Nano Banana 2 prompt library adds ARRI Alexa 65 sports-ad specs and merch templates

Amir Mushich’s merch prompt reads less like a caption and more like an art-direction brief. It specifies garment weight at 320–380gsm, a matte black hanger on a chrome pipe, a slightly low camera angle, a 4–6 degree counter-clockwise tilt, f/5.6 depth of field, and a dark cyclorama background chosen to contrast with the brand color. The output target is photoreal studio merch that still shows wrinkles, weave, pilling and cracked screen-print ink rather than “AI plastic sheen,” with the full scaffold available through the prompt link.

The sports-ad version uses the same fixed-scaffold logic. MayorKingAI’s template keeps the “ARRI Alexa 65” framing, sharp subject plus motion blur, dramatic lighting and “hero composition,” then swaps athlete description, action, angle, lens and grade. In the tips post, he narrows the variables further: describe the peak moment instead of the sport, prefer low or side angles, and escalate from “dynamic motion blur” to “pronounced dynamic motion blur” when the frame feels static.

The clearest pattern is to freeze most of the prompt and change one slot. Mushich explicitly frames the merch setup as “change a single variable, get consistent results” in his example, and similar outputs show the same hanging-shirt composition reused across Ellesse, Toblerone and Corona-style mockups. That makes the prompt function like a modular product-shoot template rather than a fresh generation brief every time.

Underwoodxie96 is applying the same approach to character photography. The bathtub prompt and hotel mirror-selfie prompt lock perspective rules, anatomy constraints, lighting mix, fabric behavior and background elements so tightly that the prompt behaves like a scene bible. GlennHasABeard’s snow globe prompt does it for macro builds by keeping the dark table, 85mm lens, shallow depth of field and moody product-photography setup fixed while swapping environment, vegetation and light-source variables. Even narrower effect studies like the frosted-glass example fit the same pattern: hold the imaging recipe steady, then tune a single visual effect.