What do you do when you have free access to cutting-edge AI with all-encompassing knowledge? You make some sweet AI wallpapers, of course! Jokes aside, there are many cool things Gemini can do on Android, and one of them is generating images. Sure, it’s not the only AI with image-generating capabilities, but it’s mostly free and produces some of the best results. Besides, you may already have it installed if you have one of the best Android phones.
No Thanks, Keep Reading
Why bother with Gemini when a bunch of great Android wallpaper apps are available for free from Google Play? Well, because with Gemini, you can create something truly unique, and the creation process can be lots of fun. And even though Android 14 introduced a built-in AI wallpaper feature, Gemini can generate a broader type of images. Here’s how you do that.
Related
7 things I learned from using AI-generated images for the first time
What you may not know about AI image generators
How to generate images with Google Gemini
The basics of requesting AI images
Generating an image with Google Gemini is quite easy. Generating an image of something specific can be trickier, but knowing how to structure your prompt helps greatly. The prompt is the instruction you type into the Gemini app. It doesn’t have to be super long, but giving Gemini more and clearer instructions tends to return better results.
“Generate an image of…” is a good way to begin. Then comes the object or scene you want visualized and, if applicable, an action they’re performing. After that, you may include details like colors, style, background, camera angle, and the year or historical period. You may also specify negative modifiers – things you don’t want in the image. Trust Insights has a helpful in-depth guide on structuring prompts depending on which AI image generator is used.
Here is the prompt I used to generate the image on the phone in the photo above:
- Generate an image of a snowy nature scenery with river flowing down the middle, pine tree forest, at sunset, in wintertime
If you’re not getting the result you expect, you may try rearranging your prompt or adding details for clarity and then requesting a new image. Tap on the image to download the full-resolution version.
Generating wallpapers with Google Gemini: tips and tricks
How to optimize prompts to make great wallpapers
Even a visually pleasing image can be a bad wallpaper. You don’t want to generate anything too busy or too bright as it can make app icons less discernible and text more difficult to read. That’s why it’s a good idea to include specific colors and their placement in your prompt. Specifying the time of day is also a reliable way of getting the desired color tones. You may add “bokeh” or “shallow focus” to blur a portion of the image.
Example prompts:
- Generate an abstract image. Pattern of triangles, small and big. Top half is red, orange, yellow. Bottom half is black
- Generate an image of a fantasy castle on top of a steep, impenetrable mountain, with a red dragon circling above it, dark color tones, at night
- Generate an image of red wool fabric with yellow glitter, shallow focus, extreme close-up
Dealing with Gemini’s fixed aspect ratio
Currently, Google Gemini can generate only square images. Setting one as wallpaper will inevitably crop out parts of it. To increase the chance of your subject remaining visible, you may clarify that you want a symmetrical image or your subject in the middle. Alternatively, you can put different parts of the image as lock and home screen wallpapers—that’s what I did in the picture above. This works well for scenes of nature and landscapes.
Example prompts:
- Generate an image of a lion, bright neon yellow, orange and red colors, flat style, black background, symmetrical
- Generate an image of a fantasy village in the style of Van Gogh’s Starry Night, lone tree in the foreground, mystical mountain in the background
Specify a style to guide Gemini’s algorithms
It’s a good idea to specify a style for your image. In my experience, this yields eye-pleasing results and limits the chance of Gemini generating something wildly unexpected. For example, you may request an abstract, minimalist, fantasy, or photorealistic image, a cartoon, or a watercolor or oil painting. Naming an artist or specifying a famous painting as a reference is also an option worth experimenting with.
Example prompts:
- Generate an image of a labrador in the style of the Mona Lisa
- Generate an image of hot air balloons flying over a seaside village, expressionist style, oil painting
- Generate an image of the android logo, rainbow colors, watercolor painting style, textured paper background
- Generate an image of a sakura forest at sunrise, Minecraft style, high and wide camera angle
- Generate an image of a cute cat in the style of the Pokémon cartoon, pastel colors, white background
Using a reference image for inspiration
Google Gemini lets you upload an image, and the AI algorithms will use it as a reference. Uploading a photo is an alternative way of defining the objects and composition of the generated image. You can still request a style in your prompt — for example, you can turn a favorite photo from your camera roll into a cartoon or a painting. Or you might ask the AI to remix a famous painting.
I got the image above by uploading Botticelli’s “The Birth of Venus” and asking Gemini to “generate an image like this, but with cats instead of people.”
What images can’t Google Gemini generate?
Even the best AI has its limitations
Like every AI image generator, Gemini has its weak spots. One annoyance is that all images are square in proportion and only up to 2048 by 2048 pixels in resolution. This is good enough for 1080p screens, but not ideal for the best phone displays we see on high-end flagships. Another drawback is that text and patterns are more likely to be glitchy.
There are a couple of other restrictions, too. People can appear in your images, but only if you have a paid Gemini Advanced subscription, and images of public figures are not allowed, but popular cartoon and game characters turn out fine.
Google Gemini is not just a gimmick
Making wallpapers with Gemini represents just a fraction of what Google’s AI tool can do. In fact, the biggest changes in Gemini 2.0 make it more useful than ever. The update added better processing of audio and image information, agentive AI features, and deeper integration with Google’s ecosystem, among other goodies. Using the best Gemini extensions, you can be more efficient both at home and at work.
Although one could argue that Gemini is getting complicated too quickly, especially for something that’s been around for only about a year. Whether you feel the same or not, it’s guaranteed that AI is here to stay, and it’s only going to get more capable in 2025.
