Google Just DROPPED Nano Banana Pro… And It’s Actually INSANE 😳

Quick Overview

The creator tested Google's Nano Banana Pro AI model by feeding it prompts to generate images based on their own photo and clothing, concluding that the image generation capabilities are "insane" and "pretty sick," though noting some slight face distortion in one attempt where the camera angle was pushed backward.

Key Points: The creator tested Google's Nano Banana Pro AI model by providing a prompt to place the creator in an existing outfit using their photos as reference. The AI successfully generated an image of the creator in a blue blazer and khaki outfit, noting the result looked exactly like him, though the background was slightly off. The creator then tested the model's ability to render a 'head to toe Barbour country wear' outfit using a reference photo, which resulted in a good image, although the face appeared slightly distorted or zoomed out. The creator also tested the AI on generating a complex thumbnail prompt for 'Gemini 3 Upgrade??' which resulted in a visually impressive, futuristic banana image, indicating strong visual generation capabilities. Another successful test involved prompting the AI with an image of the Lions Gate Bridge and asking it to generate a blueprint-style overlay showing engineering details, which the model executed 'pretty sick' and 'super sick'. The creator reviewed comments on a previous video, noting concerns about Semrush's relevance and a user whose 4,000 doorway pages were penalized, suggesting the AI image generation results are currently superior to some current SEO tools.

Context: The video features the creator testing the capabilities of Google's new AI model, referred to as 'Nano Banana Pro,' focusing specifically on its multimodal features like image inpainting and image generation based on provided reference images or detailed text prompts. The creator uses his own images to test identity preservation while changing outfits or environments, and uses complex descriptive prompts to evaluate the model's creative and detail-oriented visual output.