Mini Beijing AI Image Prompt

Create a surreal macro food-art miniature cityscape titled {argument name="headline text" default="MINI BEIJING"}, showing Beijing rising from a vast desert of fine golden cane sugar. The centerpiece is a detailed edible Temple of Heaven in the middle: round tiered white sugar-stone terraces, carved staircases, a circular white balustrade, and a dark indigo-purple ridged roof with a golden finial. Surround it with exactly 12 distinct miniature edible Beijing-inspired structures and landmarks: 1 white pagoda tower on the left, 1 small red-and-gold palace pavilion in the front-left, 1 larger red-and-gold palace hall on the right, 2 small red temple buildings flanking the central temple, 1 orange-red ceremonial gate in the foreground over a turquoise pond, 1 green lattice-pattern modern arch resembling the CCTV Headquarters on the right, 1 tall golden needle-like skyscraper in the rear-left center, 1 glowing red tiered tower in the rear center, 1 tall amber-lit skyscraper in the rear-right, 1 purple-lit vertical tower in the far background, and 1 rounded glass dome building behind the temple. Make all architecture appear crafted from edible materials such as sugar crystals, candy, wafer, icing, jelly, fondant, sesame, and translucent hard candy, with warm glowing windows and candy-like details. The ground is a textured golden sugar desert with scattered white sugar pearls around the central plaza, orange candy-brick paths, tiny green broccoli-like trees and shrubs, edible grass reeds, small rocks, and a shallow bright turquoise candy-water pond at the bottom. Use cinematic shallow depth of field: sharp focus on the Temple of Heaven and foreground gate, softly blurred skyscraper skyline in the back, atmospheric pink-purple mist and warm orange light, high contrast, glossy macro photography look, square composition. Place the headline text in the upper-left corner in elegant spaced white serif capitals, two lines, reading exactly {argument name="headline text" default="MINI BEIJING"}. No people, no vehicles, no extra labels, no watermark.