5个谷歌 Gemini AI 照片提示词上手

Google's AI tool, Gemini, has drastically improved its ability to create and edit photos. And people are looking for, and posting about, the best ways to use the tool. That's because Google recently integrated Nano Banana — the company's AI-powered image generation and editing tool — into Gemini's new Gemini 2.

5 Flash Image. Simply put, if you go to Google AI Studio or to the Gemini App, you should be able to make better photo creations. In Mashable's tests, Nano Banana works particularly well as an image editor.Now, I'd probably argue that instead of playing around with AI, you might simply go out into the world and take pictures.

Enjoy some sunshine, if you're so lucky. But hey, whatever floats your boat or drains your water supply. So, if you're serious about using an AI image generator to get interesting results, we've got some good prompt ideas to get you started. Google Gemini AI photo prompts to try Ask it to make trading cards: If you're really into sports cards, fantasy card games, Pokémon, or similar hobbies, you can feed Gemini photos, then ask it to create trading cards.

As usual, specificity is key: You can ask Gemini to include stats, change outfits, or make tweaks just how you like them. Make all your childhood dreams come true. The idea is to be as specific as possible. AI-GENERATED IMAGES Credit: Google This Thread, for instance, suggests telling an AI tool all the different categories, attacks, and other necessary info to get a good Pokémon-style card.

If you wanted a football playing card, you might tell Gemini to include stats — specifying categories like rushing yards, attempts, receptions, or touchdowns — and give other details to include, such as a fun nickname. Mashable Light Speed Add things to the background: As Mashable Tech Editor Timothy Beck Werth showed, Gemini does a solid job of adding new details to images.

In his example, he added dolphins in the water alongside a boat. But you can add just about anything to an image. In general, try to include details about how you want it to look. Typically speaking, I've gotten the best results specifying if I want it to look photorealistic. For instance, I added realistic-looking rain by asking for "easily visible rain that appears photorealistic."

Remix photos of your pets: Here's an example straight from Google itself. In a blog post about fun Google Gemini AI photo prompts, Google provided this prompt idea along with the before and after results: "Prompt: Recreate this cat as a 16-bit video game character, and place the character in a level of a 2D 16-bit platform video game."

AI-GENERATED IMAGE Credit: Google Create or edit a headshot: Lots of folks need headshots for professional reasons. I'd suggest using a photographer for best results, but in a pinch, you can use Gemini's model to create or edit an image. I tried giving Gemini a random picture of myself and asked for a "professional-looking headshot" based off that.

The results were quite uncanny valley. It looked enough like me to kind of work, but it was also clearly not me. I'd suggest using AI to edit a headshot rather than invent one out of whole cloth. If you're looking for help in crafting prompts, Reddit has lots of ideas about how to get a specific output.

Combine two photos: As we've shown before, Gemini does a solid job of transposing one photo onto another. So, as we detailed, you could add a spice bottle to an image of a fire — or whatever else you could dream up. So you might ask Gemini, "Combine these two photos. Take the hand holding a coffee and transpose it into the foreground of the picture of the Grand Canyon.

Match the lighting and shadows so it appears the two images were taken as one." Enter your favorite fantasy world: Say you love Game of Thrones. Well, you or anyone else can now enter that world via Gemini. It'd be a fun bit to send to fellow fans. The prompt is pretty simple: Upload a photo and ask Gemini to make the person or thing appear to be a part of your favorite franchise.

There's plenty to try out with Gemini — after all, it can make almost anything. In my experience, it might take a little bit of playing around, but you can get useful results by fine-tuning your prompts. For more ideas on how to use Nano Banana and make and edit AI images in Gemini, the Google blog has more examples.