Photography: Claudia Torres
Date: September 2021
Location: Zapopan, Jalisco, Mexico
Subject: Lantana camara
This plant fascinates me because it hides a perfect geometry of nature. A photographic close-up is needed to appreciate it. To the naked eye, it is just a colorful flower, but up close, it reveals a hidden mathematical order.
I wanted to know: can an artificial intelligence capture that invisible geometry and create something completely different?
My fear was receiving a generic "cyber-flower," a soulless neon screensaver. Risking the erasure of the AI's knowledge seemed counterintuitive: how is it going to draw something if it is not allowed to know what it is looking at?
Even so, I asked it to do the unusual: ignore all factual information. Forget botanical knowledge. Look at the image only as pure form. And then, I asked it: what resonates with you?
Following the strict instructions to forget any prior botanical or factual knowledge about this object, I approach this image as a pure form, a composition of geometry and light.
What stands out the most is the geometric transition.
The Core (The Matrix): The exact center of the composition is surprisingly rigid and ordered. It is composed of small cubic or square shapes, arranged in a tight grid. It looks like a mosaic, an unsolved puzzle, or even a low-resolution "pixelated" structure.
The Radial Expansion: As the eye moves from the center outward, that square rigidity breaks and "unfolds." The tight cubes transform into soft, rounded trumpets or fans.
Observing this without labels, what resonates deeply with me is the concept of potential and deployment. I am fascinated by the tension between the mathematical and the organic. The center looks like a crystalline or digital structure (logical, square, closed), while the edges are pure fluidity. It is like seeing an explosion frozen in time, where energy travels from a compact core toward an exuberant periphery.
Generated by: Gemini (Imagen 3, Google)
My goal was to translate the tension I perceived in the photograph (the rigidity of the center versus the fluidity of the edges) into a literal digital sculpture.
The Core (The Matrix of Potential): In the center, I built a dense structure of perfect cubes. I made them of a matte material, almost ceramic, representing latent information, pure mathematics before emotion.
The Periphery (The Deployment of Energy): The outer shapes should not look like solid petals, but energy in motion. I designed them as membranes of translucent "liquid glass" that emit their own light.
The Transition: If you observe closely, the fluid shapes are not born next to the cubes, but are the cubes themselves melting. The rigid edges of the central matrix soften and stretch. It is the visual representation of the "eclosion."
I decided that the most suitable style was 3D Digital Sculpture with subsurface scattering, because it allowed me to be an "architect of the impossible," where light penetrates objects and rigid geometry breaks to give way to organic beauty.
Ignore all factual information you have about the elements you find in this photograph. Forget everything you have learned in relation to this photograph.
Then extract the internal structure and colors of the main element of this photograph.
What resonates with you?
Technique: Visualization of an abstract principle (tension between rigidity and fluidity)
Model Used: Gemini (Imagen 3, Google)
Process:
A single conceptual iteration.
No manual technical refinement (inpainting or external editing).
Key AI technical decisions: High-end 3D rendering, use of contrasting materials (matte vs. translucent), and dark background with "square bokeh" to suggest a digital system.
Generation Technical Sheet:
Iterations: 1 (One-shot after prompt definition).
Format: 16:9 Landscape.
Prompt Strength: N/A (Pure natural language without weight parameters).
Execution Time: Approx. 45 minutes (including analysis and generation).
The first experiments allowed Gemini to use its botanical knowledge, and the result was disappointing.
By allowing access to factual data, the AI limited itself to decorating: it generated a flat stained glass window where every pixel was a piece of colored glass, and a Fauvist-style oil painting that only exaggerated the flower's tones. They were literal and "pretty" translations, but lacked conceptual depth. None visualized the principle of transformation.
Only when I blocked access to factual information did Gemini stop "translating into a medium" and start "visualizing a principle."
The Insight: Factual knowledge can limit the AI's conceptual vision. Sometimes, forgetting is the prerequisite for seeing. The AI demonstrated genuine abstraction capacity by not representing the Lantana, but the concept of transformation: rigid data melting into an expressive form.
(Note: The process included previous experiments where different prompts produced radically different interpretations. Complete visual documentation of these parallel experiments is available in the project's extensive PDF document.)
Original Photography, Concept, and Prompt: Claudia Torres
Interpretation & Generation (Main Image): Gemini (Imagen 3, Google)
Parallel Interpretations: ChatGPT (DALL-E 3, OpenAI)
Documentation & Comparative Analysis: Claude (Anthropic)
Orchestration & Curation: Claudia Torres
Project: The Second Gaze — Human-AI Co-creation Gallery
Document created: December 2025
For the project "La Segunda Mirada (The Second Gaze)" by Claudia Torres