Google Vids Personal Avatar Tutorial: Record, Prompt, and Fix It

AI Overview

How do you create a personal avatar in Google Vids?

Open Google Vids on a computer, choose Personal avatar, scan the QR code with a phone or tablet, and complete the guided face-and-voice recording. Then enter a scene prompt and generate your first clip.

Who can use Google Vids personal avatars?

You must be at least 18, use a Google Account with an eligible Workspace subscription, and be in a supported region. Personal avatars are currently English-only and unavailable in the EEA, Switzerland, and the UK.

Why is Personal avatar missing in Google Vids?

The usual causes are an ineligible plan, an admin-disabled feature, an unsupported region, an under-18 account, or rollout timing. Personal Google accounts may offer preset AI avatars without providing personal-avatar capture.

Can you retake or delete a Google Vids personal avatar?

Yes. In AI Video, open Create, choose Avatar, select More on your personal avatar, and choose Retake or Manage. Deleting the recording in your Google Account can also affect services that use the same avatar data.

Check eligibility before recording

Run the sixty-second access gate

Before adjusting a camera or writing a script, confirm that the Google Vids personal avatar feature should exist for your account. You need an eligible Google Workspace subscription, a Google Account, a phone or tablet for capture, and an account holder who is at least 18. The feature is currently limited to English and is not available in the European Economic Area, Switzerland, or the United Kingdom.

Personal avatars, preset AI avatars, and custom avatars are not interchangeable labels. A personal avatar is created from the account owner’s face and voice recording and can be managed through the Google Account. A preset avatar is a supplied presenter that does not reproduce you. Google’s current availability page says personal Google accounts only receive preset AI avatars, so seeing an avatar menu does not prove that personal capture is enabled.

For a managed Workspace account, ask whether the administrator has enabled personal avatars. Also verify that Vids is using the eligible work profile rather than a personal profile in the same browser.

This workflow is different from converting a presentation into scenes. If a deck is your source material, use the Google Vids Slides-to-video guide before deciding whether an avatar should present it.

Illustrative generated output: a useful avatar clip needs a believable scene, action, and object—not just a talking head on an empty background.

Record a personal avatar correctly

Prepare the room, phone, and face

On a computer, open Google Vids and choose Personal avatar from the start menu. Scan the displayed QR code with a phone or tablet, then follow the guided prompts to record your face and voice. The account owner must complete this process personally.

Put the phone at eye level instead of holding it below the chin. Use even illumination that is neither dim nor blown out. Your eyes, nose, and mouth must remain visible; ordinary glasses are acceptable, but sunglasses, masks, and hats are not. Record in a quiet room with no competing voices. Remove people, portraits, posters with faces, and reflective screens from the background.

Treat the capture as identity calibration, not a performance. Follow each movement at a comfortable pace, return your head to center, and speak naturally. A moving camera or busy set introduces avoidable variation.

Recording reference: eye-level framing, visible features, a quiet background, and soft directional light create a clean identity sample.

After capture, click Next. Your personal avatar should appear in the side panel as an available character for an AI video clip. If it does not, do not repeat the recording blindly. Recheck the account, plan, age, region, language, and admin gate first.

Generate the first avatar clip

Start with one measurable action

In an open Vid, select AI Video → Create → Avatar, then choose your personal avatar. Add a prompt that describes what the presenter should do, how the presenter should speak, and what the scene should sound like. You may also add a script, choose portrait or landscape, and include visual ingredients. Select Generate and wait for the result before inserting it into a scene.

Make the first test short and diagnostic: one location, object, composition, gesture, and two sentences. A one-person product explanation reveals more than “make a professional marketing video” because likeness, mouth motion, hand contact, voice, and object geometry can be judged together.

Use this prompt pattern:

Medium three-quarter shot of the presenter at a sunlit wooden workbench. She holds a small black portable speaker in her left hand, points once to the top control with her right index finger, and speaks two calm sentences. Natural workshop ambience, locked camera, continuous action, realistic hand contact, no background people.

Illustrative prompt target: the object, hand contact, gesture, eyeline, setting, and camera behavior are all observable at review time.

When the clip is ready, Google Vids offers actions including Insert, Remove, Extend, Edit, and Recreate. Insert only after the base clip passes review. Extend a good action when more time is needed; edit a correct concept with a local issue; recreate when the staging itself is wrong.

Write prompts that direct a real scene

Separate identity from performance

The avatar already represents who appears. The prompt should control what happens. Name the shot size, location, physical action, delivery, camera behavior, ambience, and ending state. Avoid stacking unrelated actions such as walking, picking up a product, turning to a screen, and delivering a long paragraph in one short clip.

Write scripts for speech, not for reading. Use short sentences, pronounceable product names, and deliberate pauses. If the exact wording matters, lock the script before changing motion. If identity matters more, keep the head angle moderate and avoid long occlusions. For multi-clip work, repeat the same wardrobe, setting language, key object, and camera vocabulary across prompts.

The short clip below is an independent Seedance output with visible speaker motion and audio. It is not a Google Vids result or a model comparison; use it as a transferable approval example for checking whether face movement, delivery, and the complete sound path feel coherent.

Play a finished speaker-motion and audio sample

Actual Seedance output: review a speaking clip in motion and with sound; a flattering still frame cannot prove lip timing or vocal continuity.

For a dedicated voice review method, follow the Seedance 2.5 voice-consistency guide. If the plan requires original establishing shots or cutaways rather than an avatar presenter, the image-to-video workflow can supply those visual beats.

Fix missing, rejected, or inconsistent avatars

Diagnose the gate before the generation

If Personal avatar is missing, verify eligibility in this order: signed-in account, Workspace plan, age, region, English availability, and admin setting. Clear evidence at each gate is more useful than repeatedly refreshing Vids. Feature rollouts and account entitlements can change, so use the current Google availability page rather than relying on a screenshot from another user.

If capture is rejected, improve the recording conditions: eye-level phone, face fully visible, moderate light, silent room, and a background without another face. Remove headwear and sunglasses. Repeat the guided movement exactly, returning to center after each turn. Do not ask a colleague to complete the identity capture for the account owner.

If the avatar exists but the clip is weak, classify the failure before rerunning:

• Likeness drift: reduce extreme angles, motion, and occlusion; use a simpler medium shot.
• Mouth or voice mismatch: shorten the script and remove difficult names or rushed phrasing.
• Hand or object error: request one contact action and keep the object unobstructed.
• Scene drift: repeat the location, wardrobe, prop, lens, and lighting anchors.
• Unclear result: review the moving clip with audio, not a thumbnail.

When sound generation is the actual blocker rather than avatar capture, use the Veo audio-generation troubleshooting guide to isolate prompt, speech, ambience, and retry decisions.

Manage privacy and scale the workflow

Retake, delete, and document approval

To replace the source recording, open AI Video → Create → Avatar, select More on your personal avatar, and choose Retake. Choose Manage to reach the Google Account controls for the avatar. Google notes that deleting personal-avatar data can affect other services you have allowed to use it, including the Gemini app, so check reuse before deletion.

Avatar feedback may be readable by human reviewers. Do not put personal, confidential, or sensitive information into feedback. For team production, document who owns the avatar, where it may appear, the approved script, the approved scenes, the review date, and the person authorized to request deletion or a retake.

Use a repeatable approval rubric: recognizable identity, natural face motion, intelligible voice, correct words, plausible hand contact, stable objects, clean start and end, and no unsupported claims. Save the accepted prompt with the clip so later edits remain controlled variations.

Illustrative approved end frame: identity, wardrobe, object, light, and setting remain readable through the closing action.

For campaigns with several avatar clips plus B-roll, the hard problem becomes orchestration. A Google Flow and Veo scene workflow helps separate generative shots from editorial assembly. Seedance Agent can then organize references, shot briefs, generated clips, feedback, and targeted reruns so the team approves the whole sequence rather than losing decisions across separate tools.

Conclusion

A reliable Google Vids personal avatar starts with eligibility, not a prompt: confirm the plan, account, age, region, language, and admin setting; record at eye level in a quiet, evenly lit room; generate one measurable action; and approve likeness, speech, hands, objects, sound, and continuity in the moving clip. Retake or manage the account recording deliberately, especially when other Google services may use it. When the avatar belongs inside a larger production with references, B-roll, review notes, and selective reruns, start the project with Seedance Agent.