⁽¹⁾Tasting The Unseen Cookbook

Tasting the Unseen was created to challenge and redefine what a cookbook can be. Not a set of fixed instructions, but an open invitation to explore, where culinary intuition and artificial intelligence converge in a journey across flavors, images, and emotions. At its core lies a radical question: What if imperfection was the secret ingredient? Each recipe becomes a moment, not a formula: born from textures, memories, and sensations. Cooking becomes a personal, interpretive act, guided more by instinct than precision. The AI-generated visuals don’t replicate food as it is: they reimagine it. Ambiguity, surreal forms, and aesthetic “errors” shape a visual language that resists convention. In this space, AI becomes more than a tool: it’s a collaborator, surfacing what often escapes the human gaze. Pages become sensory triggers, blending tradition with speculation, stirring emotion, and opening new paths. The layout mirrors this logic: a fluid system where words and images move freely. It’s not a recipe book, but a space where process outweighs perfection, and creativity begins by embracing the unexpected.

The images in Tasting the Unseen are created through artificial intelligence. They do not aim to represent food with photographic accuracy but to evoke its atmosphere and emotional resonance. Imperfections such as unreal textures, distorted proportions and impossible details become part of their charm. Suspended between the dreamlike and the tangible, they invite the viewer to imagine, reinterpret and taste beyond what is seen.

In Tasting the Unseen, each recipe generates its own visual grid. Built from its keywords, ingredient types, flavors and textures, the grid becomes a unique fingerprint for the dish—an abstract map of its identity. Rather than decoration, it’s a structural layer, visually translating the recipe’s qualities into a pattern that can only belong to that plate.

After generating the images, we trained a custom AI model to analyze each one through a set of conceptual and sensory dimensions: Hallucination, Abstraction, Softness, Symbolism, Realism, Crunchiness, Sensorialism, and Onirism. These categories translate visual traits into a kind of “flavor profile” for imagery, measuring how surreal, tactile, symbolic, or realistic each composition feels. The model assigned a unique weight to every category for each image, producing a signature dataset that maps the visual and emotional identity of the collection.