACTION]. Camera: [LENS + MOVEMENT]. Environment: [PLACE + WEATHER]. Audio: [AMBIENCE]. Keep identity, wardrobe, and lighting consistent. No cuts, subtitles, watermarks, or extra limbs.
変数を置き換える
[SUBJECT]The main subject.
[LENS + MOVEMENT]e.g. 35mm slow dolly-in.
どのモデルを使うか
- Veo 3.1 — Video generation is a Veo capability, not a text-model one.
入力例
- [SUBJECT] = "a barista finishing a latte". Camera = "slow push-in".
出力例
私たちは実際のモデル実行による出力例のみを公開しています。このテンプレートはまだ正式にテストされていないため、出力は表示されません。上記の入力を使ってご自身で実行してみてください。
この構造が機能する理由
- A timeline gives the model temporal control instead of one static idea.
- Negative constraints suppress the usual video artefacts.
よくある失敗と修正
Motion is chaotic.
修正: Simplify to one clear action per time band.
既知の制限事項
- Video output is non-deterministic; expect several tries per usable shot.