The creator introduces and demonstrates Google's new AI image generation model, 'Nano Banana Pro,' by showcasing its capabilities through a series of examples. He tests its ability to create a technical diagram from a photo, generate realistic images of celebrities in fictional scenarios, maintain character consistency, and even render historical locations based on geolocation data, concluding that it is the first truly production-ready AI image model.
Post caption
Nano Banana Pro has been released, and I think the day we all knew was coming is finally here – 100% indistinguishable-from-reality imagery.
The model is extremely powerful. It understands physics, technical and conceptual prompts remarkably well and has superior character consistency.
You can try it yourself, today on @higgsfield.ai, who is very kindly sponsoring this video.
Comment “KONG” and I’ll send you a link to their 65% OFF Black Friday deal (running for the next few days) which is a steal.
Thank you to @googledeepmind for the early access.
I also want to shout out Google because they’ve done something really clever here. Every image generated with Nano Banana comes with an invisible watermark called SynthID.
This means, whenever you see an image you suspect of being AI-generated, you can upload it to Gemini and ask – it will tell you with high accuracy whether it was uploaded by Google AI.
If you like this video, follow @rpn to stay two steps ahead on the incredibly disruption happening in tech and how to leverage it in content and business.
Transcript
Well, we all knew this day was coming, and it's officially here. Google just released No Banana Pro, and it is 100% indistinguishable from reality. But the realism isn't even the impressive part. I uploaded a photo of this talking fi diagram of the lighting better. It nailed it. I mean, the angles of the lights, the camera positioning, insane. No, logically, the next step was to ask for Steve Jobs and the Joe Rogan podcast. And this is where we see the realism. I mean, it is perfect. No more imperfections. So I asked for 10 more unique angles of this shot, except I wanted to replace Jobs with Genghis Khan. Absurd. I mean, the character consistency, the understanding of physics, the text accuracy, and and a native two k output with four k upscaling. This is the first AI image model that feels truly production ready. I mean, it's unbelievable. You good try it yourself right now on Higgs Field, who generated these images to further illustrate how realistic this is. But Let me show you one last thing. Nano Banana Pro has incredible Google ecosystem knowledge, including geolocation rendering. So I gave it the coordinates to the Burj Khalifa, but I asked to see it at sunset in 1990. Again, even the cars are era specific. It's it's a remarkable model. It's ultra best. No, give it a try. Let me know what you rank, and I'll catch you in the next one. I good a feeling when Google VO four launches, it's a wrap.
