Transcription
The new Nano Banana Pro AI image model is here, and it's a game-changer. Whether you need an AI image editor, want to create high-quality shots for your AI films, or tell a story with your consistent AI characters, Nano Banana Pro can do it all. And after playing around with this new update, I think Nano Banana Pro is the best AI image model out there right now. So, here's everything you need to know about the new Nano Banana Pro update.
First, let's see what's changed compared to the old Nano Banana model. I made these images using the same prompts in the old and the new Nano Banana Pro model. One thing that you'll immediately notice is how the visual style of the images have changed for these photos I created of this woman in the plant store. For the older model, it tends to have these softer textures, but in the newer model, the details are much sharper and stronger.
While both the new and old Nano Banana model are really good at following your prompts, here I created an image of a woman where half of her face is disintegrating into dust. And both are able to generate what I'm asking for pretty accurately. But you can see in the updated Pro version, it's just a way more realistic image. For the older Nano Banana model, the face looks much smoother and softer. But in the newer model, you can see all of these little bumps, grains, uh wrinkles, and imperfections on her face, which really make it much more realistic.
It's also much better at rendering realistic fantasy-style images. These are shots of a fantasy character, uh, the Monkey King. And in the older Nano Banana model, it's this like 3D Pixar cartoony style image, even though I specifically prompted for it to generate a photo. But looking at the output of the newer Nano Banana model, there's way more details on the spiders, like the hairy legs, the webs that they're standing on. It's definitely been a big upgrade.
Right now, the new Nano Banana Pro model is available in Google Gemini. So, I'm inside Google Gemini, and you'll need to make sure you're using this thinking with 3 Pro. Obviously, we're going to be creating images. So, let's also select the create image option.
At this point, you can just directly type in the prompt for the image you want to generate, but I really want to test out some of the image editing capabilities, which is what makes Nano Banana so powerful. And so what I'm going to do is actually upload this image file, which was me at this music festival a while ago, and you'll see that there's a lot of people in this image. So let's see if we can remove some of these background people and play around with some of the other image editing capabilities of Nano Banana.
So, first thing, we're just going to drag myself into here. And then inside the prompt, I'm going to tell it to remove all the people in this image besides the man in the white t-shirt and the blue bucket hat. We enter that, and it'll tell you that it's generating an image using Nano Banana Pro. It's going to go through a few different steps and, uh, yeah, it's completely removed everybody else from this image. Really, really impressive.
If we look at the original image, I mean, there's like people's heads bumping up in the corners. There's small people in the background that are kind of hard to see. The fact that it's able to cleanly get rid of everybody in that image besides me is definitely impressive, and something other AI image models are going to struggle a lot with.
Now, if we take a little look closer, it seems to also have edited some of the text, uh, in the background where they removed people. Now, I don't actually remember what the text was here. Right now, it says "the gold bores ball 2011," I think, which is definitely not what it was supposed to say.
So, what I'm going to do is actually try this prompt that says, "Change the text underneath on the bronze tablet underneath the Statue of Liberty to Nano Banana Pro is here." And Nano Banana is able to put that exact text sequence I asked for onto the bronze tablet underneath the Statue of Liberty.
Now, this image is a little bland and depressing. It was kind of a cloudy day when this was taken. So, let's see if we can brighten things up a little bit and make it look like, uh, I was actually enjoying some good weather. So, what I did is I used the prompt "change the lighting of the scene to a bright sunny day and a bright blue sky." And we get these vivid, vibrant colors in the new image. I'm not sure where these shadows suddenly came from. Maybe there are some people who are standing off to the side.
The new Nano Banana model has a very strong understanding of your prompts, and it's really, really good at combining multiple elements from different images together. Here I've got an example where I've got a guy with a haircut, and I want to transplant that onto the image of myself. Uh, maybe I need a new hairstyle. So I'm just going to drag and drop these two reference photos into my Nano Banana prompt bar and then tell it to "change the hairstyle of the man wearing the black and yellow jacket to the hairstyle in image B." And so, uh, I don't actually think that, uh, this one's going to work on me. Um, but it's good to know of what it would actually look like before I go to the barber and make a horrible mistake.
I also experimented with combining multiple characters into the same image. So, here's a reference photo which has myself and also one of my AI characters that I generated. And I combined both of them into the same photo. And Nano Banana Pro is really, really good at keeping the likeness of your face accurate to the reference photos that you used.
Here's another example I tried of using a reference with three different characters. And I'm going to put them, uh, all sitting together inside the same fashion shoot again. And you'll see that the little details that make up the faces of the people in the image are very, very accurate. And also there's, uh, four hands shown in this image. One, two, three, four. And all of them are rendered pretty accurately. And by the way, if you want to combine multiple characters together, the best way to do it is actually to create a collage reference photo where you put all the different pictures of the reference characters together into a single reference sheet. This way, Nano Banana will do the best job of preserving the likeness of your characters.
Now, it does struggle a little bit, uh, with the lighting sometimes. So, here I combine myself and this other character, and the lighting on the two characters don't quite match. For myself, it looks like professional studio lighting. And for the other character, it looks a little bit out of place, like he's a 3D Blender animation.
By the way, I'm going to be sharing a lot more tips later this week about how to use Nano Banana Pro and how to make the absolute most out of it. So, make sure to subscribe if you want to see more of that.
I also tried turning my images into a different art style. So, taking this photo shoot image of me and my AI character, I asked Nano Banana Pro to "turn this image into a cartoon illustration." And here was the resulting image. My chin does look a little bit overexaggerated, but it did change the art style really well. I also tried another variation of this where I turned them into Pixar characters, and I guess it turned the female character into someone from the Avatar movies. So, this one didn't work out exactly, uh, how I expected, but I really like the 3D version of myself, though.
You can also create plenty of images of your consistent characters and scenes for your AI stories. I started by making this image of myself as a monk sitting at a temple. And so I want this to actually be a banana-themed adventure. So I told Nano Banana that there is a bunch of bananas hanging down from the man's staff. And here were the results I got: myself as a monk with a staff full of bananas.
I generated a lot more images inside Nano Banana of this character. So, here's a shot of me on a banana boat exploring the, uh, wilderness. Here's another shot I made of myself holding up this huge banana above my head. Then I created an image of myself studying and sitting inside this banana library.
Now you will notice that the details around the face, especially, uh, as the character gets smaller, get more blurry and less detailed. So, Nano Banana Pro is coming with a 2K and also a 4K resolution version. But right now, uh, if you're generating images in Gemini, if we download it and take a look at the resolution of the images, this one had the resolution of 1392x768. So, this isn't the high-resolution Nano Banana Pro version yet, but that is supposed to be coming soon.
You can also generate mockups of your products using this new version. So, this is a real-life photo of a souvenir I got from my last job. I ended up losing this unfortunately, but I want to see if I can actually generate some product pictures, especially because it's a pretty complex structure. There's all sorts of different colors, shapes, um, these curvy antlers. And so here's a product mockup I created of, uh, him at the beach. And if we take a look at the comparison, it's pretty much 99% of the way there. Some of these little scale textures underneath, um, aren't as clear in the mockup image, but the other parts of it look pretty accurate to me. I mean, it is a super complicated object to get everything correct in one shot.
Here's another mockup I made of this character inside a collector's box. Um, I didn't prompt for any text, but it actually generated text for the collector's box: "Dragon's Horde Collector's Edition."
I also made this image of myself holding the souvenir. Not sure why I'm looking so depressed in this photo. I guess I'm having a bad day. And here's another photo I generated of myself wearing a necklace that has my character in it. Um, it's a little bit too small to actually capture all the details accurately, but hopefully later when we get the higher resolution of Nano Banana Pro, it'll be able to capture all the little details of the figurine correctly.
I also made a guide with 14 tips for how to get the best results out of Nano Banana. Go watch that right now. Hey, hey, hey.