Google Gemini Image Generation: Can It Make People Again?

Yes—Google Gemini can generate images of people again, but availability depends on the image-generation feature, the account, and the kind of image requested. Google restored the ability to create images depicting people after pausing it in 2024, while continuing to restrict some prompts and applying safeguards intended to reduce harmful or misleading results.

What changed after the 2024 pause?

In February 2024, Google temporarily stopped Gemini’s image-generation feature from producing images of people. The pause followed criticism of historically inaccurate outputs, including depictions of people from historical settings and groups that did not fit the prompt. The problem was not that Gemini could never depict people; it was that the system sometimes applied its effort to produce diverse results in ways that made particular historical requests inaccurate.

Google apologized and said it was working on a revised version. Image generation involving people later returned, with Google describing the feature as improved and rolling it out over time. This history explains why older articles or posts may say that Gemini cannot make human images. Those reports describe the temporary restriction, not the feature’s present general capability.

What Gemini can generate now

When image generation is available in a user’s Gemini experience, a prompt can request a person or people in a scene. For example, a user might ask for “a fictional bookstore owner arranging books in a bright neighborhood shop” or “a watercolor portrait of an imaginary traveler beside a mountain lake.” The result is generated artwork, not a verified photograph or evidence that a real person was present.

How well a prompt works depends on its detail and the system’s safeguards. A request specifying the subject, setting, style, and composition is usually clearer than a short instruction such as “make a person.” Still, specific wording cannot guarantee a particular face, accurate anatomy, or a historically faithful result. Image models can make visual mistakes, including distorted hands, inconsistent details, or inaccurate clothing and objects.

• Fictional people: Often suitable for concept art, illustrations, or imaginary scenes.
• General portraits: The tool may produce an invented person in a requested artistic style.
• Specific real people: Results may be restricted, especially where a prompt could enable impersonation, deception, or abuse.
• Historical scenes: Generated depictions can be inaccurate; check important details against reliable sources.

How to try it

Use Gemini’s image-generation option if it appears in your account, and describe the image you want in ordinary language. Feature names and controls can change, so follow the options actually shown in the app or website rather than relying on a particular button name from an older guide.

1. Open Gemini while signed in to the Google account you intend to use.
2. Look for an image creation or image generation option, if available.
3. Write a prompt describing the subject, setting, visual style, and any important details. For instance: “Create a gentle editorial illustration of a fictional gardener watering plants on a city balcony at sunrise.”
4. Review the result for visual errors and whether it follows the request. If needed, ask for a targeted change, such as a different background or a wider composition.
5. If the request is refused, simplify it or choose a clearly fictional, non-deceptive scene. Do not try to disguise a prohibited request with different wording.

If you do not see an image option, that does not necessarily mean image generation has been removed for everyone. Features can vary by region, account type, age eligibility, language, device, and staged rollout. Check that you are using the current Gemini experience and consult Google’s current product information for availability in your location.

Restrictions, safety, and responsible use

Being able to generate people does not mean every image request is allowed. Google applies policies and technical safeguards to limit harmful, deceptive, or otherwise disallowed content. The system may refuse a prompt or produce a result that differs from the request. Exact restrictions can evolve, so a refusal should not be interpreted as proof that the entire feature is unavailable.

Be especially cautious with images that could make someone appear to have done or said something they did not. Do not present generated images as authentic photographs, news evidence, or records of real events. Avoid using someone’s likeness in a way that could mislead, harass, or violate their privacy. If an image concerns a real individual or a sensitive situation, consider whether creating or sharing it is appropriate, even if the tool permits the prompt.

Generated images may include identifying marks or other provenance information, depending on the product and output. Such signals do not make every image self-explanatory, and their presence or behavior may change. A viewer should not assume that an image is genuine—or that it is generated—based only on appearance.

Common problems and what they mean

• No image-generation option: Availability may differ for your account or location, or the feature may not have reached your account yet. Check the current product information and try again later.
• A refusal: The wording may conflict with a safeguard, or the requested depiction may be restricted. Try a different, clearly fictional and benign concept rather than attempting to bypass the safeguard.
• The person looks wrong: Image generation is probabilistic, not a precise portrait or identity-matching system. Refine the style and scene, or regenerate, but expect variation.
• Historical details are inaccurate: Treat the image as an illustration, not a reference source. Verify clothing, architecture, and other factual details independently.
• The feature behaves differently from a guide: Instructions may have been written for a previous interface or rollout. Use the controls currently displayed in Gemini.

Frequently Asked Questions

Can Gemini generate pictures of real people?
It may support some depictions involving people, but requests involving identifiable real individuals can be limited by safeguards. Do not assume it can create any real person in any situation.

Why did Google stop Gemini from making people images?
Google paused the feature in February 2024 after inaccurate image results, notably in some historical contexts, drew criticism. It later brought back image generation involving people with changes to the system.

Can Gemini make a photo of me?
Availability and results depend on the feature and the request. Do not expect an exact likeness from a text prompt; follow the product’s current rules and avoid uploading or sharing personal images without considering privacy.

Why does Gemini refuse my image prompt?
A safeguard may apply to the subject or context, or the request may be too ambiguous. A refusal does not necessarily mean Gemini cannot generate people at all.

Are Gemini-generated people images real photographs?
No. They are generated images and should not be treated as evidence of a real person, event, or historical appearance.

The practical answer

Gemini’s people-image capability has returned after its 2024 pause, but access and permitted prompts are not universal. Try a harmless fictional scene, check the controls available in your account, and treat the output as generated artwork that may contain inaccuracies—not as a dependable photograph or historical record.

Related

• Python Machine Learning Libraries: How to Choose
• AI for Network Operations: A Practical Adoption Guide
• How to Tell When an AI Chatbot Is Making Things Up