Guide 1: Mastering Cinematic AI Video with Veo 3.1 and Sora 2 Prompts
Generating cinematic results from AI video tools requires moving beyond vague descriptions and adopting a structured approach known as generative directing. By accessing premium models like Veo 3.1 and Sora 2 Pro through Syndicate AI, you gain the control needed to produce professional-grade content.
The Core Prompt Formula
A well-crafted prompt acts as a director’s script, reducing guesswork and enhancing output accuracy. Structure your prompt using the following components: Prompt = [Cinematography] + [Subject] + [Action] + [Context] + [Style]
| Component | Description | Example Detail |
|---|---|---|
| Cinematography | Specifies technical camera details. | "Anamorphic 2.0x lens," "wide-angle perspective," "tracking shot at shoulder level". |
| Subject | Detailed description of the focus (person, object, animal). | "A person jogging through a park at sunrise," detailing appearance, clothing, or posture. |
| Action | What the subject is doing; the core storyline. | "Maintaining steady pace with natural arm movement," or "He charges into battle". |
| Context | Setting, environment, and atmosphere. | "Abandoned medieval castle surrounded by dense fog," or "futuristic city skyline glowing with neon lights". |
| Style | Visual aesthetic and quality. | "Cinematic clarity," "photorealistic," "filmic cadence," or "bold outlines, and vibrant colors". |
Advanced Techniques for Veo 3.1 (Creative Control)
Veo 3.1 is specifically built to respond to explicit cinematic instructions and is available for developers and creators through the Gemini API, Flow, and Vertex AI.
- Director-Driven Structure: Use a top-down format, specifying the aspect ratio and total duration, followed by a concise 2–4 clip shot list. Script short 4–6 second beats for procedures and assemble sequences in your editor.
- Native Audio Directing: Veo 3.1 supports cues for dialogue, ambient noise, and SFX. Proactively prompt for the desired soundscape or silence to prevent undesirable sound "hallucinations".
- Ensure Consistency: Maintain brand consistency by providing reference images for elements like logos, uniforms, or locations. Veo 3.1 offers the Ingredients to Video feature, allowing you to use up to three reference images (e.g., two characters and a location) blended with a text prompt to maintain consistency.
- Negative Prompting: Veo 3.1 explicitly supports a dedicated Negative prompt line. Use this powerful mechanism to specify undesirable traits or artifacts to be excluded from the final output.
- Multi-Minute Narratives: While individual clips are conditional (e.g., 1080p resolution is project-dependent and frame rate is 24 fps in the Gemini API), multi-minute continuity can be achieved through chaining clips and scripting narrative consistency.
Advanced Techniques for Sora 2 (Physical Realism)
Sora 2 operates as a reasoning model that understands physics, making detailed instruction and temporal logic critical for success.
- Sequence Control: Segment your prompt by shots or beats to control the temporal sequence. Delineate distinct actions clearly in the prompt text, such as: Shot 1: [description]... Shot 2: [description].
- Character Persistence (Cameo): Use the Cameo feature to insert a user’s likeness or generated characters, which is key to maintaining consistent realism across multiple shots. This is crucial for narrative projects like branded training videos or serialized content.
- Dialogue Formatting: Dialogue must be described directly in a separate block in your prompt so Sora 2 clearly distinguishes spoken lines from visual descriptions. Keep lines concise and label speakers consistently.
- Refinement: Use the model’s iteration features, like Remix, rather than full regeneration, to maintain the latent consistency of the initial result while adjusting the prompt.