This guide uses a fictional analytics product and a recorded sample. It is not customer work, a repeatability claim, or a live public generation. Authenticated requests require a current Veo catalog entry, entitlement, credits, spend controls, provider availability, and enabled billing mode.
What you’ll learn
- Which settings to lock for the fast setup pass shown in the screenshots
- How to write a motion prompt instead of a still-image prompt
- How to narrate the in-progress state without leaving awkward silence during a longer render
- How to review and export the base clip without implying that merged voiceover is available
Before you start
- Open Generate
- Switch to Video mode
- Keep the setup pass shown in this walkthrough on
4sand720p - Use one short prompt that is easy to evaluate on screen
Open Video mode and start with a fast 4-second 720p setup pass
Click Video in the mode selector before typing anything. Then lock the first run to the same settings shown in the screenshot.
- Set
Durationto4s. - Set
Qualityto720p. - Choose
Veo 3.1 Fast. - Stay on
Generalfor this setup pass. - Use this faster path to validate motion direction before you spend more time on a longer export.
Expected result
The request is configured for a faster approval pass before the team commits to a longer or higher-fidelity export.
Use a prompt that names the scene, camera move, and mood
Paste this exact prompt into the textarea:
Create a 4-second product reveal for a glowing analytics interface. Slow dolly-in camera move, dark navy studio lighting, cyan UI highlights, and smooth particle motion.- Name the subject first: the glowing analytics interface.
- Name the movement second: a slow dolly-in camera move.
- Name the mood third: dark, polished, and cinematic.
- Keep
Veo 3.1 Fastselected for this setup pass. - Leave
Apply brand guidelinesoff for this isolated video demo. - Click
Generate.
Expected result
The render reads like an intentional cinematic sequence instead of a short placeholder animation.
Use the in-progress preview to explain queue and render time
Once the request is submitted, keep talking. This is where good demos either build trust or make the workflow feel stuck.
- Explain that video jobs take longer than image jobs by design.
- Call out visible progress in the preview area instead of treating the wait as dead time.
- Tell the customer to wait for the full preview before deciding whether the clip is good enough.
- Do not restart the job if the preview is actively populating.
Expected result
The customer understands that the job is still processing and that the waiting state is expected for video generation.
Review the base clip before export
Wait for the base preview, play the full clip, and inspect it against the approved brief.
- Check whether the camera move and subject remain usable throughout the clip.
- Verify any visible interface, product, logo, or text details frame by frame.
- Confirm the exported duration, resolution, and crop against the request.
- Merged video voiceover is not in the production launch path. Generate standalone single-speaker audio separately if the workspace admits it.
Expected result
The team can inspect motion, text, product details, duration, and crop before deciding whether to keep the draft.
Sample generated clip
Recorded fictional sample for inspecting the review checklist; it does not promise the same result from another run.
Recommended rollout order
- Start with one short
4s,720psetup pass to validate the motion direction quickly - Teach the preview, waiting state, and review checklist before export
- Move to longer or higher-fidelity exports only after the team likes the base concept
What to read next
- Audio & Text-to-Speech for standalone voice generation
- Branding & Guidelines for image overlays and logos elsewhere in the widget
- Docs for the exact widget configuration and events behind the video workflow