Nano Banana 2 is Here! Full Review & Testing Results

0:00 So, we had another big announcement out of Google.

0:02 They upgraded their best image model.

0:05 And here's the thing.

0:06 Many people already considered their previous image model,

0:09 Nano Banana 1 and Nano Banana 1 Pro, to be the best in the industry.

0:13 Sure, some opinions say that maybe Mourney is the more tasteful model,

0:18 but that's pretty much it.

0:18 Other than that, people really love Nano Banana 1 and Nano Banana 1 Pro,

0:22 and now we have Nano Banana 2.

0:24 So, I thought this would be worth a separate video where

0:26 we compare the outputs to all other popular image models out there.

0:29 We verify a bunch of the very impressive examples they showed

0:32 off on their launch blog post and we run the prompts ourselves.

0:35 And three, we show you a bunch of use cases for useful

0:37 stuff that the previous model wasn't able to do as well

0:40 as this one in a little live section with a bunch

0:43 of prompts that you can take away and use yourself after this video.

0:46 Lot of ground to cover.

0:47 And honestly, I like making these image generation videos cuz they're so visual

0:50 and it's so easy to see AI's progress cuz it just looks better

0:54 and works better rather than LLMs where a lot of times it's kind

0:57 of hard to pinpoint which part of the text is now better or different.

1:00 Let's do this.

1:01 So, here we are.

1:02 If you're familiar with the channel,

1:03 we have a few prompts in different image generation categories

1:05 and we just compare it to some of the most popular models.

1:08 So here on the left side you see the nano banana 2 output to this logo

1:11 design prompt and then the right side now you see m journey you see

1:14 the grock image model you see the chat GPT image model there's flux 2 and then

1:19 finally nano banana pro which was the previous

1:22 best version if you want the full sheet

1:23 with alo all the Chinese models and previous

1:25 versions we'll link it below I would say

1:27 on logo generation it's not special that's not

1:29 the biggest strength of this model but now we go to portrait photography and I

1:33 think here it actually really shines you know

1:34 we had really realistic models before even if I look at nano banana on a pro.

1:38 It looked really good.

1:38 To me, this looks even better.

1:40 Maybe it's kind of the textures,

1:41 the pores on the man's face, or it's the posing.

1:44 I don't know.

1:45 It just looks perfect to me.

1:47 Next one is a cinematic still,

1:48 which always is an interesting one cuz there's a lot

1:50 of variety in what you get from the different models.

1:52 I like it.

1:53 It's a hyper realistic movie scene.

1:55 This could be a frame from a movie, and it's not that stylized.

1:58 You know, when I look at some of the other ones,

2:00 including Nano Banana Pro, it uses a lot of saturation and a lot of style.

2:03 And this one is like rather neutral, just real.

2:06 And I think that's a thing that remains true across the models testing.

2:09 Aerial photography kind of all outputs

2:11 look very similar from the various models.

2:13 So this one is not that exciting, but yeah, it looks good.

2:15 It looks very photorealistic.

2:16 It's incredible at that.

2:17 And then when it comes to book cover, also a lot of these look similar lately.

2:21 I noticed it's really good at graphic design and book covers.

2:24 I would say this one is actually maybe

2:25 a little better than the previous Nano Banana.

2:28 It just looks a bit more aesthetic.

2:29 The whole composition and the way the colors match.

2:32 And then lastly, this one was interesting.

2:33 We asked for a comic strip illustration.

2:35 and it actually refused cuz it figured

2:36 like hey maybe there's some copyright content.

2:38 So that's the thing with this model it's quite

2:40 restricted in terms of what you're allowed to do.

2:42 Often times you'll also upload like image of yourself

2:44 you want to remix it and it'll be like

2:45 no I can't because like you're remixing a celebrity

2:48 and now just cuz you're thinking like okay yeah Eigor

2:50 you do YouTube so maybe your face is out there.

2:52 No, this happens for various people and the guidelines are

2:55 just so strict that it just cuts you off sometimes.

2:57 Often the key to resolving that is just running it

2:59 a few more times and sort of it just works.

3:01 Anyway, so that's our little comparison there.

3:03 I think except of maybe logo design.

3:05 Uh to me it's a no-brainer to just make this my go-to model

3:09 for whenever I need image generation stuff just cuz it looks so real.

3:12 It's so precise.

3:13 But now let's look at some of the examples from their blog because

3:15 they led with various characteristics that were

3:18 really beyond anything we've seen so far.

3:20 Like not a head and shoulders above anything else, but like further.

3:23 So here's one of the examples they shared in their blog post.

3:26 Like this beautiful top-down infographic but in this like handmade style.

3:32 What would you call this?

3:33 There's like a word for this um what's it called?

3:36 Various stuff like a collage scrapbooking.

3:39 That was it.

3:40 Yeah.

3:40 Yeah.

3:40 Oh my god.

3:41 Nice.

3:42 Human one chat GPT0 like a scrapbooking style.

3:46 I think this is gorgeous.

3:47 We ran the same prompt they shared in the blog

3:49 post just to check how much they cherrypicked the results.

3:52 And this is what we got.

3:53 It's still good but it's not that good.

3:54 So yeah, as per usual in the blog post, you know,

3:57 they run it a lot of times and they pick the best one.

3:59 But when it comes to like the accuracy of all the elements, like it's all there.

4:02 I guess you could be more specific on asking for this scrapbooking style.

4:06 I just really wanted to show this too because I think this is a very

4:08 creative way to add like images or B-roll

4:10 to your video content or social media content.

4:13 Like people would not associate this with AI generated images, right?

4:17 Okay.

4:17 The next one is this infographic that is more in this cartoony style.

4:20 It's an overview of three different types of clouds.

4:22 We ran the same prompt that they shared and we got this way

4:25 more details and I have to say I'm kind of impressed by the text.

4:28 It's usually like you read through it and you find a letter off here and there.

4:32 Text got so much better over the last year,

4:34 but here I kind of have a hard time finding a mistake.

4:36 I think all of this text is just flawless.

4:39 Correct me if I'm wrong, though.

4:40 Even the numbers, it's just good to me.

4:42 It seems like it went from like 95%

4:44 reliability to at least very close to 100, right?

4:47 Okay, here's the next one where they compared Nano Banana to Nano Banana Pro.

4:50 Nice.

4:51 We ran the same prompt and got this.

4:53 I personally like this one more,

4:54 but it is a different style, but it is beautiful.

4:56 like the amount of details and the composition and everything.

4:59 Like heck, if you're going to do something in this direction

5:01 of like infographicy or like handmade

5:04 scrapbooky like this model is just incredible.

5:07 And that's also what they claim.

5:08 They say that the visual fidelity has been upgraded and I honestly see it.

5:13 They claim to have more vibrant lighting, richer textures, sharper details,

5:16 maintaining high quality aesthetics at the speed

5:19 expected from flash, their quick model.

5:21 A lot of that is subjective.

5:22 You can make up your own mind based on examples I showed you.

5:24 And hey, by the way, if you're enjoying this video,

5:26 make sure to subscribe to the channel.

5:27 We make deep dive tutorials and news

5:29 updates and short tutorials every single week.

5:32 And now, let's continue with the video.

5:34 Next, let's look at the stuff that is not so subjective.

5:37 Namely, the upgrades to subject consistency

5:39 and how precisely it follows instructions.

5:42 This is the one that is really exciting to me cuz some image models,

5:44 especially chat GPT, you tell it certain things and it just ignores them.

5:48 for three months and shop for two months.

5:50 You know,

5:52 well, I have a few prompts here that I want to test with you live to both give

5:56 you inspiration for what can be done and also

5:58 test if all of these upgrades are actually present.

6:00 So, let me just go into my Gemini account

6:02 and click the create image button here at the bottom.

6:06 And yeah, this automatically has Nana Banana 2.

6:08 And I'm just going to start with this prompt.

6:10 Create a bobblehead figurine of me.

6:12 And then attach an image of me.

6:13 Now, this is not the test yet.

6:15 What I want to see is if I can bring that figurine

6:18 into various scenes without it turning

6:20 into a completely different figurine of me.

6:23 In other words, if I use this figurine in five different scenes,

6:26 is it still going to look like the same Eigor bobblehead that it creates here?

6:30 Okay, there you go.

6:31 Eore bobblehead edition.

6:33 Not bad.

6:33 That's actually kind of cool.

6:34 Here comes the real test.

6:35 Now, put the figurine on the table of the US president and the Oval Office.

6:40 You can see loading Nano Banana 2.

6:41 And there it is.

6:42 Okay, let's be super critical here.

6:44 I mean, the hair is good.

6:46 Yeah, honestly, I would say 9 out of 10.

6:48 In previous models, I would say the consistency here was like a 6 out of 10,

6:52 maybe a five out of 10.

6:53 It's not perfect.

6:54 Perfect.

6:55 If I'm super critical, I would say that maybe Wow, I don't even know.

6:59 Probably just the lighting change.

7:00 No, this is really good.

7:00 Let's try one more.

7:01 Place the figurine in the hand of an astronaut

7:04 stepping onto Mars for the first time.

7:08 Heck yeah.

7:09 Okay.

7:10 Yeah, that's still the same figurine.

7:11 The consistency is incredible here.

7:12 Okay.

7:13 Okay.

7:13 Okay.

7:13 We like it.

7:14 We like it.

7:14 So, you can totally do this.

7:15 You know, create one asset and then put it into various situations.

7:18 Amazing for, you know, product photography and e-commerce.

7:21 You can kind of just add one image and all of a sudden,

7:24 because this stuff stays consistent, finally, you can place it anywhere.

7:27 No specialty tools, no crazy workflows

7:29 and apps that take multiple days to learn.

7:32 Just good old Gemini on a $20 plan that I think you can get for $10.

7:36 And I think Dana Banana 2 even works on a free plan.

7:38 Yeah, even on the free plan, you could just try this.

7:40 Amazing.

7:41 Let's switch it up a little bit and say

7:42 create four more bobblehead figurines of the following characters.

7:45 Penguin, cat and a hat, a turtle, and a lion.

7:48 And here's the fun thing.

7:48 They claim that the model can now handle up to 14 objects like this.

7:52 This is the first time I've ever heard a claim even close to that.

7:55 14 objects that it consistently can place into various scenes for you.

7:59 So, we're going to have like these four figurines plus me as the fifth one.

8:03 Look at them right here.

8:04 Beautiful little lineup there.

8:05 Especially the cat with a hat and maybe even the penguin.

8:08 Yeah, I think the penguin is actually my favorite.

8:10 We prepared a pretty precise prompt here.

8:12 And this is another thing that I told you about

8:14 that the model is really good at adhering to details in a prompt.

8:17 We call this prompt adherence.

8:19 So when I put in this detailed prompt here, let's have a look.

8:21 Create a playful beach scene with these five characters.

8:24 Actually, I'm just going to replace

8:24 this and say these five bobble head figurines.

8:28 All characters except the penguin are wearing sunglasses.

8:31 Okay, so we're adding glasses to all of them.

8:33 Okay, the penguin is holding a tropical drink with a small

8:35 umbrella in its right hand and sipping from it.

8:37 And then a bunch more details about the rest.

8:39 Let's just run this and see if it

8:41 can actually do all of these detailed instructions.

8:43 Edit to the figurines that exist and then place them

8:46 in a completely different scene

8:48 while maintaining character and detail consistency.

8:51 All right.

8:51 Wow, this looks impressive.

8:52 So, editing team, I would love a good checklist on the side.

8:55 Let's see if all of these things are present.

8:58 Starting out, playful beach scene with all five figurines.

9:01 Playful beach scene set.

9:03 1 2 3 4 5.

9:04 Perfect.

9:05 All characters except the penguin are wearing sunglasses.

9:08 Check.

9:09 Not bad.

9:10 The penguin is holding a tropical drink with a small

9:12 umbrella in its right hand and sipping from it.

9:15 Yep, that worked.

9:16 The turtle is lying on a sunbed tanning while reading a book.

9:19 Oh my god, he's totally Yeah.

9:21 tanning and reading a book.

9:22 It's perfect.

9:22 He actually put the book away, but like it's accurate, right?

9:25 He's in the process of reading the book.

9:27 Sure, he could be holding it, but no, that was ambiguity in the prompt.

9:30 We said he's primarily tanning,

9:31 so actually this is super accurate to the prompt.

9:33 The cat and the line are playing beach volleyball together in the sand.

9:36 Yep, they're definitely playing.

9:38 Looks like the cat is about to score.

9:40 Oh my god, this is so good.

9:41 My bubblehead version of me swimming in the ocean nearby.

9:44 Yep, right there.

9:46 Redressed me into a swimsuit as appropriate here.

9:49 The setting is a sunny tropical beach with blue water,

9:51 palm trees, and a bright summer atmosphere.

9:53 Style should be colorful, fun, and slightly cartoonish.

9:55 Nothing short of perfect.

9:57 That's really impressive, actually.

9:59 Like, I've seen so many of these models in an action,

10:01 and it's always something that just doesn't perform as you would expect it to.

10:05 You ask for something, it doesn't show up.

10:06 You ask for character to be shown up,

10:08 all of a sudden they look like a slightly different person.

10:10 Look at that.

10:10 Even this cat, the hat is the same.

10:12 Look at these red highlights and the ribbon here on top.

10:15 Same thing.

10:15 Red highlights, ribbon here on top.

10:17 The penguin has a little red bow tie.

10:19 Look all these details.

10:20 They just carried over.

10:22 This is so freaking good.

10:24 I guess I found one thing.

10:26 If I were super nitpicky,

10:27 it would be this turtle seems like a different character.

10:30 But we can make edits, right?

10:31 Make sure the turtle is consistent with the original bobblehead.

10:40 I'm not even sure if this is going to work.

10:41 I just want to try.

10:42 Okay, added these boxes.

10:43 Didn't change the turtle.

10:44 So, fair enough.

10:45 That didn't work.

10:45 But I'll show you one more.

10:46 What you can totally do is make edits to, for example, the format.

10:50 So, change the format to fit an Instagram story of this image.

10:55 I'll add that, too.

10:56 Update.

10:57 Yeah, there it is.

10:57 Just extended it.

10:58 Amazing.

10:59 You can just post this on the story if you want with text up here.

11:02 So good.

11:03 M so good.

11:05 Now look, I want to say one thing about the prompt adherence.

11:07 Sometimes it can be a double-edged sword, right?

11:09 Because it's so precise.

11:11 Every word that you enter there really matter.

11:13 So if you don't express yourself precisely, it will have to interpret it.

11:16 So be precise with your language, then this is amazing.

11:18 And in case you don't like something, then go back and change the prompt so

11:22 that the words exactly represent what your idea is.

11:25 And then there's two more fun things that I

11:26 found on Twitter that I want to show you.

11:28 One of them is from Chris Kashanova

11:29 that always experiments a lot with AI image models.

11:31 Great following X by the way.

11:33 She created an anime style panorama of a location.

11:36 And what I'll do here is just upload this Google Maps

11:38 image of Midtown Manhattan in New York somewhere around Central Park here.

11:42 Let's create a anime style panorama of this location.

11:45 Oh, look at that.

11:46 Not bad.

11:46 Kind of a good illustration.

11:48 Honestly, I don't know this street well

11:49 enough to judge if this is entirely accurate.

11:53 Definitely looks great.

11:54 Kind of a fun one.

11:55 One more thing that I wanted to show you which is

11:57 its ability to replicate handwriting can be also kind of scary, right?

12:00 But let me just show you this.

12:02 Okay, I'll just scribble a quick sentence.

12:03 All right, so this is the image.

12:04 Hey, what's up?

12:05 I wonder if the font will match.

12:06 Let's see if it can continue in a font similar to mine.

12:09 Add text to this picture of a thank you note for moving help.

12:13 What I'm hoping for is that it's going to match this font and extend

12:16 the writing so that it looks exactly as if I hand wrote it.

12:20 Okay, let's try one more time.

12:21 That's not perfect.

12:22 Obviously, match the font color and style.

12:24 That's way better, but not perfect.

12:26 You can see it's good at handwriting.

12:28 If you want to make images of handwritten notes, it's really good at that now,

12:31 but this is not a perfect match to my handwriting,

12:33 which honestly is probably a good thing.

12:35 But yeah, it's really good at handwriting notes as long

12:38 as you're not trying to perfectly match what you're doing.

12:40 I guess some of the examples on the internet are a little

12:42 overblown on this one cuz people state that it can match any font.

12:45 I don't know about that.

12:46 And that's it.

12:47 Nano Banana 2.

12:48 Pretty impressive, right?

12:49 Several things you can try on social media or to update your website.

12:53 I think the prompt adherence is incredible.

12:55 And the fact that it can maintain all of these characters, wow,

12:58 really opens up a lot of new opportunities and makes it so much more practical.

13:01 Product shots for online shops or just

13:03 telling stories with images that are consistent.

13:06 Amazing update.

13:06 And if you found some unique use case for it,

13:08 please share it in the comments below.

13:10 That being said, my name is Eigor and I hope you have a wonderful day.