Best AI Image Generator? Gemini 2.5 Flash (Banana Nano) vs ChatGPT-4o

For the longest time now, ChatGpt 4 has been one of the best image generators for content creators on the internet. But Google's new Nano Banana has just released inside of Gemini. And today, we're going to test it to see which image generator is actually the best because I've been seeing some pretty crazy insane things that you can do with Nano Banana. And I'm not sure that you can do those same things with Chat GPT40 image generation. So, let's jump into the video and let's run a couple of prompts and see which one comes out on top. With the first prompt, I'm going to

upload an image of myself and the prompt says, "Create a professional LinkedIn style headsh shot, but change my outfit to a business suit and add a confident, friendly smile. " So, we've added the exact same image and the exact same prompt to chat GPT. And we're going to send it over. And this is what Google Gemini created. Not pretty bad. Not pretty bad at all. clean up my background and make it look like I'm standing in front of a Sears backdrop. And this is one of the places where Nano Banana or Google's image generation supposedly stands out the most because it's able to edit the images so clean that a lot of people are saying that it's going to be a Photoshop killer. I really don't like this. This makes me look like this is my funeral photo or something like this is an obituary. Can we change the background and clothes? so that I look more like a SAS founder. Maybe put a black t-shirt on me or something. And if you look at it, this is pretty darn good. But now, let's head back over to Chat GPT. Look at what Chat GPT did. Like, Chat GPT didn't even try. This is not me. Like, nowhere even near me. Okay

let's upload another photo I took. And this time the prompt is simply transform this image into a black and white cinematic portrait with a motivational vibe and add beads of sweat on my face to give it a determined look. We're going to send that prompt over. Paste the prompt. We're going to upload the same image. Send it over. And so this is an image of me smiling. It's definitely a faithful representation of me. It didn't change the photograph too much. I can tell that it's been altered somewhat slightly. But now we're going to ask it to change my facial expression so that I'm not necessarily smiling, but I'm more tired, exhausted. I'm in the middle of a workout, but I'm working hard and I'm determined to keep going. And once

again, when we come back over to Chat GPT, ChatGpt keeps creating the images of me as a white male. I don't know what's going on. So, I'm going to try something. It's on ChatGpt 5 thinking. I'm going to go to legacy models and I'm going to choose 40. And then what I'm going to do is I'm gonna come back to the top. I'm going to grab this LinkedIn prompt. As a matter of fact, I'm just going to start a brand new conversation just to be on the safe side. No excuses for you, Chat GPT. No excuses. All right, adding the first image again.

Sending it over. And maybe ChatGpt has a problem reading these iPhone files, these HIC files, but Gemini doesn't have a problem reading it. But let's see if selecting the 40 model actually helps the image generation some. We can tell now that chat GPT just cannot work with personal images. And I had a video that I was going to make about two months ago about chat GPT because it would not let me upload images of myself. It would not edit the images of myself because it said that it violated their content policy. And this was around the time that President Trump signed that new law where the Take It Down Act where you can't post certain material about people online that you create with AI and so forth. And oh man, I seen a thumbnail today or yesterday with Elon and Sam Alman that was very suggestive. They might want to take that down. On another note, we can see here that Gemini has altered my photo again, but this time they have made my nose excessively hard.

This next prompt is simply replace the background of this photo with a beach sunset scene and adjust the lighting on my face to match the sunset glow. Seriously doubt that chat GPT is going to do any image generation with my face. Uh I don't know why. Maybe it needs to be a pi png file, but it's just not working. But I'm still going to run the prompts. test them out. At the end, what I'll probably do is convert the images to PGs and then run all the prompts at one time and let you know if it did any better. And I'll show you. All right. And so, here we go with Gemini. Now, it did place me, you know, kind of in the scene. It did adjust the light a little bit, but you can still tell that I'm not there quite. I'm still sitting in the chair. Come on, CH. GPT. Listen, man. Gemini is making chat GPT just look really bad right now. I'm sorry. Like, I don't even want to do anymore. I don't want to do anymore because it's obvious which one is the photographers's AI. Like, Gemini is the one that you're going to use if you want to edit an image in a realistic or a creative way. Chat GPT is just doing what it wants to do. Like, who is that? That is not me. So, I went ahead and I converted the images to ping files. And you can see the preview now. So, I'm going to run the LinkedIn prompt again. And let's see if chat GBT actually takes the image of me and modifies it to create the LinkedIn headsh shot. And if it does, I'll follow up with the same prompts that I use inside of Gemini to see if chat GPT can actually keep pace because maybe it's the photo uh file type. Okay. I wasn't able to generate the updated image for your LinkedIn headsh shot due to an internal error in the image generation tool. I haven't attempted to retry it yet. If you still like to proceed, just let me know and I'll generate it again based on your latest request. So, this is what I'm going to do. I'm going to paste the prompt. I'm going to copy the image. I'm going to paste run the prompt. Okay. So yeah, chat GPT tried, but this is not the same as this. So when it comes to consistent character, chat GPT is just out of there. It's going to make a general image of somebody. It is not going to make the image of you. I don't even want to test chat GPT anymore. But there are some other things that we want to test it with. So I'm going to start a new

convo, and I'm just going to paste in this prompt. And it's basically creating a pink elephant fighting a superhero. And I'm going to do the same thing over here in Google because I don't want to just test images of people. We know that chat GPT cannot keep pace with Google's new image editor. It just can't. Nowhere even close. It used to be able to do it, but they rolled it back for whatever reason. Google is going all in with it. But now I want to test each of them with basic image generation. And so when it comes to a typical image generation, you can see the image here of the pink elephant with the superhero. And surprisingly, this looks how it did the last time I actually ran this. So, it's kind of consistent. And as much as they scream about content policies, I'm very surprised to see that this superhero has what looks like the Superman emblem on his chest. And this guy is very much built like Superman. the blue, the red waistband, the boots, the cape, and you can even see some of the yellow kind of muddied in there, the dark hair. They are absolutely drawing off of Superman. Whereas this superhero, even though he has the same red boots, blue pants, he has red trunks, red cape. The mask on his head and the way the cape kind of comes over on his chest and the gloves help you realize that this is not Superman, whereas this guy is blue straight down. This guy looks like Superman. This is an image of Time Square on New Year's Eve that is AI image generated. Let's get the prompt for it and see how well Gemini does. And so you can see here that Google's image of New York Times New Year's Eve is a lot more realistic, I would say. Like you can look at the smoke, the crowds. We're not zooming in, of course. You can look at the detail here with the people with cameras. You see the stage, somebody performing right here. and just the depth of field with the brightness in the air. This looks very realistic, more so than chat GPTs. The next prompt was about showcasing America's most breathtaking coastal destinations. And so it has these right here. You have Destin, Big Su, Kauaii Island, I think Texas, South Carolina, South Padre Island, Destin. Now, one thing I will say, I'm looking at the image and the image looks good, but one thing I'm going to say that I have noticed as an improvement with this image generator, with Google's image generator, is the ability to spell these words accurately. It was not doing this the last time I tested this on August 20th. I just ran the prompt again with chat GPT, and this time it did not perform well at all. Oh, we can see that the elephant even got blended into this photo. And the elephant is nowhere in this prompt. The elephant is just in this conversation.

So, this last prompt is titled the four panel comic layout. Let's just scroll down and look at what we got. We got basketball in three of the images. There are no touchdowns being scored here. And if you look at the letters, it's u when the your fif. So, it's some other type of language. It's just kind of putting words together, right? And it did this exact same thing the last time I did this. So, now let's go over to chat GPT. And I'm pretty sure that chat GPT is actually going to have the writing better. And it does. Ace throws for six, right? Whoosh. Pick six. And then, right, uh, some of us are competitors, entertainers and spelled entertainers wrong. I'm here to compete. Old Dylan Gabriel. Right. So anyways, chat GPT is definitely better when it comes to something where the text has to be right. And it seems as though Gemini still struggles with writing basic English words. Even though it got Destin, Big Su, Texas, Kauaii Island, South Carolina, South Padre Island correct, I think because Destin doesn't have Florida, big su doesn't have California right here. So, aside from

these strangers here that I do not know, chat GBT is still the best when it comes to creating infographics like this, I want to try one more prompt with Gemini. And I'm going to upload a photo of me. Turn this into a Lego version of me with gold chains and shades and dress my Lego figure inside a run DMC outfit to make me look like an 80s rapper with Adidas on. I want to see how creative you can actually get with Google Gemini's image generator. But look at this. Look at how Gemini was able to take a photo of me and turn it into a Lego version of me with the Run DMC, the Adidas. This is exactly what I had in my mind. But let me know what you guys think about these two image generators. I personally am looking forward to using Google's imagine for a lot more because it's just better. It is just better. Unless I'm creating something that has words on it like an infographic or something like that, then I'll use chat GPT40 image generation. But that's a thing of the past now. And there are so many different updates that Google is going to be rolling out that it won't be long before Google will actually be a better AI product, a better choice for a subscription than it will be to pay $20 to Open AI. I'm just saying. But let me know in the comments which one do you like? Which one are you going to stick with? And speaking of Google becoming better than OpenAI, one of Google's products that I believe is better than ChatGpt in a lot of use cases right now is Notebook LM. And if you want to watch a video to see how I'm using Notebook LM to do next level research and get insights that you can't get from chat GPT, then make sure you check out this video right Here.