Create Images in Google Bard: A Step-by-Step Guide

Google Bard was renamed Gemini, and its image-generation feature lets you create pictures from written prompts without using a separate design app. You describe the image you want, review the generated options, then refine or download a result; availability and controls can vary by account, language, and region.

What You’ll Need

A Google account and access to Gemini at gemini.google.com or through the Gemini app where available.
An internet connection and a browser or app that is up to date.
A clear idea of the image you want. You do not need to upload a reference image to generate one from text.

Step-by-Step

Open Gemini. Go to gemini.google.com and sign in to your Google account, or open the Gemini app. If you still search for Google Bard, use Gemini: Bard was the former name of the service. If the site asks you to choose an account or confirm access, follow its on-screen instructions.
Check that image generation is available. Start a new chat and look for the prompt box. You can usually request an image directly in the conversation, without turning on a separate image tool. If Gemini says it cannot create images, check whether the feature is supported for your account, location, language, and age settings. A work or school account may also have restrictions set by its administrator.
Write a specific prompt. State the subject, setting, visual style, composition, lighting, and intended use. For example: “Create a bright, flat illustration of a red bicycle leaning against a brick wall in a small city courtyard. Use a limited color palette, soft morning light, and a square composition.” Keep the request focused; you can add or change details in follow-up messages.
Ask Gemini to create the image. Enter your prompt and submit it. Use direct wording such as “Create an image of…” or “Generate an illustration showing…”. Gemini may take a short time to respond. Depending on the interface and account, it may provide one or more results or explain that it cannot fulfill the request.
Review the result. Check whether the main subject, setting, style, and framing match what you asked for. Look closely at details such as hands, faces, objects, signs, and lettering. Generated images can contain visual errors, and text inside an image may be misspelled or distorted.
Refine with a follow-up prompt. Tell Gemini what to change and what to keep. For example: “Keep the bicycle and courtyard, but make the wall blue and move the bicycle to the left.” Changing one or two things at a time makes it easier to see whether the edit worked. If you want a fresh interpretation instead, submit a revised full prompt.
Save an image you want to use. Open or select the result, then use the download control if it appears. On a computer, you can also use the browser’s image-saving option where permitted; on a phone, use the app’s share or download controls. Check where your device saves downloads, and confirm that the file opens before closing the chat.
Review the image before sharing or publishing. Make sure it does not include unwanted details, inaccurate text, or a person or logo you did not intend to depict. Follow the service’s current terms and any rules that apply to your use. Do not present a generated image as an authentic photograph of a real event when that could mislead people.

Common Mistakes

Using a vague prompt: “Make a nice picture” leaves the subject and style open to interpretation. Name the main subject and add a few useful visual details.
Putting too many conflicting instructions together: “Photorealistic watercolor, minimal and highly detailed” can point in different directions. Choose one main style and make the other requirements compatible.
Expecting exact lettering: Image models may render words incorrectly. For a poster or label that needs precise text, generate the background first and add the wording with a separate editing tool.
Assuming the result is factually correct: A convincing image can still depict objects, labels, or details inaccurately. Inspect it rather than treating it as evidence.
Assuming Bard is a separate current product: Google Bard is now Gemini. Look for Gemini’s current website or app, not an old Bard page or outdated instructions.
Repeating the same prompt when an edit is needed: If most of the image is right, specify the exact change and say which parts should remain unchanged.

Tips

Put the most important subject first, then describe the setting, style, lighting, and framing.
Use concrete visual terms such as “top-down view,” “warm sunset light,” or “simple geometric shapes” instead of several subjective adjectives.
For a consistent look, repeat the same style description across prompts, such as “editorial flat illustration with muted colors.”
Ask for one variation at a time. If the result misses the mark, change a specific detail rather than rewriting everything.
Avoid including private information in prompts. Use only material you have permission to share, especially when working with uploaded references or images of other people.
Keep a copy of the prompt that produced a useful result. It can help you make related images with similar direction, though the output may still vary.

Troubleshooting

Gemini does not create an image: Confirm that you are signed in and that image generation is available for your account and location. Try a direct request such as “Create an image of a yellow kite over a field.” If you are using a managed work or school account, ask the administrator whether Gemini features are restricted.

The request is refused: Re-read the prompt and remove any element that may conflict with the service’s current content rules. Try describing a harmless, fictional scene in neutral terms. Do not try to disguise a prohibited request with different wording.

The image is unrelated or misses key details: Shorten the prompt, put the central subject at the beginning, and make the requested composition explicit. Then send a follow-up that identifies the single most important correction.

The result looks garbled or contains unwanted text: Ask for an image without writing, letters, or labels. Add any exact wording afterward in an image editor, where you can check spelling and placement.

There is no download button: Select or open the image and check the available controls, or try Gemini in a current desktop browser. If saving remains unavailable, check the app’s permissions and your device’s download or photo settings. Availability of controls can differ between devices.

The page is not responding: Refresh the page, check your connection, and try again later. If the issue continues, try another supported browser or the Gemini app. Avoid repeatedly submitting the same request while a response is still loading.

Frequently Asked Questions

Is Google Bard still available?
Google Bard was renamed Gemini. Use Gemini on the web or in its app where the service is available.

Can I create images in Gemini for free?
Access and limits can change and may depend on your account, region, or plan. Check the current Gemini interface and Google’s service information for what is available to you.

Do I need to install an image-generation app?
No, if image generation is available in your Gemini account, you can request an image in a web browser or the Gemini app. An image editor is optional for later changes.

Can Gemini edit an image after creating it?
You can ask for changes in the conversation, and Gemini may generate a revised result. Describe what to change and what to preserve; the exact editing options can vary.

Why does Gemini refuse to make my image?
The prompt may conflict with current usage rules, or the feature may not be available to your account. Review the request, check availability, and try a safe, clearly worded alternative.

Can I use a generated image commercially?
That depends on the service’s current terms and the laws and rights relevant to your use. Review Google’s terms and check for recognizable people, trademarks, or other material that may require permission.

Method: the standard formula for this calculation, rounded as each result states. See how the calculators work or report an error.

More cards in this drawer

The whole Numbers and Powers drawer →