OpenAI Image 2.0 vs Nano Banana Pro - TESTED

OpenAI just released Image 2.0 and we're going to compare it to two other models which is going to be Gemini's Nano Banana Pro along with their flash version of it as well. And this is going to be in the context of a real app, the one I've been working on the last month. It's a followalong course. I the project I'm going to show you exactly how to build it and I'm going to release it as an actual SAS. So a realworld demonstration. Let's get started. Okay, so we have eight different examples and I'll show you the original image of the landscape that we uploaded and then I'm going to show you the three variations. All right, so here's the back of my house. It's the picture I took. And for this first prompt, this one's going to be open AI's image 2.0. And I'm going to show you the prompt and what it changed. Okay, so add a row of these juniper bushes. So basically my system allows people to you know add these different assets in with the context and and add it in crushed granite. So the the actual prompt is add a row of these juniper bushes to the back side of the house. They should be planted in crushed granite which also lines the back of the house. Don't remove or change the wooden enclosure. So this is the line of bushes in the actual crushed granite. And it did an excellent job. It's exactly what I imagined. This is what it looks like before. Right here there's nothing. And then here's what it did after. All right. So now let's go ahead and check the Nano Banana Pro. It's not quite as good. It seems a little rough here. Seems a little bright. This looks way more realistic, which is OpenAI's new model. This one doesn't so much. Here is Flash. All right. So this one, in my opinion, Flash right here. These are both from Google. I this one did a better job even though this this bed is way too like far out. That kind of doesn't make sense. But the overall winner here is going to be this one which is image 2.0. I will say this one takes image 2 point image 2.0 from open AAI takes long time. It takes about five times as long as the flash version in Gemini and probably about at least twice as long as Nano Banana Pro 2 Pro. So that's the first example. Let's check out another one. Um the same the next example is going to be about removing this tree. This tree right here is complex because it it's all over the place. It has branches all over the place. And I asked it to remove the tree and change nothing else. All right. So here is OpenAI's image 2.0. Tree is removed successfully. Looks pretty good kind of although it changed this to windows. I think the garage and again that that was area was really covered up. Here's image 2.0 know or Nano Banana Pro rather. It did a better job. It still has the car in there. Uh this time it actually beat it. And then here is Flash version. Flash did a very good job as well. I And guess what? Or yeah, Nano Banana 2. This is actually Gemini 3.1 Flash. This did a very good job as well. So in this case, it didn't do as quite as a good job as the other two. So in this case, image 2.0 failed. All right, here's an entirely new example. Um, right here, just some ugly bushes. And I asked it to replace the bushes with a specific type of tree or plant. And also remove this white gutter. All right. So, the plant is a Frasier fur. So, I could just type that in. Fraser fur. And you can see what it looks like right here. And this is what it did. It didn't replace this dead bush. And that's okay. The grass is also improved as you can see. Very good job. Now, here is Nanobanana Pro. This time it replaced all of them and it added four a little bit close to the house. Got rid of the the gutter. And then here's Nano Banana the uh Gemini Flash version. And again, it did a pretty good job as well. You'll notice that um the Gemini models are a little bit brighter and this in my opinion it's more realistic photorealistic. This is like it looks like it's just like a little bit too polished in terms of lighting. All right, let's check out the next one. This is the front of my house and this is when it was in winter. This one I have these presets and I chose contemporary lux for it. We'll see what these do. All right, so the first one's um OpenAI's new model and again it really makes it contemporary there. I also specified in this uh nighttime, fall and rainy that's why it's dark and and considerably different. So, if we compare these two right here, get the hell out of my way. This is what it transformed. Pretty good job, you know, based on what I want. That's That would be very expensive, though, because I did ask, do not go overboard with an extremely expensive refactor, but it looks like the system prompt with Contemporary Lux is kind of fighting that a little bit. Um, let's check out Nano Banana 2 Pro. All right, good job here. I I like it a little bit better. It's a little bit more realistic. is a little bit more tamed, even though we do have like this infinity pond thing there. Uh, and then here is flash. Again, I would say the the Gemini Flash model, while very fast, it's not quite as realistic. This kind of looks almost fake here, but still, these aren't, you know, not too far apart from each other. All right, here's a picture here back of my sister's house, actually. And, uh, let's see what we changed. Oh, okay. This one's cool. So basically if I compare these, this is image 2.0 from OpenAI. All I did is I asked it to put in this metal fire pit. Now if I I type that in fire. This is the actual fire pit I asked it to replace or add in. And it did a very good job of creating it with fire. And I also asked it to put in flag stone. So here's the prompt. Uh I'm not going to read it, but it it did a pretty good job. I and I think it also has contemporary lux which is why you have all the the um the lighting here. So multiple presets. I think it did a great job. Let's check out Nano Banana Pro. Nano Banana. This is like one of the worst ones. Check this out. First of all, it's not the same uh fire pit and these people are clearly 3D. I And yeah, because we have the contemporary Lux added this. That's ridiculous. This is terrible. Absolutely terrible. the scale's wrong. Like this looks like it could be it could be a real image and then this is terrible. Now what's interesting is Pro did a way worse job than Flash. Flash actually has real people and but it did not use anything similar to our flag stone or that. So the when it comes to attaching additional images, I sometimes it will just come up with its own and will not adhere to it. Image 2.0 know adheres to the actual image assets that you pass in with the context. So that was way better. Let's check out the next one. All right. So the next one I also had to integrate a predefined water fountain right here and also a predefined I think it's called a pergolo pergola or something. It's a wooden structure. You'll see. All right. So here's image 2.0. Yep. And you can see here it is. I this if I type in PR real quick, you will see that this is the structure and it's really integrated like great. And then same thing with the waterfall feature. If I type in water, here it is. And there it is here. Perfectly integrated. I like it. It's also um evening, fall, and overcast. So that's what we changed it to. I'll show you the before and after right here between these. Uh very nice. I think it did a great job. It's kind of what I would imagine it to be based on my request. Um, here is Nano Banana Pro. Again, it's not adhering exactly to, you know, the asset we included. And also, the waterfall feature is completely kind of ignored. Um, it has its own integration, but again, it's not adhering to the asset that I specified. Here's uh the Flash version, Gemini Flash. Again, not the same one. roughly the same one. So, a little bit more hallucinations with the Gemini models. All right, here's another one. This one I simply asked it to keep the house, keep the fence, but remove all the uh the palm trees and the vegetation and then put in a small swimming pool. So, let's check out our OpenAI image 2. And this is, you know, did a pretty good job. It kept these little things right here. Not bad overall. Uh, also it's morning time, spring, and partly cloudy. Yeah, good job. Very simple. And then let's check out the other one. Now, this one is Nano Banana uh 2 Pro or Nano Banana Pro. And um you can see it has a small swimming pool as well. It did a pretty good job here. Um this is a little bit more realistic setting. This seems a little bit staged. Uh and then here's the flash version. This is not a small swimming pool. It's actually pretty large. Um, but again, you're getting way quicker results and it's way cheaper as well. Now, also about cost, um, all these, uh, right here on Open Image 2.0, they actually have a low, medium, and high setting. These are high settings. Um, but even with Nano Banana Pro, which is also more expensive, this is more expensive slightly, uh, image 2.0 for the high setting versus Nano Banana 2 Pro. Um, and then of course this is way cheaper. Okay, let's check out this next one. Um, this one is super interesting. So, basically I used one of my presets and I used the lowmaintenance evergreen preset. All right, so year round structure, evergreen shrubs, mulch beds, minimal plumbing, uh, automated irrigation. Okay, so that that's what I asked to do. I also asked it to make it like rainy and overcast. So, here is the OpenAI image 2.0. Notice it it changed it to shrubs. It's raining. You can see there. Yeah. Did a great job. Now, let's check out the other one. H. Okay. So, this is Nano Banana Pro. It This is not low maintenance. This did a terrible job. I Flash did a better job. Gemini Flash. At least it didn't replace the yard with more like, you know, upkeep. But notice brighter. Both are brighter. This is more realistic. It's definitely better. You can even see like the water raindrops coming through. And that is it. So, take from that what you will. There was just eight examples here. I know for me personally, you know, building out this app, I'll probably stick with maybe a fast and a slow version. like a fast cheaper option which could be flash and then stick with open AI's image 2.0 know as you know the slower but better result but also more expensive result as well an option to use and so that's how you would you know integrate something like this into a project where you know you're using image generative models same thing applies to video as well and so yeah take from that what you will hopefully you found that interesting I know that I found it interesting and now I have a couple options now um for this landscaping app for people to use so if you want to learn how to build this app from scratch. Definitely take take a look at the uh the early access course. You can take it now for a discounted price. And I will see you all very soon.