Welcome to the ultimate showdown between ChatGPT and Nano Banana! 🌟 In this battle of intelligence and innovation, you’ll discover which one truly comes out on top! Dive into an engaging comparison that uncovers the strengths, capabilities, and unique features of both contenders. 💪✨
This video is perfect for anyone curious about AI technology, its applications, and the future of intelligent systems. Don't miss out—watch until the end to see who reigns supreme! 🔍🚀
👉 Discover more about the power of AI with Nano Banana:
Registration Link: [https://Tava.short.gy/aibanana]
If you enjoy this content, please subscribe, like, and leave a comment with your thoughts on the showdown! Your support means the world to us! 🙌💖
#AI #ChatGPT #NanoBanana #Innovation #Technology #Comparison #FutureOfAI #TechShowdown
This video is perfect for anyone curious about AI technology, its applications, and the future of intelligent systems. Don't miss out—watch until the end to see who reigns supreme! 🔍🚀
👉 Discover more about the power of AI with Nano Banana:
Registration Link: [https://Tava.short.gy/aibanana]
If you enjoy this content, please subscribe, like, and leave a comment with your thoughts on the showdown! Your support means the world to us! 🙌💖
#AI #ChatGPT #NanoBanana #Innovation #Technology #Comparison #FutureOfAI #TechShowdown
Category
📚
LearningTranscript
00:00OpenAI just dropped a brand new image model and it's called OpenAI ChatGPT Images 2.
00:06Great name. But it is coming for the crown that Nano Banana 2 and Nano Banana Pro have had for
00:12months now. And in this video, we're going to go side by side, compare both models with the exact
00:17same prompts and do a blind test. By the end of this video, you will know which model is better,
00:22Nano Banana 2 or ChatGPT Images 2. Let's get right into it. All right, now before we get into
00:29the blind test side by side, it's got seven rounds. I want to give you a little bit of
00:32context on GPT Images 2. This came out yesterday, so on April 21st. And from everything that I've
00:40seen so far, it looks really, really impressive. So I haven't done a ton of testing yet. I've done
00:44a little bit here and there. But today, I really want to see, is this better than Nano Banana Pro
00:48and Nano Banana 2? Because ultimately, that's what we've been using at Ad Class. We spend a ton on
00:53paid advertising. We use a lot of AI image generation. And basically, what I want to know
00:59is, is this the new king, right? And so that's what we're going to be doing today. So what we
01:03made here is a little Claude project, right? And so Claude wrote all the prompts for these,
01:08you guys are going to see that I can show you the prompts that we use in each one, right?
01:12And so the
01:12prompt is up here. And then the actual image generation that I got from using that prompt for
01:18each one is down here. Now, these are blind. I don't know which is which. My team helped me create
01:22this. So I'm just going to be voting in real time on this video, which one I think is more
01:26impressive
01:27based on what the prompt was. So this first prompt, I'll read it for you. Ultra realistic
01:30lifestyle product photo, matte black ceramic coffee mug with the words morning fuel embossed on the side,
01:36sitting on a reclaimed wood desk next to an open MacBook, soft golden hour window, light from the left,
01:41shallow depth of field, steam rising from the mug. So we have two options here. We've got side one,
01:48which pretty good. I mean, I want to zoom in a little bit here on it.
01:52I'm not seeing many issues and the keyboard is a little bit blurred out. So I can't tell
01:56exactly what the keys are saying, but it does look pretty accurate for me. The actual cup itself in
02:00the wood looks real. I mean, like very hard to distinguish that from anything fake. So overall,
02:06I'm pretty impressed with this image. Image number two here. Let's see.
02:13I mean, yeah, overall, this also looks pretty good. I'm not noticing anything like out of the
02:20ordinary. It looks pretty accurate. I mean, both of these are really good. Honestly, like I'm gonna
02:27have to pick on which I'm more drawn to in from like the sort of like depth of field and
02:33blur.
02:33I kind of like side a better. I don't know why, but I'm going to give it to side a.
02:39Okay. So let's see.
02:41Reveal side a chat GPT images too. Okay. So side B is nano banana too.
02:47Um, so you can see here with my, move my face here for you guys. You can see at the
02:51top right here,
02:52nano banana two versus chat GPT images too. So I'm giving it to side a. So, so a chat GPT,
02:58you've got the first round. Let's go on to round two. Okay. So this is a text heavy poster.
03:05We've got the same text. So let's show the prompt bold, modern event poster. Uh, I changed this to
03:12be one by one. So headline at the top and large condensed font, uh, AI media buying summit sub
03:18headline 2026 playbook for performance marketers. Okay. And it's telling, telling it to do hosted by
03:23Dakota Robertson and 12 guest operators. Don't know who that is, but let's go. Okay. So we've
03:29got side a here. All the texts looks good. I don't see any issues in the text. So nano banana
03:34two was
03:35already pretty good at text. I was impressed with it when it came out for sure. I don't see a
03:40ton
03:40of issues with the text in the banana two, unless you really get my new with it, which we will
03:43do in
03:44this video. Um, so really I'm just looking for typos or anything like that. Both look good.
03:50Um, then it's just like styling, which one catches my eye more. Right. I mean like,
03:54ultimately both are decent. Um, honestly, I'm, I'm going side B. I just like the way the font
04:01works a little bit more. I think it's, I think it catches my eye a little bit more. I think
04:05it
04:05looks more like a designer made it, whereas this one looks a little bit more just like
04:09AI made it. Honestly, it just looks like text kind of plopped in there. Um, so I'm gonna go side
04:14B.
04:15There you go. Uh, chat TPT images 2.0. There you go. Interesting. So this is nano banana two.
04:22So once again, chat TPT gets this round. Oh, here we go. All right. We're onto complex multi-subject
04:29scene. So this one, uh, this is interesting. This is, uh, basically a bunch of things happening here.
04:35Let's look at the prompt. We've got three media buyers sitting at a long shared desk in sunlit office
04:41viewed from straight on. Okay. Wait. So already viewed from straight on in the prompt. You can
04:47see one of these is viewed from straight on. And one of these is not interesting from left to right.
04:53So on the left, there is a woman in her late twenties with short pink hair, wearing a cream
04:58cable knit sweater. Okay. We have that in both holding a coffee mug. Yep. Holding a coffee mug.
05:04Um, looking down at a laptop screen, showing a red downward chart. Ooh, look at this. Hold up.
05:11She is holding a coffee mug here, but she's also, I don't know if you can see that there's like
05:18a coffee
05:19mug artifact on her sweater. And then over here, she's also got a coffee mug. And then on the back
05:24of her screen, there's a downward chart. Okay. So that one, not, not so great. Not so great.
05:30Um, I should say too, I only, I only let one generation for each of these, right? So we didn't
05:36do multiple generations and pick the best. We just did one shot. That's it. Uh, so that's kind of
05:40interesting on this side, we do have the girl. And then we also have a screen kind of facing us
05:44with the downward chart, which honestly is a little bit better. Um, a man in his thirties with a buzz
05:50cut
05:50and a beard. Okay. Good on both of those wearing a black t-shirt that reads CBO only. Yep. Good.
05:56They're laughing while pointing at the third person's screen. Okay. Yeah. I would say on both
06:01of those. And the third person is a woman in her early forties with long curly black hair,
06:04wearing round glasses and a mustard yellow blazer, leaning back in her chair with arms
06:08crossed, looking unimpressed. Yeah. I'd say it nailed it on both of these Florida ceiling
06:13windows behind them, soft documentary lighting photo realistic. Okay. I will say both are pretty
06:20good, but because of the weird artifact stuff on this one, I don't think I can give it to this
06:25one,
06:25even though it did have the prompt of like being straight on better. This one seems to be just,
06:31uh, there's just not any weird artifacts on this one. I, you know, I can't see anything
06:35on first glance that looks fake. Um, it doesn't look as good lighting wise. That's for sure. Like,
06:41I like the lighting and styling of this one more, but I have to give it to this one.
06:46So reveal set a, Hey, let's go. This is nano banana too. So Google gets this, this round.
06:53Okay. Don't call it a comeback. We're two to one. Let's go to round four.
06:56This is an infographic. It's supposed to have real data, but the prompt
07:00just says clean modern infographic title at top top five largest U S cities by population
07:05below the title, a horizontal bar chart with five bars representing the top five most populated U S cities,
07:10each bar labeled with the city name on the left in the population number on the right bars and
07:17gradient from deep Navy to teal white background, generous white space, uh, geometric sans serif type.
07:24Cool. Well, first off, the thing I'm noticing when I'm looking at both of these is that the numbers
07:29are vastly different. So my question is, where did the numbers come from? The second thing I'm noticing
07:33is that the gradients are different, right? So this, this one has the gradients in the bars.
07:37This one is like gradients from each one. Um, and then the third thing is, is mostly just that
07:42there's a little bit of a weird artifact here. Like this text is not real legible. Um, just based
07:48on population numbers, let's do a quick check. Let's look at LA. Uh, let's look at LA population
07:57population, oops, population that way. 3.88 million. Okay. So this one's accurate, uh, compared to this
08:07one, Chicago, let's look at Chicago population of Chicago 2.72 2.72. Okay. Hold up. So this actually
08:16has accurate data from this prompt, by the way. So this is doing some like Googling or something.
08:22It's, it's got data on the backend. That's actually being used to create the image. That's kind of crazy.
08:27And it doesn't have any artifacts. So this is the easy winner. And there you go. This is chat GPT
08:31images,
08:32too. Wow. Okay. That's kind of crazy. Actually. If you think about it, it used like, all I did was
08:39give it a prompt and said, pull, like, make this image. And it actually made it with accurate data,
08:45which is the first time that I've actually been a little bit mind blown by image generation prompts.
08:51Okay. Crazy. Let's go on. We're moving on. We're moving on. Round five. Oh, oh, what's going on here?
08:58There's a little, we got a little steering wheel in the back, a little steering wheel in the back seat.
09:02Let's go. Um, okay. Prompt iPhone front camera, selfie style photo, 32 year old man with five
09:10o'clock shadow wearing a Heather gray hoodie sitting in driver's seat of a parked car. Okay. First off,
09:16he's not in the driver's seat over here and there's a steering wheel in the back. So like,
09:19obviously this, this one's going to win the rain, the round slightly off center, natural midday lighting,
09:24holding a phone with his left hand mouth, slightly open mid sentence. Like he's recording a video,
09:29slight motion blur. I mean, this one is just a lot better. Like the graininess of this doesn't
09:36really work. And obviously the artifacts are weird and he's in the wrong seat. This one kind of nails
09:42all of it. I don't see any real issues. Honestly, it looks really good. Okay. Let's get, let's,
09:50oh, there you go. Chatty images too. There you go. So it wins another one. We're at four to one
09:56going into round six. Okay. For this one, this is a character consistency test. So for this one,
10:02what we did was we had to make a first prompt a, which is a photo realistic portrait of a
10:0628 year
10:06old woman. So that's what I did first. And then prompt B is the exact same woman using the reference
10:11image from prompt a with a new sort of prompt. And so let me actually pull up the two images
10:18of the
10:19women. So you can see those. Okay. So there's one of the women and then the next one here. Okay.
10:29So
10:29you can kind of see, you know, and each one was made by the respective, the respective one. So, um,
10:35we can see basically woman one here, and then we used her for consistency in that same model.
10:41So you can see, let's, let's kind of double check the consistency, right? Um, honestly,
10:48pretty good. Same earrings,
10:52same eyes for the most part. It looks, it looks really similar. It looks,
10:56looks pretty much like the same person. I'm trying to see if I can find any differences. Honestly,
11:02the freckles are slightly different.
11:04Like if you kind of look at the freckling on the forehead and on the nose,
11:09it is slightly different. There's more freckles in this one. And then also her teeth are a little
11:13bit different. There's not by much though, honestly. So, okay. That's one of them. And then
11:18there's this one and she is this, this, uh, girl, I have a hard time calling, calling it a girl.
11:24It's a
11:24fake person, but you get it. Um, okay. So this, this is kind of a similar, similar look,
11:30similar haircut. The haircut is exactly the same. And the, honestly, let's see.
11:37Yeah. It's, it looks like the same person. I mean, I don't know. I'm in my head on this
11:45one a little bit. Okay. So I think the character consistency is really good. So let's just look at
11:50the prompt and see if we can find anything. Um, show her from the waist up sitting in an outdoor
11:54coffee shop table, holding a latte with both hands, golden afternoon light, blurred city street
11:59behind her. Same hair, same freckles, same earring, same blazer. Huh? This one's tough.
12:09The only thing is that I just feel like this girl looks more real, but I'm getting, I'm,
12:16I'm being pretty picky. Yeah. I really don't know. Like I like this image more, but it looks more like
12:21it's AI to me. So I'm going to give it to this one. Hey, there you go. That's nano banana
12:25too.
12:26Okay. Four to two, four to two. So the game is over already. We, we already know chat GPT one,
12:32but I have one more, one more image. Okay. And so this is a similar thing where I actually made
12:36a fake product. Okay. So the fake product is this ad class greens, your daily green simplified.
12:43Um, now I did use, I'm sorry to report. I did use a chat GPT images too for this. Okay.
12:49So just,
12:49just know that that's how I made this image, but, um, I made this one and I wanted to see,
12:55can, uh,
12:57can we keep a product really consistent between these two models? So I gave it that product.
13:01And I said, using the uploaded product as a photo, generate a scroll, stopping e-commerce creative
13:05for meta place, the exact same product, uh, into a vibrant lifestyle steam lifestyle scene,
13:11morning kitchen counter, soft, natural window light from the left glass of water. You get the idea.
13:16Okay. So let's see which did better. We've got this one here, 30 super foods, one scoop.
13:22I want to look really closely at this. It all looks pretty good. There's, I can be a little bit
13:28picky with the fact that I can't read what's on this little label and it's supposed to say natural
13:32lemon lime flavor. Okay. But again, this is like, it's very picky, but you know, that, that is a note.
13:39The glass looks good. The greens look good. Everything else looks good. Let's look at this one.
13:4530 super foods, one scoop also looks pretty good. Um, but I can't even read this. I can't read that
13:54one there. Um, and I also can't really read that. So both of them kind of a fail here in
14:01my opinion,
14:02but again, I'm being very nitpicky. Um, I can at least on this one, I can at least read supports
14:08energy, supports immunity, supports digestion, supports overall wellness, which matches up with
14:12these, um, pretty well. Yeah. I think I'm going to give it to this one just because I like the
14:20font
14:20and the actual like image better. I think it looks better compared to this one, but yeah,
14:26it's a tough one. Okay. Chat GPT. There you go. So have it chat GPT wins five to two on
14:35the image tests.
14:38Okay. Kind of crazy. I mean, overall I'm impressed with both. I still really like nano banana.
14:43If you're wondering like my workflow for this, I'm using Higgs field. Higgs field is a great, uh,
14:47app because you can choose all of the different image models. So you can go in there and just
14:51split test these. That's what I've been doing. I'm basically anything that I might make in nano
14:55banana pro or nano banana two. I'm just testing in, uh, chat GPT images too. Now from this,
15:01it's pretty clear that chat GPT 2.0 is beating nano banana two, and we might have a new king
15:06at least
15:07until Google drops the next update. Hopefully this has been helpful for you. Comment below
15:12if you have any questions, uh, and I look forward to seeing what you guys built. Peace.
Comments