#435 – AI Image Generation with Nano Banana

#435 — AI Image Generation with Nano Banana

AVID Open Access
12 min

Transcript

The following transcript was automatically generated from the podcast audio by Otter.ai. Because of the automated nature of the process, this transcript may include unintended transcription and mechanical errors.

Paul Beckermann 0:00 Welcome to Tech Talk for Teachers. I’m your host, Paul Beckermann.

Transition Music with Rena’s Children 0:05 Check it out. Check it out. Check it out. Check it out. What’s in the toolkit? Check it out.

Paul Beckermann 0:16 The topic of today’s episode is AI image generation with Nano Banana.

Transition Music with Rena’s Children 0:23 Here’s the, here’s the, here’s the tool for today.

Paul Beckermann 0:26 All right, so what is Nano Banana? Nano Banana is Google’s brand new generative AI image generator. It’s a creative alias for the Gemini 2.5 Flash image model. It specializes in photo realism and maintaining character consistency across edits. In practice, it allows you to enter a prompt and receive an image generated based on your description. For instance, if you type “generate an image of a dolphin jumping out of the ocean water,” you’ll get a version of that image.

If you don’t like the result, you can type in a prompt asking for a revision of that image or an entirely new generation based on a new description. The more detailed your descriptions, the more likely you’ll get what you envision. You can even indicate what style, lighting, or coloration you’d like. For instance, you might follow up with “revise this image in animated style. At night, under moonlight,” you’ll get that same image with the requested edits.

Instead of generating an image from scratch, you can also upload your own image and ask for Nano Banana to edit that image. For instance, you might draw a sketch of something you’d like to create, upload that, and then ask the program to flesh it out in a style of your choice. This is a great way for students to give their input into what that image will look like. They can draw that original image.

Another option would be to upload two images and have Nano Banana combine them. For instance, you might upload a picture of a person holding a fishing rod and then another picture that features a flag. For the prompt, you could ask the AI to combine the images by replacing the fishing rod with the flag in the other image, resulting in a picture of the person holding the flag.

Other options include removing items from an image, adding new elements to a picture, changing outfits, locations, and hairstyles, applying styles and textures to an existing image, adjusting facial expressions and poses, changing color schemes, colorizing a black and white image, and even restoring and enhancing old photos.

All right, so if this sounds good to you, how do you access this? Well, there are two ways to do it. You can access it via Gemini or Google’s AI studio.

If you access Nano Banana through Gemini, it will remember your conversations and automatically save them for reference. Provided you’re logged into your account in Gemini, the option appears as a banana icon in the word “image” at the bottom of the prompt field. By clicking the plus icon, you also have two options for adding your own reference images: uploading and getting them from Google Drive.

If you access Nano Banana from Google’s AI studio, you’ll click the option that says, “Try Nano Banana”. It’s located as a quick access option right below the prompt field. Using this option will mean that your conversations are not saved by default, and instead, you’ll engage in a temporary chat. When I did this, I was prompted by a warning: “your work is currently not being saved and will be lost. We recommend enabling Google Drive”. You will always have the option to turn on temporary chat for conversations you do not want to save in your chat history. In AI studio, if I chose to enable the Google Drive, I could indeed save my conversations and generations. But it’s a bit more involved.

When you click the “try Nano Banana” option here, it’ll take you to a screen with just the prompt field in the guiding words “generate an image of a banana wearing a costume”. It’s a Getting Started idea for you. There’s a plus sign on the right side of the field. When you click that, you’ll get options to add an image from your drive, to upload a file, to take a photo with your device’s camera, or to choose from a palette of sample media. Instead of clicking enter to run the prompt, you’ll click a yellow “Run” button. Or you could choose the keystrokes control plus enter.

In short, if you already have a Gemini account and you want to save your history, it might be simpler to begin in Gemini. If you’re more interested in having the additional image options, you might want to use the option featured in Google AI studio. Regardless of the path you take, the image generation experience will be similar.

Transition Music 4:36 How do I use? How do I use? Integration, inspiration, integration ideas.

Paul Beckermann 4:44 So how can teachers and students use Nano Banana? Let’s start with a few ways teachers might benefit from using this tool.

One would be lesson visuals. This might mean quickly generating custom images to match content in a document or slideshow.

Number two, engagement hooks. Maybe you start class with a funny or surprising AI-generated image tied to a lesson theme to spark curiosity. For instance, language teachers might put vocabulary words into one common image.

Three, assessment prompts. Use generated images as writing, discussion, or problem-solving prompts.

Number four, project support. Provide visual resources for student presentations or classroom displays without having to search stock photo sites.

So those are teacher ideas, but the real power often comes from students using these tools. So how might students use Nano Banana?

Number one, creative writing. They could generate illustrations of characters, settings, or scenes for their stories. I really like that idea.

Number two, multimedia projects. Students can add visuals to slideshows, digital portfolios, or videos.

Number three, STEM representation. Maybe they create diagrams, models, or concept illustrations for the work that they’re doing.

Number four, art exploration. They can experiment with different styles and aesthetics, blending art and technology together.

Number five, collaboration. They can work in groups to generate visuals that represent a shared idea and then refine and justify their choices.

Number six, critical thinking. Students can compare AI-generated images with real photographs to practice media literacy and evaluate accuracy and bias. This last one is really powerful, especially as tools like this make it harder for consumers to know if an image is authentic or not. Students can even help develop criteria for identifying images as authentic or not, and then they can practice that.

Using a tool like this also helps to demystify it a bit and create more critical consumers of that type of information. Tools like this are always great in theory, but I wanted to test it out myself to see if it was as great as it’s been advertised to be. For my tests, I decided that I would be a student generating images for a story that I had written. I could see this being a really strong application of the tool. Students would first write their story and then they’d use Nano Banana to illustrate it. Not only would this process help to bring their stories to life, but it would force them to think very critically about how to craft a prompt and get the image that fit well with the story that they’ve written. This undertaking would likely lead to lots of iteration and disappointment before resulting in success, and this process in itself, can be a great life lesson.

To begin my experiment, I tried generating an image from scratch. I prompted the AI to generate an animated image of a skunk spraying a bear in the woods to go along with a story narrative. The outcome was a very nice animated picture of a bear and a skunk in the woods. However, the skunk was spraying the bear out of its nose. This would not work for my story. And no matter how many times I tried revising the prompt, I could not get the skunk to lift its tail and spray as it normally would do. When I followed up with similar prompts, like an animation of a bear walking alone in the woods, it did a great job, but it could not get the skunk right.

I wanted to test this some more, so I entered the simpler prompt that I opened this episode with: “generate an image of a dolphin jumping out of the ocean water”. The result was perfect, realistic, and as I described. I followed this up with “revise this image in an animated style at night, under moonlight”. Again, the revision was perfect. In fact, I really liked how it turned out, complete with the glow of the moon on the water and the moonlit outline of the dolphin. This would work really well for a story about that dolphin.

From there, I moved on to uploading my own image. I added a picture of a penguin that I took on a recent trip to South Africa. The original image was a side view, and I prompted the AI to quote, “change the angle of this picture so it shows the front of the penguin rather than the side”. Unquote, this one worked out perfectly as well. I could not tell the image was AI-generated. It simply looked like I was in front of the bird when I photographed it.

Next, I prompted it with this: “Now place this penguin in a scene walking down a busy city street”. The result was impressive, complete with realistic shadows and an appropriately blurred background. Really nice.

Then I prompted, “keep this penguin identical and place them on the ice in Antarctica”. Again, it was very impressive. It looked very realistic, complete with shadows and all. I tried one more iteration, typing, “add a snow storm”. This resulted in the same penguin walking in a near white-out condition, its head slightly bowed down to shield from the wind and more of a side angle than the first. While this one was the most different of the set, it was still clearly the same character and would work for a student’s story. I was pleased that I could change the compositions, but keep the character the same from image to image, something that would be really important when illustrating a story.

Overall, Nano Banana seems like it could be a useful tool, especially as a way to enhance and illustrate larger projects. It can save time and also still require creativity, imagination, and problem-solving when used by students. If you do decide to use this tool with students, be sure to check with local, state, and federal guidelines to make sure it’s okay to have students engage with it.

If students are not able to engage individually with Nano Banana, you might consider facilitating a full class experience with you being the driver of the tool and the students in the class helping you to develop prompts and iterative follow-ups.

However you proceed, this will be a tool to keep an eye on with the rate that AI multimedia generation is evolving. It’s sure to continue changing and improving.

To learn more about today’s topic and explore other free resources, visit avidopenaccess.org. Specifically, I encourage you to check out the article collection, “AI in the K-12 classroom,” and, of course, be sure to join Rena, Winston, and me every Wednesday for our full-length podcast Unpacking Education, where we’re joined by exceptional guests and explore education topics that are important to you. Thanks for listening. Take care and thanks for all you do. You make a difference.