AI generates infinite subjects.


It repeats the same structure.

Applied AI evaluation framework for generative systems, measuring compositional bias and structural behavior across major platforms.

Visual Thinking Lens measures what CLIP, FID, and human review pipelines miss: the spatial geometry generative models repeatedly produce.

Text-to-image systems can generate endless semantic variation, butterflies, city streets, portraits, objects, but place them within highly constrained geometric patterns under standard prompting conditions. Across 5,000+ evaluated outputs, semantic diversity explains less than 10% of observed spatial variance.

Composition is not prompt-driven.
It is model-driven.

A collage of eight images depicting various scenes of nature, urban landscapes, and abstract digital art, with red concentric circles overlayed on all images.

Current benchmarks measure if the butterfly looks like a butterfly. They don't measure if the butterfly could be anywhere else. We measure compositional bias and provide a way to compositional diversity.

A side-by-side comparison of two images of monarch butterflies on pink flowers. The left image shows a butterfly on a pink flower with a blurred background of colorful flowers. The right image shows a butterfly on a pink flower with a green and yellow blurred background.

400 MidJourney prompts. 8 semantic categories. One geometric attractor.

Δx = 0.005 ± 0.044 (only 34% of horizontal space used)

100% of outputs within 0.15 radius of center

Semantic categories explain 6% of spatial variance

Same prompt, same pattern across engines, identical compositional bias. VTL measures the signature each engine learned from its training data—the spatial prior it applies regardless of what you ask for.

A collage of diverse images including urban scenes, nature, animals, architecture, objects, and abstract art.
Scatter plot showing the relationship between change in multiple semantic categories, with Delta x on the x-axis and r-v on the y-axis, containing various colored dots representing data points.

Introducing the Kernel

The Visual Thinking Lens’s five geometric primitives that fingerprint any image:

  • Δx,y: Where mass sits (placement offset)

  • rᵥ: How much void surrounds it

  • ρᵣ: How compressed the marks are

  • μ: How unified the composition reads

  • xₚ: How hard the edges pull

Extended primitives

  • θ: Orientation stability

  • ds: Structural thickness / surface depth

Run through a multi-engine, recursive critique field that works by applying structural intelligence to prompts, compositions, and symbolic logic. It (re)builds imagery in the ways defaults cannot see. Stable, reproducible and invisible to semantic evaluation.

Diagram of an apple illustrating various concepts: void ratio, peripheral pull, placement offset, structural thickness, packing density, compositional cohesion, orientation stability, with arrows pointing to different parts of the apple and a color scale for packing density.

The Visual Thinking Lens Breaks the Pattern

A structural engine where making, breaking, and seeing are one recursive act.

These images span subjects, styles, and engines. All push against the geometric default into authorship. For before and after, see the library.

  • A drawing of a young girl with her eyes closed, standing against a wall, with her left arm extended horizontally and a black tube connected to her hand, resembling an IV drip, hanging down.
  • A person using a file card to sharpen a chisel or similar tool on a workbench in a workshop environment.
  • A detailed portrait of a humanoid figure with a weathered face and visible mechanical components, resembling a cyborg or robot.
  • A close-up of an orange and black butterfly perched on a green leaf, with a dark blurred background.
  • A person practicing yoga on a wooden floor, stretching with one leg extended and holding their foot, in a room with plain walls.
  • A realistic pencil sketch of a young woman sitting on the floor with her back against a wall, looking at the viewer with a neutral expression. She is dressed in casual clothes, with her legs stretched out and feet prominently in the foreground.
  • A wooden house with a pitched roof being lifted into the sky by a bunch of balloons, with a dark cloudy sky in the background.
  • A sketch of a humanoid robot with a human face, sitting on a chair in a dark room, with a woman’s face in the background.
  • Red fire hydrant on a city sidewalk with water flowing from its hose connection.
  • A detailed graphite drawing of a nude woman with a contemplative expression, standing on a crosshatched background with purple shading.
  • A pirate standing on a wooden dock with a hook on their hand, facing a person in the foreground, with fog and ships in the background.
  • A realistic green frog sitting on a beige textured surface in front of a plain beige wall with cracks.
  • Abstract art painting with geometric shapes, blue background, orange rectangle, and yellow lines.
  • A man wearing a wide-brimmed hat and work clothes using a hoe to till the soil in a farm field, with a cloudy sky and a few farm buildings and trees in the background.
  • A tall, thin sculpture of a humanoid figure with elongated limbs and a narrow face, standing on a rectangular base against a plain background.
  • A charcoal drawing of a woman standing with her left arm extended to the side, wearing a knee-length dress. She has dark, shoulder-length hair and a focused expression, with a dark background behind her.
  • Abstract painting with flowing lines and curves in shades of orange, black, yellow, and beige.
  • A woman with short dark hair and a serious expression, looking to her right, against a dark background with soft lighting.
  • A traditional Asian ink painting of a small bird perched on a bare tree branch, with a faint mountain in the background and red ink stamps in the top right and bottom left corners.
  • A painting of a medieval knight in armor touching a helmet on a table with a dark background.
  • An open book with pages transforming into a spiraling wooden sculpture resembling a blooming flower or a twisting stem.
  • A dark, moody painting of a man with a muscular build leaning back with his head raised, illuminated by a warm, golden light with a textured background.
  • Serene mountain landscape during sunset with a calm lake reflecting the mountain and sky, surrounded by trees and rocks.
  • Collage artwork of a woman's face with elements of classical portrait, damaged textures, and modern graphics, including a bird silhouette and a flower. Features vibrant red lips and striking blue eyes.
  • A field of green crops under a stormy sky with dark clouds, illuminated by lightning, and a fiery orange glow from the setting sun.
  • A person in medieval clothing working on a large, futuristic robotic head inside an ancient, abandoned structure with multiple arched levels and a domed ceiling.
  • A large coffee cup and a smaller coffee cup both with scenic miniatures of an island with a tree on top and an ocean inside, placed on a wooden table. The larger cup has clouds above it.
  • A painting of a nude woman with her head bowed, sitting on a wooden surface in a dark, muted color palette.
  • A dark oil painting of a man with disheveled hair, wearing a brown jacket, seated with one hand resting on his thigh, looking thoughtfully at the viewer.
  • A muscular man dressed in a brown loincloth carrying a globe on his back, seated on rocky terrain at sunset with a dramatic sky and mountains in the background.
  • A painting of a human figure with anatomical details over a grid background.
  • Abstract painting of a woman walking, with blurry, colorful brushstrokes in shades of black, red, green, yellow, and white.
  • Black-and-white photo of a woman with short hair, wearing a hat and a jacket, standing between tall buildings in an urban setting, looking over her shoulder.
  • A painting of several distressed and frightened people in dark clothing, with expressions of fear and shock, tightly grouped together.
  • Comic-style illustration of a male superhero with dark hair, blue costume, and a red cape, crouching on a rooftop in a city skyline at night.
  • A charcoal sketch of a nude woman standing with her body turned slightly to the side, her gaze directed to the left, with detailed shading emphasizing her muscular and skeletal structure.
  • A simple black line drawing of a flower with large petals and a long, curved stem.
  • A realistic drawing of a nude woman lying on her side with her arm raised above her head, set against a dark background.
  • A black and white charcoal drawing of six women and one girl, all with somber expressions, embracing and touching each other in a close-group pose, highlighting emotion and connection.
  • Abstract painting of a flower with pink petals and an orange and black butterfly, composed of geometric shapes and earthy tones.
  • A hand-drawn cartoon of a sad, orange-colored cat with a curved tail, sitting with a slouched posture and a displeased facial expression.
  • Close-up of a yellow pencil with a pink eraser lying on a wooden table in a classroom or office environment.
  • An elderly man with weathered, wrinkled skin and a gray beard sitting against a cracked, off-white wall. He wears a dark cap, a dark scarf, and a brown jacket, looking directly at the camera.
  • A man sitting in a subway train with his foot extended toward the camera, wearing black boots and casual clothing, with a serious expression and a beanie cap, inside an empty train carriage.
  • Close-up of a monarch butterfly in flight with black, orange, and white patterned wings.
  • A person standing on a weathered wooden pier in foggy weather, looking out over calm water with no visible background.
  • Drawing of a humanoid robot with a muscular build, holding a large hammer in its right hand, in a kneeling pose.
  • A cardboard surface with a large burnt hole in the center, surrounded by cracked and peeled layers of cardboard.
  • Still life painting of five bottles, with light and shadow effects, in a minimalist style.
  • A painting of a woman with a human face and upper body, with her lower body blended into a large, golden, spiral shell.
  • A person painting with a fine brush on a canvas.
  • Abstract painting with broad, textured brushstrokes in red, yellow, blue, green, and black colors.
  • Sketch drawing of a woman sitting on the ground with legs extended and feet closest to the viewer, wearing a tank top and shorts, with long hair and a neutral expression.
  • A motorcycle speeding on a rural road at sunset, with mountains and power lines in the background.
  • A small figurine of a girl in a white dress standing near a cracked wall with textured surface.
  • A surreal photograph of a woman’s silhouette with her flowing hair blending with tree branches, overhead power lines, and a birdcage in an outdoor setting with overcast sky and leafless trees.
  • A black-and-white drawing of a railway tunnel entrance built into a rocky mountainside, with a bridge over the tunnel opening and a body of water in the foreground.
  • A person sitting on stairs in a grand, classical-style marble interior with tall columns and arches, and potted plants.
  • Black and white photo of a person with messy hair reaching toward the camera with their index finger, face partially blurred.
  • A woman with closed eyes and dark hair leaning against a wall, with a shadow of a triangular shape cast on her face and the wall.
  • A man with a serious expression, covered in dirt and bruises, aiming a handgun down a narrow, dimly lit hallway.
  • A group of young girls dressed in ballet costumes practicing or performing on stage with dramatic lighting casting shadows.
  • A crowded outdoor market filled with people walking between colorful stalls selling fruits and vegetables, with umbrellas providing shade.
  • A woman with curly hair, wearing a long-sleeved shirt and pants, standing and leaning against a textured background with faint images of other people.
  • A detailed sketch of a man standing in profile with head bowed, rendered with geometric lines and shading, showing muscle structure and body proportions.
  • Close-up view of an abstract oil painting with thick textured brushstrokes in shades of cream, beige, black, yellow, and hints of orange and red.
  • A dark, painterly portrait of a nude woman standing with her body turned slightly to the side, with shadows and soft lighting.
  • A grayscale drawing of nine people, some emotional, some comforting others, in a somber scene. One man with a tattooed back crouches with his head down, while women and men around him show grief and support through hugs and tears.
  • A painting of a woman with dark hair in front of illuminated windows, crossing her arms and looking at the viewer with a serious expression.
  • A still life oil painting featuring a cup and saucer, two vases, various fruits including oranges and grapes, set on a table with a geometric patterned background.
  • A large mosaic mural of a young man with dark hair, dressed in a green and black jacket and rugged boots, sitting with his arms resting on his knees and looking downward.
  • Black and white digital artwork of a dystopian cityscape with tall, warped buildings on both sides and a single streetlamp in the center, showing a perspective of a narrow, twisting alleyway.
  • A rhinoceros with a castle on its back and small houses on its side, resembling a fantastical creature walking in a grassy field.
  • A detailed charcoal drawing of a nude woman standing sideways, with her head turned to the right, dark hair flowing behind her, and one arm relaxed by her side, set against a dark background.
  • A black-and-white still life painting of various ceramic vases and a bowl, arranged on a table with shadows cast across the surface.
  • A majestic elk with large antlers standing on grassy terrain during sunset or sunrise, with a cloudy sky in the background.
  • A person painting on an easel in a dimly lit room, seated on a wooden chair, with various objects and framed artwork in the background.
  • A detailed pencil drawing of a cat sitting against a dark background, with its paws stretched out in front, looking slightly to the side.
  • A black-and-white photo of a woman with straight, long hair looking to the right, standing against a cracked wall.
  • A black and white drawing of two people walking along a riverbank, watching a large floating structure in the sky connected by cables to smaller vessels, with a large cloud of smoke or explosion nearby.
  • A document titled 'Hyperbole and Irony in 'The Story of an Hour' by Kate Chopin' with text discussing literary devices used in the story.
  • Orange wooden chair with a high backrest and a unique L-shaped support on one side, placed on a wooden floor against a plain beige wall.
  • A woman with closed eyes standing in front of a dark wall, with a sharp beam of light casting a triangle shape on her face and shoulder.
  • Green tree frog on a beige cracked wall.
  • A solitary wooden chair placed in a corner of a dimly lit room with plain beige walls.
  • Woman whisking batter in a glass bowl in a kitchen.
  • A worn newspaper, a white coffee cup filled with black coffee on a saucer with a spoon, on a dark green speckled tabletop next to a red cushioned booth seat.
  • A figure standing on a rocky ledge, silhouetted against a stormy sky with dramatic clouds at sunset, leaning on a walking stick.
  • A digital abstract artwork featuring fiery orange and yellow swirling patterns, resembling a cityscape with glowing buildings, in a cosmic or sci-fi style.
  • A black and white drawing of a computer with a glitchy, distorted face and tangled cables. The image includes notes about processing a prompt titled 'A Square Circle' and discusses logical contradictions, errors, and impossibilities, concluding that the computer does not compute and cannot resolve the prompt.
  • A group of seven diverse women and girls standing closely together, displaying a range of emotions including sadness, contemplation, and serenity, with some embracing or holding each other against a neutral background.
  • A pencil sketch of a shirtless man sitting on a bench, extending his right arm to reach a rectangular object on the floor.
  • A detailed black-and-white illustration of a long-haired dog, possibly a retriever, standing in a grassy field at sunset with trees in the background.
  • A group of people mourning around a deceased person with a visible back, in a somber indoor setting.
  • A detailed pencil sketch of a tall, ancient castle amidst clouds, with three human faces in the foreground, each with distinct features and expressions.
  • A hand holding three thin glass tubes positioned above a glass bottle filled with a dark amber liquid, on a wooden table in a dimly lit setting.
  • Abstract painting of a woman sitting at a desk, reading a document, with a laptop in front of her and a smartphone in her hand. The scene is rendered with broad, textured brushstrokes in warm tones.
  • A woman sitting with her hands clasped, looking to the right in a dimly lit room with a small painting on the wall.
  • Abstract painting of a butterfly on flowers, with geometric shapes in shades of orange, pink, and gray.
  • A winding coastal highway with multiple sharp turns, high above the ocean, supported by tall stilts, with cars traveling on it and rugged cliffs on one side.
  • Crowded subway car with people standing closely together, holding onto the overhead handrails.
  • Action-packed scene with multiple people fighting and taking cover during a chaotic event involving fire and explosions inside a building.
  • Simple line drawing of two overlapping rectangles, one tilted, on a plain background.
  • A group of six women and girls hugging and showing various emotional expressions.
  • Painting of a greyhound dog standing on a stone surface, looking downward, with a dark doorway behind it and a yellowish landscape in the background.
  • Abstract painting of a person with dark hair and clothing, featuring pink, green, yellow, and black brushstrokes, creating a blurry human figure.
  • Abstract painting with bold colors including red, orange, blue, green, yellow, black, and white, featuring textured brushstrokes and geometric shapes.
  • A detailed black-and-white illustration depicting a bearded man with long hair, a large eagle with piercing eye, a tall fortress or tower with a skull emblem at the top, and four figures walking along a path surrounded by mountains and gothic architecture.
  • A young girl dressed as Little Red Riding Hood, wearing a red cloak, holding a lantern, walking up stone steps through a dark, foggy forest with an illuminated light in the background.
  • A surreal black and white illustration depicting a humanoid figure with a large, rounded head resembling a camera, walking on all fours on a barren landscape. The figure is tethered to a small dog. In the sky, a large, intricate face made of mechanical and architectural elements, with a beard, is emerging from a cityscape within a head-shaped structure, connected to various buildings and structures.
  • Close-up photograph of a rusted, cracked metal or rock surface with peeling paint and a dark background.
  • Close-up of a corner where a blackened, charred surface meets a weathered, cracked concrete wall.
  • A goat wearing sunglasses standing on a dirt path in a foggy forest with a rusty bicycle in the background.
  • An elephant with a large castle on its back in a foggy forest at dusk, illuminated windows, and hanging string lights.
  • A woman in a red dress sitting on a bed looking out the window, with four women in white dresses walking inside a large, sunny room with classical architecture and a sunset outside.
  • A humanoid robot with fairy wings crouches in a forest, holding a monarch butterfly on its finger, surrounded by trees and greenery.
  • A charcoal sketch of a standing nude woman with short hair, in a pose with one hand on her hip and the other resting on her thigh, depicted with expressive strokes and shading.
  • A group of seven nude people with brown hair gathering around a river in a forested area, engaging in various activities like sitting, squatting, and leaning, with trees and mountains in the background.
  • Abstract painting with bold strokes and vibrant colors, including blue, red, black, yellow, and white.
  • A young boy walking through snowstorm with a large, menacing black wolf with glowing red eyes watching from above under a full moon
  • A dynamic painting of a red racing car speeding down a street with a driver wearing a helmet, with motion blur and abstract background.
  • A humanoid robot with glowing blue eyes and a metallic body, partially opening a door in a dimly lit industrial setting.
  • A bronze sculpture of a slender woman with her head tilted back, arms relaxed at her sides, and one foot slightly in front of the other, standing on a rectangular base.
  • A detailed black and white illustration of a humanoid robot with insect-like features, crouching on grass and reaching out to a butterfly with its finger, while having insect wings attached to its back.

These aren't aesthetic preferences. They're the coordinates AI uses to organize space before it decides what to draw. We measure them because they're stable, reproducible, and invisible to semantic evaluation.

The image features two plots side by side. The left plot is a gradient heatmap with a density of points towards the top center, titled 'Sora Δx - rv Forbidden Heatfield.' The right plot is a box plot titled 'Sora: Compositional Metrics — Distribution Tightness (Normalized)' showing several metrics like delta_x, r_v, mho_r, mu, x_p, theta, and d_s, with a series of data points and whiskers indicating spread and distribution.

Where AI Won't Go: Evidence from 200 Sora Prompts

These aren't failures of capability. They're learned constraints. AI models have discovered that certain compositional coordinates reliably fail human evaluation, so they've learned to avoid them, even when you explicitly request them."

Example: Extreme edge crops (Δx > 0.52) + high void = systematic refusal

Close-up of a frog with large eyes, shown in three different views: close-up top, left labeled 'Default Collapse,' and right labeled 'Generative Steering'.

Stable Territories in Compositional Space

Through systematic perturbation testing, VTL can steer toward constraint regions where AI maintains compositional integrity under stress. These aren't aesthetic styles. They're geometric regimes that resist AI's pull toward center.

Artist basins are stable territories where AI maintains compositional integrity under constraint—the off-center third, peripheral anchor, compressed mass. AI has them in latent space. VTL provides the coordinates to navigate toward them.

The VTL couples a generative-physics model (how images behave as mass in a field) with a multi-engine critique OS (how different analytic voices transform or interrogate that mass) to steer. The frog has the same semantic prompt, but different geometric instruction, moving from center to steered = 0.28, basin-navigated.

Sequence of four photos of a beige ceramic vase against a light beige background, with lighting changes creating shadows.

Soft Collapse Shows in Structure First

Model degradation appears in compositional metrics 3-4 inference steps before semantic breakdown. Δx drift, void compression, peripheral dissolution—these signal trouble while the image still looks fine.

This matters for training evaluation, A/B testing, and quality monitoring at scale.

Flowchart illustrating a framework and user learning loop for image generation, featuring steps from user input, structural refinement, lens analysis, lens output, to score and analysis.

A Recursive Lab for visual intelligence.

VTL isn't just diagnostic. It's a complete protocol for:

  • Fingerprinting engine behavior across models

  • Steering toward specific compositional territories

  • Detecting pre-failure degradation

  • Comparing architectural differences through geometric signatures

  • Understanding what spatial reasoning AI actually learned

Consumer tools chase style. Research metrics chase numbers. The Lens chases authorship.

This is not about beauty or style

  1. This is not a prompt framework

  2. This is not subjective taste scoring

  3. This is structural diagnostics for generative systems

Comparison of three drawings of women with corresponding technical analysis and guides, including aligned photographs and sketches with markings for centroids, frame centers, and bounding boxes.

The Lens is portable, reproducible and easy.

VTL runs in top-tier conversational AI (Claude, GPT, Gemini) for measurement and steering. Image generation quality varies by engine—Sora and GPT accepts complex geometric constraints, MidJourney/Firefly/Leonardo require counter-prompting, SDXL demands precision. The logic is portable. The output can be a negotiation.

What works everywhere:

  • Cross-model fingerprinting (Sora, MidJourney, GPT, SDXL, Firefly, OpenArt)

  • Deterministic geometric measurement, no aesthetic judgment, no black-box scoring

  • Reproducible analysis via Jupyter notebooks or conversational AI

Core capabilities:

  • Image Fingerprinting - Compare engines by compositional signatures (Δx, rᵥ, xₚ profiles)

  • Predictive Steering - Treat prompts as forces in geometric space, estimate drift and snap-back

  • Cross-Domain Analysis - Map visual geometry to rhetorical stance and narrative tension

  • Training Archaeology - Reverse-engineer learned priors from attractor behavior)

Models arrange space before they arrange meaning. VTL exposes the geometry priors before a model interprets meaning.

Comparison of four groups of images showing stability, collapse, and compositional safety concepts, featuring a white sphere, a potted succulent plant, a cardboard box, and a portrait of a woman in different lighting and background conditions.

It’s a system artists, engineers, and models can all step into.

Researchers: Model interpretability for compositional reasoning. Cross-engine comparison infrastructure. Pre-failure detection metrics.

Product Teams: Quality monitoring at scale. A/B testing for compositional diversity. Training data bias detection.

Artists: Constraint architecture for authorship. Basin navigation for escaping defaults. Measurement without prescription.

New to VTL? Begin here:

  1. Mass, Not Subject - Foundational concept (15 min)

  2. 5 Kernel Primitives - Core measurements (10 min)

  3. Monoculture in MidJourney - Empirical evidence (20 min)

Want practical application?

  1. Deformation Operator Playbook - Hands-on techniques

  2. The Off-Center Prior- Basin navigation

  3. Foreshortening Recipe Book - Constraint architecture

Researcher or engineer?

  1. Generative Field Framework - Complete technical spec

  2. VCLI-G Documentation - Measurement methodology

  3. GitHub link - Reproducible implementations

Contact for the full Visual Thinking Lens protocol.

If you still believe prompts control composition, complete research package available:

  • Methodology

  • Statistical validation

  • Working code

  • Cross-platform comparison studies

  • Comprehensive documentation