How to use the Gemini Live camera for visual help

To use the Gemini Live camera, open a Live conversation in the Gemini mobile app and tap Camera to share what your phone sees. Tap Camera again to stop; Google’s Live instructions↗ (opens in a new tab) also describe the starting control as Turn on video streaming. Show one object or scene, then ask a specific question about it. fenbs can keep the observations you checked and the next step, so you do not have to repeat the whole demonstration later.

How do you start and stop the camera?

1. Open the Gemini mobile app and choose Live, or swipe left to enter Live.
2. In the active Live conversation, tap Camera to start sharing. Google’s introductory instructions also use Turn on video streaming for the start control.
3. Review any device request for camera access before proceeding. Include only the scene you intend to discuss.
4. Use Switch to rear or front camera when you need the other view.
5. Ask one question about the visible object. When you have shown enough, tap Camera to stop sharing.

Begin with an object you can inspect yourself, such as a public brochure or a sample of packaging. That lets you check whether the answer matches what is in front of you. Ask permission before including another person, as Google’s guidance instructs. Move unrelated papers and private details out of view before you start, rather than trying to correct an accidental disclosure afterward.

Which device and account should you use?

Google documents camera sharing in mobile Live for Android and for iPhone or iPad. Its iOS Live guide↗ (opens in a new tab) uses the same Camera control for starting and stopping sharing. These instructions do not establish camera access in an ordinary desktop Gemini web chat. Use the mobile experience for the workflow described here.

Google names signed-in personal accounts and eligible work or school accounts in the Live requirements. The mobile availability page↗ (opens in a new tab) includes the US. The cited Live guidance does not name a separate camera subscription requirement, but account eligibility still matters. Check current requirements for your own account, especially when your school or employer manages it.

If you are still choosing which Google tool fits the job, start with the separate Google AI tools overview. Camera sharing solves the narrow problem of discussing a physical view. It is not a substitute for choosing a research or document workflow that fits the material.

How do you make the visual question useful?

Say what you are showing and what you need to decide. “What do you think?” produces an open-ended discussion. “I am comparing two draft package labels; tell me which wording is easier to scan and explain why” gives you a concrete result to review. Make clear whether you want a description, a comparison, or a list of questions to investigate.

Keep one subject in view long enough to discuss it. If you move to a different object, say so. Read any important small text yourself and compare it with the answer. These simple checks help separate something actually visible from an assumption that entered the conversation.

• For an explanation: “Describe what is visible first, then explain the part I point out.”
• For a comparison: “Use these three criteria and tell me what you cannot judge from this view.”
• For a review: “Separate observations from suggestions, and do not treat a suggestion as my decision.”
• For a finish: “Summarize what we checked and the next step that still needs a person.”

Treat those as suggested questions, not special camera commands. When the task needs source-based research rather than a live view, the separate ChatGPT versus Gemini research guide covers that choice. A camera conversation is only one part of deciding what evidence you need.

What happens when you pause or leave the app?

Google’s Live guidance says camera sharing stops when you put Live on hold, leave the Gemini app, or lock the device. It resumes automatically when you continue a held conversation. Simply returning to the app or unlocking the device does not, by itself, restart the camera. Check the sharing state before assuming Gemini can see the scene again.

That distinction matters during a practical review. If you set the phone down, open another app, and come back with a different sample, explain the change and restart the camera as needed. Do not continue asking about “this one” without confirming which object you mean. A clear spoken description helps you keep track of the work even when the view changes.

If you need help with something visible on the phone rather than in front of its lens, use the separate Gemini Live screen-sharing guide. That shares a different input and has its own stopping behavior.

What should you check when visual help is wrong?

First check the input: is the intended camera active, is the relevant object visible, and has sharing stopped because you left or locked the phone? Then ask Gemini to describe the scene before interpreting it. If the description does not match the object, correct the misunderstanding before building another question on top of it.

If camera access is the obstacle, review your device’s permission request and camera access controls. Follow the controls on your own device rather than expecting Android and iOS to use the same system dialogs. Do not grant unrelated access simply to try to make a visual question work.

Finally, check the conclusion yourself. A useful suggestion about a sample does not prove that a measurement, material, or instruction is correct. Refer to the original label or other appropriate source before acting. Keep anything unresolved as a question to check, especially if the answer would affect a physical change.

What should you save after showing the scene?

Save the meaning of the session in words: what you showed, what question you asked, what you checked, and what remains uncertain. For example, a packaging review might end with “The small ingredient text needs a closer review; compare the printed proof before approving it.” That is a hypothetical next step, not evidence that a visual review already happened.

Keep the relevant reference with that note when you are allowed to retain it. Avoid vague phrases such as “use the second one” that depend on a vanished view. This makes the result useful when you move into a different AI; transferring a Claude chat to Gemini covers a separate conversation-transfer workflow, while Claude versus Gemini Canvas covers working on an editable artifact.

How to keep visual findings with fenbs

1. Create a free account.
2. Write what you showed, what you learned, and what you checked. Keep guesses clearly marked as guesses.
3. Add the next action as a task, with enough detail that someone can do it without seeing the camera session.
4. Connect Gemini to fenbs. Use a personal account that meets the connection requirements.
5. Before stopping, save where the work stands so you or another connected AI can carry on from the same place on any device.

Google’s custom-app documentation↗ (opens in a new tab) explains the US personal-account restrictions for that connection. Adding it does not guarantee it can be used inside Live. fenbs keeps your checked findings and next steps; a quick description that creates no ongoing work may not need a separate record. For more practical examples, browse AI at work.