How to prompt cinematic AI video in Veo4 Studio
Strong prompts say what is on screen and how it should feel—camera distance, motion, light, and sound. Veo4 Studio’s controls match that pipeline: choose **Text to Video** (prompt only), **Image to Video** (start frame required, end frame optional), or **Reference to Video** (up to three reference images); set **Aspect Ratio** to **Auto**, **16:9**, or **9:16** (Reference mode offers **16:9** and **9:16** only—no Auto); pick **Fast** or **Quality** (**Quality** is available for Text and Image only); and select **720p**, **1080p**, or **4k** resolution. Below is how to write prompts that line up with those settings—especially for Text mode—so you waste fewer credits on mismatched expectations.
Build prompts in layers: subject, setting, action
Start with who or what is on screen, where the scene takes place, and what happens over a few seconds. One clear sentence for each layer reduces ambiguity. For example: “A lone cyclist on a coastal road at golden hour, riding toward camera as waves crash below.”
Add camera and lens language
Words like wide shot, slow dolly in, handheld, aerial drone, shallow depth of field, or rack focus give the model concrete visual intent. You do not need real gear names—plain language works as long as it describes framing and movement.
Describe light and mood
Specify time of day, weather, and palette: soft overcast, neon reflections in rain, warm tungsten interior, high-contrast silhouette. Mood words—tense, playful, documentary, dreamy—steer tone when paired with visual cues.
Mention sound when you care about audio
Veo 3.1-class generation can include synchronized audio—call out ambient sound, dialogue style, or music vibe in the same prompt: distant thunder, quiet café chatter, lo-fi beat under dialogue. That aligns picture and sound instead of leaving audio random.
Match generation type to your assets
**Text to Video:** the prompt carries the whole brief—use the layering habits above. **Image to Video:** upload a **Start Frame** (required) and optionally an **End Frame**; describe how motion, lighting, and subject should evolve between them. **Reference to Video:** add up to three reference images and spell out what must stay consistent (look, character, palette) versus what should move or change. In the UI, **Quality** mode unlocks only for **Text** and **Image**; **Reference** stays on **Fast**, matching the “Text/Image only” hint next to Quality.
Set aspect ratio, mode, and resolution
**Auto** lets the model pick framing; choose **16:9** for landscape YouTube-style shots or **9:16** for Shorts/Reels. **Reference to Video** shows **16:9** and **9:16** only—no Auto. There is no **1:1** export. **Fast** uses fewer credits per run; **Quality** costs more and suits final polish when Text or Image is selected. **720p**, **1080p**, and **4k** set output sharpness—pair higher resolution with prompts that reward detail (textures, faces, wide shots).
Iterate instead of stuffing one mega-prompt
Long prompts are fine if structured. If results drift, remove conflicting adjectives, change one control at a time (lighting only, or resolution step only), and reuse a winning prompt prefix across runs.
Related guides
- Vertical video for Shorts & Reels (9:16)
Frame for mobile-first viewers: safe margins, hooks, and motion that reads on a narrow screen.
- Common AI video prompt mistakes
Fix contradictory wording, vague subjects, and overloaded prompts before you burn credits.
Generate video from your prompts
Use Veo4 Studio to turn structured prompts into HD clips with realistic audio.