Google Veo 3.1video workflow

Google Veo 3.1 video trials: modes, shot prompts, and output checks

Distinguish text-to-video, image-to-video, and reference modes in the Veo 3.1 entry. Prepare shots, inspect motion, and compare Seedance or Kling with the same brief.

Sep 3, 2026 · Updated Oct 20, 2026

Illustration for video shot planning and motion checks

The on-site Google Veo product entry currently displays Google Veo 3.1 and retains its legacy URL. This guide covers mode selection and trial checks, without assigning Veo, Seedance, and Kling a fixed quality or speed ranking.

Choose the input mode first

Mode	Prepare	Inspect
Text-to-video	A clear shot description	Subject, action, and setting
Image-to-video	An image and motion instructions	Subject shape, details, and movement
Reference-to-video	Reference images and text required by the form	Preserved reference features and continuity

The current catalog supplies callable, priced modes with forms for these three tasks. Other preview or advanced catalog labels do not automatically mean a selectable workbench option. Check the current form for resolution, aspect ratio, count, duration, and image requirements.

Three steps to prepare and check a trial

Define one shot: subject, action, camera direction, lighting, and details to preserve.
Choose a matching mode, check the actual duration options and cost, and submit.
Inspect subject stability, camera path, background continuity, and unwanted elements. Change one factor per revision.

Shot prompts to adapt

The original examples below are writing exercises. Seconds and sharp-logo requests express creative intent, not guaranteed duration choices or results. Adjust them to the current form. Veo 3 in an older example refers to the series; the current entry is identified above.

Glass skincare bottle on light marble, rim light from the side, slow push-in until the logo is sharp, shallow depth of field, cinematic grade, stable camera, about 4–5 seconds, no subtitles or watermarks, no extra fingers or warped bottle.

Dusk city rooftop, light wind on fabric, distant lights warming up, camera slowly tracks sideways, subject holds a stable stance, natural light plus neon bounce, film framing, about 5 seconds, no dialogue captions.

Write one 5-second Veo 3 prompt: subject is a smartwatch, camera pushes from a 45° high angle to a dial close-up, soft studio light, emphasize metal texture and screen highlight; include don’ts: no warped strap, no garbled UI, no watermark.

White-backdrop studio smartwatch at 45°, sharp logo, soft light, commercial hero shot—lock framing in Nano Banana or GPT Image 2, then drive motion via image-to-video.

Three working recipes

Recipe A: check one shot

Turn a real need into an executable brief, generate a trial, and inspect subject and motion against your criteria. Record the result. The account link below can help locate past work where the current interface supports it.

Recipe B: compare with the same brief

Where inputs and settings allow, try the same shot requirement with Seedance, Kling, or Veo. Compare deformation, camera deviation, background changes, and actual cost. Do not assume one model is always faster or more cinematic.

Recipe C: stills, video, and music

Confirm a still with Midjourney Image, GPT Image, or Nano Banana, then select a suitable video mode. For music, prepare a no-vocals Suno brief and check the track against narration in the edit.

Common issues

Only adjectives such as cinematic: specify a push-in, track, or locked-off camera and the lighting.
Warped subject or packaging: check the input image and motion range, then revise one factor.
Missing duration or parameter: use the current form rather than inferring options from an example.
Unclear model choice: compare against the same checklist instead of applying unverified rankings.

Back to blog All Models