Veo 3.1 Guide: Cinematic 8-Second AI Video Up to 4K

Use Veo 3.1 when a single shot needs high-fidelity detail or a 4K deliverable. In the Spicy Studio Video Lab it turns a prompt or a start image into one 8-second clip, and it's the only video model there with a 4K setting.

It's a good fit for shots carried by place, light and camera movement: - establishing shots of cities, coastlines and mountains - building exteriors and interiors for property or architecture - product hero shots and slow reveals - B-roll to cut between talking segments

For a person's performance or several actions in one clip, a Kling model usually fits better.

Prompt: the scene, action and camera. On its own, it gives you text-to-video.

Start image (optional): composition and product details come from your picture, not the model's guess.

End frame (optional): used with a start image, it gives the clip a destination.

Aspect ratio: 16:9 or 9:16 (vertical).

Length: always 8 seconds.

Price per clip: 80 credits at 720p (10 a second), 110 at 1080p and 200 at 4K.

The 25 free credits on a new account don't reach one clip. The €2 Spark top-up (90 credits that never expire) covers a 720p test, and the €10 Spicy Basic plan's 600 monthly credits cover seven 720p clips or three in 4K (pricing).

The Video Lab has no sound setting for Veo 3.1. For sound generated with the picture, use Seedance 2.5 (audio on by default), or add music and effects in your editor.

Veo 3.1 is a Google DeepMind model, announced on October 15, 2025 (Google's announcement). Google offers it in the Gemini app and Flow, and to developers through the Gemini API (Veo page). Google's own tools add extras the Video Lab doesn't offer, such as extending a clip and guiding it with several reference images.

For developers, Google charges per second: as of October 2026, its Gemini API pricing page lists Veo 3.1 at $0.40 a second at 720p or 1080p and $0.60 at 4K, with a cheaper Veo 3.1 Fast tier and no free tier. That route needs an API key and code; in Spicy Studio you work in the browser from one credit balance.

Eight seconds holds one shot: the place and subject first, then one camera move, then light and lens. Several moves would each get too little time.

Camera moves: push-in, pull-back, crane up, low-angle tracking shot, slow orbit, locked-off tripod shot.

Light and look: blue hour, overcast daylight, neon reflections on wet pavement, 35mm film grain.

Landscape, 16:9: “A fjord at blue hour, fog lifting off still water as lights come on in a small harbor village. The camera cranes up slowly from the shoreline to reveal the mountains behind. Cool tones, soft haze.”

Interior, 16:9: “A sunlit mid-century living room with oak floors and tall windows. The camera glides from the doorway toward the windows while sheer curtains move in a light breeze. Late-afternoon light, no people.”

Product, 9:16: “A matte black wristwatch on wet slate, raindrops bouncing around it. Low-angle macro shot; the camera orbits a quarter turn to the left. Rim light from behind, dark background.”

When the frame must match a brand, a real product or a storyboard, don't leave the composition to Veo: make the start image first.

1. Create it in the Image Lab in the clip's 16:9 or 9:16 shape, with Nano Banana 2.1 (12 credits at 2K) or Spicy Imagine Pro (8 credits at 2K).

2. Attach it as the start image and describe only what should move: “Steam rises from the cup, the camera pushes in slowly, morning light shifts across the table.”

3. For a transition, add an end frame: the same street by day and by night, or a closed gift box and then the open one. Veo fills the 8 seconds between them.

More on preparing stills: image-to-video tutorial.

Each run costs the full clip price, so test at 720p (80 credits) until motion and framing work, then finish one of two ways.

Re-render at 4K (200 credits) for native 4K. A new run won't repeat the draft exactly.

Upscale the approved clip with Video Upscale to 1080p, 2K or 4K for 16, 28 or 56 credits. It keeps the exact motion you signed off on (how upscaling works).

A person or character in motion: Kling 3.0 animates your start frame for 5 or 10 seconds with controlled movement, from 60 credits (Kling 3.0 guide).

Several actions in one clip: Kling Omni O3 chains up to five prompt segments within 3–15 seconds (Kling Omni O3 guide).

Longer takes, sound or references: Seedance 2.5 runs 4–30 seconds with native audio (Seedance 2.5 guide), and Seedance 2.0 costs less for clips of up to 15 seconds at 720p.

Every model side by side: AI video generators compared.