AI prompt for 3D glass logos shared

A creative director’s copy-and-paste prompt for turning brand marks into glassy 3D mockups is spreading as a practical recipe for premium-looking AI product visuals. (ai-primer.com) AI image prompts are text instructions that tell a model what to render, much like a shot list tells a photographer what to capture. Google says Nano Banana is built for image generation and editing, with support for logos, posters, mockups, and clear text placement. (gemini.google) In the April 2026 post, Amir Mushich shared a prompt that tells Nano Banana to keep a logo’s official geometry unchanged while rendering it as a suspended glass object under a blue sky. AI Primer, which documented the thread on April 11, said the examples included brands such as Adidas, Nike, and McDonald’s. (ai-primer.com) The prompt is structured like a production brief. AI Primer said it breaks the image into subject logic, glass material, environment, and technical camera-style details, including beveled crystal glass, sunlight, refractions, and a Phase One XF with a 120 millimeter macro lens at f/5.6. (ai-primer.com) Google’s own prompting guide, published March 5, 2026, recommends building image prompts in layers and specifying the subject, setting, medium, and overall feel for tighter control. The same guide says Nano Banana 2 and Nano Banana Pro are aimed at uses such as storyboards, ads, and product mockups. (cloud.google.com) That helps explain why a single reusable prompt drew attention: it translates a general-purpose image model into a repeatable brand-mockup workflow. Google says Nano Banana can generate 1 kilopixel, 2 kilopixel, and 4 kilopixel visuals and supports multiple aspect ratios used in marketing layouts. (cloud.google.com) The post also lands in a broader market for packaged prompts rather than one-off experiments. A GitHub repository called “Awesome Nano Banana Pro” had about 9,700 stars when indexed this week and describes itself as a curated library of high-fidelity prompts sourced from X, WeChat, Replicate, and prompt engineers. (github.com) Google’s consumer product page pitches the same model around style transfer, resizing, photo combination, and text rendering, which are the building blocks for mockups that look like ad photography without a physical studio. It tells users to start with a simple prompt formula and then add specifics. (gemini.google) The catch is that polished logo renders are not the same as approved brand assets. Google says users can create logos and marketing images in Gemini, but brand teams still control trademark use, and Mushich’s prompt is framed around preserving official proportions rather than redesigning the mark. (gemini.google) For now, the appeal is straightforward: one detailed prompt can turn a flat logo into something that looks lit, shot, and retouched like a campaign visual. The more the model follows exact geometry and material instructions, the closer AI prompting gets to art direction by template. (ai-primer.com)