Everyday AI Podcast – An AI and ChatGPT Podcast - EP 598: Nano Banana! Real Use cases for Google’s new Gemini 2.5 Flash Image

Episode Date: August 27, 2025

Nano Banana is no longer a mystery.Google officially released Gemini 2.5 Flash Image on Tuesday (AKA Nano Banana), revealing it was the company behind the buzzy AI image model that had the internet talking. But... what does it actually do? And how can you put it to work for you? Find out in our newish weekly segment, AI at Work on Wednesdays.Newsletter: Sign up for our free daily newsletterMore on this Episode: Episode PageJoin the discussion: Thoughts on this? Join the convo and connect with other AI leaders on LinkedIn.Upcoming Episodes: Check out the upcoming Everyday AI Livestream lineupWebsite: YourEverydayAI.comEmail The Show: [e-mail] with Jordan on LinkedInTopics Covered in This Episode:Gemini 2.5 Flash Image (Nano Banana) RevealBenchmark Scores: Gemini 2.5 Flash Image vs. CompetitionMultimodal Model Capabilities ExplainedCharacter Consistency in AI Image GenerationAdvanced Image Editing: Removal and Object ControlIntegration with Google AI Studio and APIReal-World Business Use Cases for Gemini 2.5Live Demos: Headshots, Mockups, and InfographicsGemini 2.5 Flash Image Pricing and LimitsIterative Prompting for AI Image CreationTimestamps:00:00 "AI Highlights: Google's Gemini 2.5"06:17 "Nano Banana AI Features"09:58 "Revolutionizing Photo Editing Tools"12:31 "Nano Banana: Effortless Video Updating"14:39 "Impressions on Nano Banana"19:24 AI Growth Strategies Unlocked20:58 Turning Selfie into Professional Headshot24:48 AI-Enhanced Headshots and Team Photos29:51 "3D AI Logo Mockups"32:22 Improved Logo Design Review35:41 Photoshop Shortcut Critique38:50 Deconstructive Design with Logos44:01 "Transform Diagrams Into Presentations"46:12 "Refining AI for Jaw-Dropping Results"Keywords:Gemini 2.5, Gemini 2.5 Flash Image, Nano Banana, Google AI, Google DeepMind, AI image generation, multimodal model, AI photo editing, image manipulation, text-to-image model, image editing AI, large language model, character consistency, AI headshot generator, real estate image editing, product mockup generator, smart image blending, style transfer AI, Google AI Studio, LM Arena, Elo score, AI watermarks, synthID fingerprint, Photoshop alternative, AI-powered design, generative AI, API integration, Adobe integration, AI for business, visual content creation, creative AI tools, professional image editSend Everyday AI and Jordan a text message. (We can't reply back unless you leave contact info) Start Here ▶️Not sure where to start when it comes to AI? Start with our Start Here Series. You can listen to the first drop -- Episode 691 -- or get free access to our Inner Cricle community and all episodes: StartHereSeries.com Also, here's a link to the entire series on a Spotify playlist. 

This is the Everyday AI Show, the everyday podcast where we simplify AI and bring its power to your fingertips. Listen daily for practical advice to boost your career, business, and everyday life. Meet Firefly AI Assistant, now live in Adobe Firefly, the All In One Creative AI Studio. Just describe what you want to create and the assistant handles the rest, orchestrating multi-step workflows across Photoshop, Premiere Express, and more in one conversational interface. You direct the outcome. The assistant accelerates execution. Nano banana is no longer a mystery.

So if you follow AI at all, you've probably seen the very viral images making their rounds of some mystery AI image model called nanobanana. And after some not so subtle hints from the Google team over the last two weeks, it was finally, released on Tuesday. So the real name, although I kind of wish they just kept it as nanobanana, is Gemini 2.5 flash image. So on today's show, we're going to be going over it's Wednesday. So we're going AI at work on Wednesday. So we're going to be going over, hey, great timing. What's new in Gemini 2.5 Flash image and also giving you some real use cases and live demos for this impressive new model from Google Gemini. All right.

It's time to go bananas. I told myself I wouldn't do this. Make banana jokes. Let's leave it there. Welcome. What's going on, y'all? My name's Jordan Wilson. Welcome to Everyday AI.

If you're new here, we do this thing every single day, Monday through Friday. It's unedited, unscripted. I like to say it's the realest thing in artificial intelligence. It's a daily live stream podcast and free daily news that are helping us all keep up and get ahead when it comes to all of these AI updates. They're coming at us faster than bananas in the old donkey con game. Another one. All right.

So if you haven't already, please make sure to go to your everyday AI.com. Sign up for the free daily newsletter. We're going to be recapping the highlights from today's show as well as everything else happening in the world of AI to keep you ahead of the game. So if you want the AI news, that's going to be in the newsletter. So let's get straight to it. Nanobanana is no longer a mystery. Google did have a pretty splashy announcement of their very impressive new Gemini 2.5 flash image model.

So on today's show, we're going to detail what's new in Nanobanana, aka Google's new image-generating model, Gemini 2.5 Flash image. We're going to do a live demo on this AI at work on Wednesdays. And we're going to end the show with five helpful use cases that both non-technical and non-creative people can take advantage of, right? I think one of the hardest things when it comes to AI is number one, there's like 50 new things a day. But number two, it's like, okay, how do we actually use this, right? The everyday business leaders like you and me. Well, AI at work on Wednesdays is a new-ish segment. where I show you ways that we're using new AI models and modes and tools.

And for the most part, we kind of stick to OpenAI, Google, Microsoft, Anthropic, for the most part, right? Because I don't want to go down too many rabbit holes. So at least for me, this one's exciting because a lot of times we're doing more text-based large language models. This one's going to be a little bit more visual. So if you are listening on the podcast, this might be a good one. Go to our website.

We always put the video from the live stream. You can watch it on our YouTube channel, LinkedIn, et cetera. So I'm going to do my best for our podcast audience. But clearly, this is going to be a visual one. So I'm going to do my best. All right. Enough to chat. Let's get into it.

And let's look at what is new. So pretty good write up actually from Google DeepMind on the new release. Google actually does a much better job than most when it comes to new models, what their capabilities are, et cetera. So make sure you go check that out. and we'll put that in a newsletter. But here is the details. So like I said, it's no longer a mystery.

So essentially this nano-banana model was being tested by early evaluators and on LM Arena. So if you don't know what LM Arena is, it's where you get put in a prompt and you get a side-by-side output. You don't know what they are. You vote for the winner. And that gets what it's called an ELO score. So a lot of times, sometimes weeks before a new model comes. out, whether that's a text-to-text AI model, a text to image, image to video, etc.

They're on LM Arena, but sometimes with code names. So this new Gemini 2.5 Flash image model that was just officially released have been on LM Arena as Nanobanana. But yeah, unfortunately, you couldn't just go select it and use it. You had to go in that battle mode and you would just randomly happen upon nano-bonata. So when people saw it and how good it was specifically at editing, and manipulating images, it rightfully so kind of took over the AI internet. So it is the benchmark leader and it's not even close.

I'm going to share those benchmarks later. So yeah, when we talk about ELO scores, what humans prefer when it goes to head to head, blind competition. This model, Gemini 2.5 flash image is not off the charts because it's number one, but it's off the charts compared to everyone else. some pretty advanced features. So the biggest one, it's a multi-modal model at its core. And that's important.

So Google actually has other image models, right? They have a great model called Imagine 4. But the Google's version of Gemini 2.5 flash image works a little differently. It has a much better understanding of natural language. You can work with it in natural language, much better than you can with their other imagine for it has a better understanding of the world and physics and you know if you're making a series of images you don't have to explain you know a new scenario you could say hey put this basketball on mars and players are probably going to start floating around right because they're like oh okay well

gravity is gone is gravity gone on mars i think it is so a lot of advanced features but i think one of them is well it is a multimodal large language model at its core which is extremely important when it comes to working and iterating with nanobanana. I'm just the rest of the show, do you guys mind? Can I just call it nanobanana? I don't know. Gemini 2.5 flash image is a mouthful. Nanobanana is just fun to say.

All right. So nanobanana, the other big thing is it maintains character consistency, which is huge. So if you want to do multiple photos of someone in different scenarios, but, you know, if you've ever used AI image, generators, even the very impressive and extremely viral. Also, GPT4O image gen from a couple of months ago, it really struggled with character consistency. Or if you uploaded a photo of yourself, it always seemed to change a couple of things. Not so with nano banana. Also, the ability to blend multiple

images and just, like I said, using Gemini's world knowledge is huge. Let's talk a little bit about availability. Well, it's available now. So it's already rolled out. today across Gemini apps for all users, plus developers via the API inside Google AI Studio, as well as obviously on the front end, Gemini and Gemini apps. And also, many of the third parties are integrating it, including, interestingly enough, Adobe, right? Adobe, I think in this one, they have no choice. A lot of people are calling this a Photoshop killer.

I won't go that far. But if Adobe doesn't do something, they're going to run. into some problems. Adobe's stock actually took a pretty big hit yesterday when Nanobanano was announced, believe it or not, that something called Nanobanana, you know, I don't know how much market cap it took off Adobe, but it was a couple billion. So usage and limits. So this has been going back and forth a little bit. We'll probably have some clarity, you know, by later today. And if we do get a little more clarity, I'll put it in the newsletter. But for the most part, it's free. So right now, free users get 100 daily edits inside Google AI Studio and then paid users get

a thousand. And then on the developer side, so if you're using this on the API side, it is extremely cheap. It is about $3.9.9. So let's just say $4 per image. I believe that's like a quarter of the price of OpenAI's GPT4O image. And also way faster, right? That's the problem. I really did love GPD 4O image gen, but just because I guess, you know, they're having

scaling issues, right? So it can sometimes take 30 seconds, 60 seconds, multiple minutes to generate an image in GBT40 ImageGen right now, even though, I mean, we'll see. I'm going to do it live. Even though it came out and I know that their servers, Google servers are getting slammed, they're able to handle it scale. It's fast. Right. And usually when something this popular comes out in the first 24 to 48 hours,

you can expect a lot of slowdowns, a lot of throttling. I haven't seen that. Things are generating really quickly. Also, it is important to know that all outputs include AI watermarks and synth ID fingerprints. Well, save that for another, another day, another episode, because I think as powerful, and I'm not just talking about nanobanana here, I'm just saying this genre of AI image generating as exciting. and powerful as it is, it also can have a pretty dark side. But like I said, we'll save that for another episode because this is the fun nano-banana episode.

Let's talk a little bit more about the capabilities. And this is where it really shines. So precise object control. So being able to instantly remove people from photos, delete clothing, stains, blur backgrounds, change poses, add color to black and white images using simple text commands. I mean, this is like, and this is coming a little bit from my background, right? So I've been using Adobe Photoshop. Let me do the math here for, I think, more than 20 years, just about 20 years.

So 20 years, give or take a year or two. Some of these things, right, like removing people from photos or, or, you know, changing the color on something and preserving the texture, that is not easy. the lighting and the composition when you're changing or putting a character in different scenes. It's not easy, right? I've myself, you know, I have a kind of a marketing tech background. So I worked at a nonprofit, but I was the national marketing director and then the executive director.

But we worked on a lot of huge campaigns with Nike and Jordan. So essentially, we just became an activation and marketing agency for Nike and Jordan. And I literally, I'm thinking of it right now, so many projects that I spent probably dozens of hours on single projects, right, doing one of these things, right? Removing someone from a photo, changing something in Photoshop, right? Take seconds now. The capabilities are wild, right? I do think the cap now is your imagination.

And I do want to hit a pause here and say this. This is not just for creatives. I think this changes the way that non-creative people can actually create. And that is huge. I think that breaks down one of the larger or more daunting barriers that a lot of non-technical or non-creative people have faced for decades, right? You know, you're like, oh, if only I could, you know, be creative. I'm great at executing on the back end. But man, I can't strategize. I can't create, but I can execute. Okay. Well, you can strategize now with large language models like Gemini, like GBT5, like Claude 4.

But then you can use something like nanobanana and you can actually create things. It is a multi-modal input output powerhouse, right? I'm not even talking about Google's industry leading V-O-3 video model, right? And how you can use nanobanana to start by creating or editing or putting yourself in a different image, right? So let's say you're the CEO with small company and, you know, you have an older, you know, kind of corporate video and it's kind of boring and you want to update some of the shots. Okay, well, take a picture of yourself with your camera.

Load that into nanobanano. Nano banana. You know, go into different scenes. And there you go. You have new B-roll. You don't have to go out and spend $30,000 and many hours hiring an expensive production crew just if you want to update, you know, the old training video. So some other things, aside from character consistency, multi-step editing, so you can start with an empty room, then paint the walls blue, add a bookshelf, place a couch, etc. Great for interior, exterior design, you know, scene setting, et cetera, style transfers.

So you can apply textures from one image to create something new. So as an example, you could use a butterfly wing pattern for clothing or flower petals to redesign rain boots. So if you have different textures or if you give a photo of yourself, you're like, hey, I wonder what I'd look like and, you know, these kind of shorts or this kind of, you know, top, well, you can do that, right? The whole this kills, I think, well, kills, but also create so many opportunities. It's going to kill some current, you know, startups and in apps out there, but it's also going to create so many opportunities of what you can build on top of this software. It's not terribly hard. And then also kind of already mentioned this, but smart image blending. being able to upload multiple images and blend them together.

All right. Let's look live. What could go wrong? All right. So I will tell you one thing that I've noticed in my limited experience, I don't, I think I need to reach out to Google folks, right? We have them sponsoring our podcast all the time. But for some reason, like, I don't always get the Google stuff early. Sometimes I do.

But yeah, so I unfortunately didn't get nano banana early. like I know a lot of other people did. So I've only been able to play with it for a few hours. And I will tell you this, in my experience so far, I don't, at least for me personally yet, I haven't gotten a lot of impressive kind of one-shot results, right? Got an extremely impressive results when you iterate with it a little bit. But that's the, like I said, that's the, you might look at that as a con,

but that's actually a pro of using a multimodal by D. default model like Google Gemini and not using like, you know, Imagine 4. Nothing against Imagine 4. It's one of the top rated, just text to image models in the world, actually now only behind Nanobanana, but it's so much different working with an image editing and image creating model inside of a large language model and not inside of a platform like a mid-journey or a, you know, stable diffusion, right? So that's the power is just being able to talk to the model with natural language

and not having to speak, you know, prompt, you know, with these, you know, S-ref codes and, you know, all these, you know, adjectives throwing out of, you know, 1970s paintings. It's like, no, just be like, yo, make this look like this and it'll do it. All right. So let's look live. So before we start demoing, all right, like I was. mentioning, here is the LM Arena. Elo scores.

And I don't know the last time I saw something come in at like 170 points above the next highest variant. But that is where Gemini 2.5 Flash image preview, aka Nanobanana, came in over the next highest. And this is for image edit. Text to image, obviously, Nanobanana was also number one, but I'll only buy a small margin over Google's own Imagine 4. So they're different products.

Keep that in mind. But on the image edit side, I mean, absolutely bonkers. I didn't say banana there. I thought I was going to say absolutely bananas. I didn't. I'm refraining. That one's absolutely bonkers. It is 170 points above the second place model in Flux 1 context.

I don't remember the last time a model came in that much higher. in an established category. If it tells you how freaking good it is, I don't remember the last time we saw this. On the text side, right, as an example, you have Gemini 2.5 Pro with a two point lead over GPT5.

Right? So pretty much tied. You know, they're 1A and 1B, which is normal. So, you know, two, five points, pretty big,

20 points. You're like, oh my gosh, 170. That means it is the category, right? There's no competition. So real quick, benchmarks, similarly, much better, right? I don't know if y'all follow image benchmarks.

I do not as much as I follow kind of the traditional large language model, but on overall performance on LM Arena, visual quality on Gen AI bench, text to image, text rendering, pretty much Gemini 2.5 Flash image is leading in most of those categories. But surprisingly, Imagine 4 is pretty up there as well. All right. Let's get demoing. What could go wrong? So I'm going to be working inside both Google Gemini and Google AI Studio.

And I'm going to show you why. But a lot of times I'm also, hey, when this is putting AI at work on Wednesdays, when something is brand new like this, I don't even necessarily know the best way yet. And unless you were developing the product at Google, most people don't. So, you know, if sometimes you feel like you're behind, don't worry, y'all. I do this every single day and I'm still learning like everyone else. So for a podcast audience, I'm going to be flipping between Google Gemini and Google AI Studio. So Google AI Studio is free.

And I dig it to a little bit of a makeover. So now you just need to make sure there's a lot more information when you're selecting models. You can go to all models feature Gemini images, et cetera. So you just need to click on the Gemini 2.5 flash image preview if you are in Google AI Studio. And I do believe it's at a hundred cap a day. And then I am on a paid plan on Google Gemini. So I'm not going to run into any caps. So that's another reason why I'm doing it in both. So we can hopefully not run into any issues. All right. So, Adobe just introduced an entirely new way to create, bringing the power and precision of its creative suite into one conversational experience. Meet Firefly AI assistant,

now live in the Adobe Firefly app, the all in one creative AI studio. Powered by Adobe's creative agent, Firefly AI assistant lets you start with your vision, just describe what you want, and shape the outcome as it takes form with the assistant. The assistant orchestrates multi-step workflows drawing on 60 plus pro-grade tools across Adobe Creative Cloud apps, including Photoshop, Illustrator, Premiere, Lightroom Express, and more to help bring your ideas to life. You can also get started with creative skills, a growing library of pre-built workflows for common creative tasks like batch editing photos, creating mood boards, portrait retouching, and creating social variations. Every step the assistant takes is visible so you can refine, redirect, or take over at any

time. You stay in the driver's seat as the creative director. Adobe Firefly AI assistant now in public beta. See it today at firefly.adobie.com. I'm going to, where is this photo? All right, give me a second. I got to find a photo here. All right, there it is. So essentially, let me drag this on so the audience can see. This is just an audio photo I shared on LinkedIn or something when I got to Nvidia GTC this past year. So just a selfie picture of me and a hoodie. You can kind of see the everyday AI logo in the hoodie, which is why I'm actually using this.

I want to see how well it can preserve that and kind of put me in different scenes. All right. So I'm going to go ahead and drop that image. I know you, you know what? I'm actually going to try to turn this into a professional headshot. All right. So probably something a lot of people might be interested in. All right.

So I'm going to say, give me five. professional headshot options, but still in this exact hoodie. Don't add anything new on the hoodie. That's not, I'm going to say that's not already there. Make this, I'm going to say make this appear. It's shot on an iPhone. I'm going to say give me different angles, make them flattering but realistic. All right.

There we go. We'll see what happens. So I'm running that in Gemini. I should probably go into the right model. the Gemini 2.5 flash. And then I'm going to click Tools and then go on to create images in the tool selector. That's it. I'm going to copy that prompt.

And then I'm going to jump over in Google's AI Studio. I have the Gemini 2.5 Flash. The good thing is with this photo, you can just drag and drop it, which is nice. And there we go. And then we're going to jump between the two different and see what the results are. Cool. So we already ran into this. some fun. Let's see what I did wrong here.

Didn't look like I did anything wrong. So I was doing a little bit of testing in Google AI studio before. Wasn't running into any issues. It just gave me nothing. It just gave me blank in Google AI studio. So we're just going to run it again. Uh, yeah, Gemini 2.5 flash image preview. All right.

And let's check on. All right. Hey, you got to love the demo demons. They got me right before, right before started this recording. Everything was working great in Gemini 2.5 flash image preview inside Google AI Studio. It's not anymore. So this is going to make the demo interesting. All right.

So here we go. Let's look at Google Gemini using the new nanobanamoid. All right. So it gave me a couple different images. And you'll see here, this actually looks like me, right? For the most part, if I did this in a GPD 40 image, it would not really look like me, right? I don't know, last year, audience, if you watch, does this look like me? I'd say it pretty much does, right?

So again, we have the watermark. We have a photo. It took the same hoodie, different pose, right? But got me looking just like me. It looks like I'm in an outdoor park setting for the first one. The second one, same setting, straight on in the same kind of park area. Cool. All right.

So it looks like we got kind of five different, really, really good. It looks like we have a little weird kind of artifact on the hoodie on one of these, but overall, really good. So now what I'm going to say is I'm going to say, make this more Chicago-e. You guys know I like to do these on. live on the demo sometimes. Just say like make it more Chicago and see what happens. Presumably it's going to put the Chicago skyline in the background, but we'll see what happens. But this is what you can do. You just natural language. You don't have to go in, you know,

speak prompt to it. So these are photos that in theory I could use as, you know, professional headshots. If I wanted to, you know, update my LinkedIn photo, these are obviously better than the, you know, random selfie. Um, you know, know, that I shot at the NVIDIA GTC conference. There we go. We got the, the bean photo, obviously, but, you know, pretty, pretty decent outputs here from Nano Banana. All right.

And then if I wanted to, I could go further and I could say, hey, you know, actually, let me do this. I'm going to say, you know, make this a team photo and add to more people in the same outfits. All right. So again, this is the power of iterative prompting. Right. So I actually, we'll see how this does. I didn't try this one out yet while I go check on Google's AI Studio. That's just not working, just not working right now. So fantastic. I might have to launch one of my other accounts and see and see what's going on. But for whatever reason, it's not working. And we'll try it with with the next one.

All right, but there we go. It added two team members. I'm looking distorted the logo in the hoodie slightly, but it just added kind of, you know, two more people in the same clothing like I told it to. And in Chicago, the thing I like is it actually gave me, it kept the same hoodie. I like the ones that kind of go up to the chin. All right. So let's go to our next demo.

And again, I'm just going to. over use like use cases i might actually use this for right i'm still uh kind of figuring this out myself so i'm going to launch a new chat in both google gemini um and in uh google a i studio we'll see if uh this works so the next thing i'm going to do is i'm going to say redesign my cover art for my podcast give me two options um and i'm actually just going to i had one saved here so uh here we are There we go. I thought it made sense since Logan killed Patrick from Google,

um, helped seemingly lead a lot of this, at least from the developer relation and product management side. I've had Logan on the podcast a couple of times. So I'm going to upload, uh, that image and then essentially have it redesign my, um, my podcast art, right?

Some I've been thinking about for a while. Um, do I want to do it? Do I want to hire someone else? or will nano banana be good enough to do it itself? Let's see. So pretty simple here. All I'm doing, again, you can upload multiple photos.

You can combine them. But right now, I'm just uploading one text prompt, one image with our current podcast cover art, and essentially saying, like, change a bunch of stuff. And I'm also saying you can extend the photos. So Google AI Studio for whatever reason is being absolutely bonkers. At least it worked at this time, but it didn't really change anything. So not sure why. Let's check in on Google Gemini.

Same thing. Didn't really change anything. So interesting. In my original prompting, or my initial test, it did really well. And I'm going to say, you must make the layout design. very different. All right. So I'm going to do a follow-up prompt here, see if we can get anything because essentially when I ran this the first time, works great, gave me some cool layouts. One of them I actually liked and I'm like, oh, I should give this to an actual designer or just,

you know, continue working with this and, you know, maybe this could be the new cover art. But the second time around, different, right? This is the fun part of doing it live. All right, there we go. I do like this one a little bit better, even though it's a little bit better, even though it's little screwed up. So it gave me some, uh, some different options where it kind of has the, uh, the podcast title text, uh, over, uh, the person's face, uh, which isn't terrible, but, you know, maybe that's a cool, uh, cool layout where I can put the, uh, the guest face really big and, you know, prominent. Uh, and then just, you know, put my little mug, you know, somewhere smaller there. So, uh, not great output on that side from Google's AI studio, but at least

It worked on the second time. Okay. Got something decent-esque from Google Gemini. I wouldn't use these per se, but it at least gave me one, you know, slightly different layout for my podcast design. Cool. Let's do another one.

Here's one that I think just about everyone might be able to take advantage of. Again, I'm trying to find some general use case. that will be helpful for you all. So now what I'm doing is I'm essentially going to drop my everyday AI logo in here. And I'm saying put this on a bunch of mock-ups, you know, t-shirts, put it on, you know, mugs, et cetera, right? So, so nothing, nothing crazy. But I am saying make sure it's 3D, make sure it's realistic, right?

And put it on actual people, right? And I want to see how good of a job it does at just transferring the logo and making it look real. So it's already done on Google Gemini in that 10 seconds where I was copying and pacing the prompt and going over to Google's AI studio. It's already done in Google Gemini. So the photos themselves, let's take a look. It did a good job of transferring the logo. The logo looks a little bigger than it probably should.

I'm a fan of like smaller logos. But the photos look, you know, fairly okay. They look obviously AI generated ask, right? If these were on social media and small, you wouldn't be able to tell. I can tell. Here's another one of a woman. Again, the logo is pretty big. And then here's one of the water bottle.

So I'm not a huge fan of the giant logo. So I'm just going to say, I'm going to say make the logo, I'm going to say make the logo much smaller and realistic. and the logo lettering in black and white and ensure it looks more stitched, stitched on the clothing. Not screen printed. Yeah, we gotta get that good stuff. All right, so just iterating on the prompt, let's see what Google AI Studio did.

Google AI Studio did a pretty good job here. They gave me kind of a water bottle on the desk with a person in the background. Nice kind of boca effect, right? That looks really nice. Again, logo looks a little comically large, for a water bottle. Maybe not. I guess I'm holding up my kind of water bottle I got here, my Chicago Bears one. That's a ridiculously big Chicago Bears logo. So I don't know, maybe that's the right size.

Another one here, pretty good. The logo just looks screen printed, though. So that's why in Google Gemini, I asked for it to be kind of stitched. All right. So I iterated the prompt here. Did a good job. It made the logo black and white, and it looks less screen printed on this original photo. It looks like, you know, two college age, you know, kids in a city. And it made the logos a more appropriate side and made it black and white. And it looks stitched now versus screen printed. So overall, pretty good, pretty good. Better than the podcast cover art, if I'm being honest.

So you can see so far, you know, a lot of times you have to reiterate, right? You're not going to be knocked out of the park with your first, depending on what you're doing. Right. I'm trying to do more general use cases. I could cherry pick things that I knew would be extremely impressive, but I want to give examples that at least for our podcast audience and for our non-technical, non-created people watching here on the live stream that, you know, something that might resonate. So here we go. Let's do this one.

You know what? I'm going to switch this one up. So I have a photo of me in front of a green screen. All right. And let's just go ahead so you guys can see this. There you go. So photo in front of a green screen. And I'm just going to say, let's see, there we go.

I'm improvising here. So I'm going to say make this a full body shot of me walking around in downtown, downtown Chicago. All right. I love typing live. All right. Downtown Chicago. There we go. So this photo is from like the shoulders up.

I'm in front of a green screen. The rest of my body is obviously not showing. So let's see how nanobanana in Gemini and in Google AI Studio handle this. So there we go. All right. I did it in Gemini. I did it in Google AI studio. And Gemini done, y'all.

That took again so fast. So fast. All right. So it looks like a pretty good photo. It made me like six foot two, which I'm here for. It looks like in this photo, it looks like I'm like blinking or something like that, but not bad. So I'm going to say move me off center and a happy expression with wide eyes, right? So let's see how this does.

does. And again, I'm looking for this consistency, right? It's weird to say character consistency when it's myself. But there we go. Okay, so now I'm slightly, slightly off centered, a little bit to the right. I should have said, you know, using like rule of thirds, go to the, you know, right third. But here we go. So again, does this look exactly like me? Not necessarily, but it actually did a pretty good job. inside Google Gemini of moving me to the right. So now I can say something like make it nighttime and keep everything else. Same. So this is where you can see how you can save a ton of time,

maybe doing something that you might normally have to do in Adobe Photoshop, right? With a simple text prompt, it didn't take long. This one actually doesn't look good. It's done. It looks a little. from a lighting perspective, it doesn't really resonate. So it actually did a really good job with the background objects and the people in the background

and kind of better illuminating them under kind of darkness of Chicago. But at least for this version of me in the foreground, I don't think it did that great of a job. Again, maybe I'm nitpicking a little bit here. All right. So let's see how Google's AI Studio did with the same. prompt, just make this a full body shot of me walking around downtown Chicago. This one actually looks much more like me. Pretty good, actually, I'd say.

So I'm just going to iterate on this one. Same thing. I'm going to say make this nighttime. Keep everything else the same. See how it turns out. I was going to do, actually, let me just do one more quickly. I'll make this one quick and explain exactly what I'm doing, just to show you some of the other capabilities. All right, Google's AI studio did a much better job converting that one to nighttime.

Again, just taking a photo of myself kind of from the, you know, forearms or from my elbow up. Looks pretty pretty much like me, not 100%, but much better than GPT40 image would do. Put me in Chicago, change it from, you know, daytime to nighttime did really, really well. Here's one other capability that I really like. And I think a lot of people could find some use for. All right. I'm just going to let it run because it might take a second. But all I'm doing here is I have a prompt and I set at the very top of this.

All right. I'll just read it out here. I said this is text from my website, which mainly contains the transcript of a podcast episode of Everyday AI. Please create one detail of infographic or a series of images that accurately displays the main takeaway messages from the transcript. be specific, here's the text, right? And then there's a 40-minute transcript of my podcast episode from yesterday on the MIT study of 95% of AI pilots failing.

Bad study, by the way. Anyways, so I kind of left it up to nanobanana to just be like, yo, okay, interesting. So it didn't. That's because, okay, that was my bad. I forgot inside Gemini. to have images selected. So I have to still have that images selected in the toolbar, where in Google AI Studio, you don't need that.

So let's see what Google AI Studio finished. Okay, this is pretty good. So I'll have to go through and check for accuracy. But this is really, really good. So what it did is it kind of created, it looks like it kind of took four main takeaways. And it really deconstructed. The entire, this is so good.

So it pulled, you know, it pulled logos, right? Because I was talking about, you know, a Boston consulting group study. It pulled BCGs. It pulled snowflakes logo. It put the actual stacks that I use, although there's a couple of letters that maybe don't look like are 100% correct, even though I know what they're saying. But it made a four-column graphic. It looks professionally designed, maybe not the highest end designer.

But this looks like something you might get with a junior designer if the letters were all correct. It looks actually really good. It's an infographic. There's, you know, kind of some icon, some icons and circles and logos and big pullout stats and little quotes. It looks really, really good, actually. I'm surprised at how good this one turned out and even put the everyday AI logo on there or the website. So that's nice. Let's see the Google Gemini version.

Okay, interesting. This one is actually even more visual. So I'm clicking on it. At the top, it says debunking the 95% of AI pilot failure myth. It put the episode number. It says why this study is flawed, five major red flags. But let's see, it looks like out of the five major red, oh yeah, they're all right there. One, two, three, it looks like they just got one of the number.

is wrong, but they got them all in there. It kind of dissected the viral claim, talked about the stock market drop, and it illustrated that pretty good. And then on the side, it said key lessons for business leaders, dig deeper, go beyond sensationalized headlines, examine data window, understand realistic ROI timelines, examine data window. So it duplicated that headline, validate benchmarks, misspelled benchmarks. Okay, so a couple, a couple misspellings, but overall, y'all, I just, you know, I just

just dumped, you know, with some iteration and saying, hey, change these things or, you know, this is wrong, this isn't. But the other thing, if you have a library of, of visual assets that you've already deployed in your organization, you just upload those and you say, here's the new info and update this, right? And it might get one or two words wrong, but, you know, I know, I know as someone that used to design a lot before Canva, right, bless up Canva. But I remember in Photoshop putting together, you know, flyers or infographics, right? Like like this. And it would take hours, hours, right?

So if you have something that you've already designed professionally, it looks great, you can literally upload that as a template or inspiration, throw a bunch of new document, new information at it. And it will, you know, won't get it 100% right each time. But it's probably going to save you 80% of the time or, uh, you can even export it and then bring it to something. And this is what I, I do right? now bring it to something like Canva because there's a great feature in Canba called their magic

text grab. So let's say you get, you know, one of these, uh, infographics, you're like, oh, this is perfect. You know, we talked about this at our meeting last week. We said it's going to take, you know, $5,000 project and we're going to hire our, you know, contractor, designer or something like that. And you're like, okay, well, I can either just do it in house or give them a very good looking version to start with, you know, so you can export this, uh, from, uh, nanobanana. Bring it into Canba. Use the AI magic text grab. Correct any of those things that are slightly off or any misspellings.

And you're good to go. All right. So demos happen at about a B minus. But that's to be expected. So let's wrap with five other use cases. So I showed you, it's Wednesdays. I showed you ways that I'm going to be using nanobanana. I think the last one, I'm probably going to be using a lot more,

putting in a bunch of podcasts, transcripts, newsletter, having pre-designed assets, upload them as kind of inspiration, and then putting all of our information in those, and then, you know, doing some slight tweaks in Canva. That's how I'm going to be using Dano Banana, but here's some other use cases. So real estate marketing, this is huge, y'all. You can create professional listing cards by placing the same property in different lighting conditions, seasons, or with different furniture arrangements, all while maintaining the exact same architecture details and proportion.

So the same way that you can have character consistency with humans, think about the thing you can do on the inside, right? If you're building out new warehouse plans, whatever it may be, this is great. Product catalog creation. You can transform a single product photo into multiple marketing angles and environments. You know, you can show your furniture in various room settings or place your, you know, your water bottle or whatever thing of a jig your company sells into different lifestyle contexts, etc.

Yeah, this is obviously big for e-commerce. Next, training material visuals. You can convert complex diagrams, workflows, and instructional content into clear professional presentations by combining screenshots, arrows, text overlays, and educational graphics in one seamless image. So yeah, that's another thing that you can do. If you have like an older asset that you need uploaded, you can literally just throw in Canva or just annotate it and being like, change this, make this brighter, you know,

put this new product in here. So you can literally annotate on an old image, put it some additional smaller assets in there and nanobanas going to do it. Right. So, oh, here's this great, amazing, you know, photo that we use. It was the front page of our, you know, magazine, but, you know, it has a bunch of old products or, you know, whatever. You can literally just circle them in red Sharpie and then upload the new products.

and you know, banana, replace it. All right. Interior design mockups. I already kind of talked about this, but specifically for interior design versus real estate. So you can help clients visualize office layouts, retail spaces, or home improvements by digitally placing furniture, changing wall colors, et cetera. This is something I'm not going to be doing this for everyday AI,

but I'm going to be doing this for my home, right? Me and my wife been talking about, oh, you know, where should we hang, you know, these photos, what should we hang, all these things? And a lot of times it's just like, you know, decision overload. It's like, you know, just leads to nothing. So this is going to be great. I'm going to be using that almost immediately in that regard. Then also another fun one that I'll probably be using is instant AB test variations.

So again, if you have a designer, marketer on your team, an old ad that you ran previously, find your best ad, upload that, upload other smaller assets. And you can just generate a bunch of different variations. of winning assets without having to go through and do it manually. All right. I hope that was helpful, y'all. So nano banana is no longer a mystery. It is Google's newest model, Gemini 2.5 flash image.

And like I said, this is a completely off the charts, useful model. If you're looking for one shot easy, jaw-jopping results, you might be able to get those in predetermined scenarios. But this does require working with the model and just getting your reps in. Right. So right now, I don't have enough reps in. So that's why, you know, the demos are a little clunky and I don't even have the best use cases worked out. But I hope throughout this, you did see the potential.

And for people who could watch it. And again, podcast audience, thanks for sticking with me. But make sure you go watch the YouTube version. or go check this out on LinkedIn or on our website at your everyday AI.com. I think you'll be able to see from the results, you know, you're not going to always get to an A plus, right? Something that you'd pay two, three, 20 grand a designer for, you know, or a photographer for. But you can get 80% of the way there in 10% of the time for 1% of the cost. And by 1% of the cost, like 0% of the cost, right?

because like I said, Google AI Studio available for free, you can go and use Nano Banana. So I hope this was helpful. If so, tell someone about it. If you're listening on the podcast, please subscribe and follow the show. I'd appreciate that. Thank you for tuning in. Hope to see you back tomorrow and every day for more everyday AI. Thanks y'all.

Meet Firefly AI Assistant. Now live in Adobe Firefly, the Allman One Creative AI Studio. Just describe what you want to create in your own words and the assistant handles the rest, orchestrating multi-step workflows across Adobe Creative Cloud apps, including Photoshop, Premiere Express, and more in one conversational interface. You direct the outcome while the assistant accelerates execution. Stand control with the ability to step in and refine at any time. See it today at firefly.adobie.com.

And that's a wrap for today's edition of Everyday AI. Thanks for joining us. If you enjoyed this episode, please subscribe and leave us a rating. It helps keep us going. For a little more AI magic, visit Your EverydayAI.com and sign up to our daily newsletter so you don't get left behind. Go break some barriers and we'll see you next time.