This project explores how generative AI tools can transform a real subject into radically different material identities while preserving anatomical accuracy, silhouette integrity, and defining characteristics.

Using a real reference image as the anchor, I created a series of stylized renderings and subtle motion studies across multiple material interpretations, including feather layering and refractive crystal surfaces.

The core challenge was control: maintaining identity while pushing aesthetic transformation.

To test how far a subject’s surface material and visual style can change without losing recognizability.

Constraints included:

Preserving unique ear shape and silhouette

Maintaining accurate facial proportions

Avoiding anatomical distortion

Keeping expressions natural

Ensuring animation-ready compositions

Rather than letting the model “hallucinate freely,” this project focused on controlled transformation.

ChatGPT for structured prompt development, constraint refinement, and image rendering

Morphic for image rendering and animation

The project began with a clean, neutral reference image emphasizing:

Slender proportions

Large upright ears

Distinct facial structure

Clear subject separation

Prompts repeatedly reinforced constraints such as:

“Do not alter ear shapes”

“Preserve anatomical proportions”

“Clean unobstructed silhouette”

“Photorealistic structure”

Maintaining these constraints across styles required iterative refinement.

Layered pastel petal construction, emphasizing softness and volume through overlapping texture density

Aqua-toned feather realism, introducing more natural color gradients and fine structural detail

Translucent crystal refraction with controlled light dispersion, simulating rigid geometry and refractive physics

Texture density and layering behavior

Light interaction and shadow depth

Surface reflectivity and transparency balance

Control of AI-driven detail exaggeration

The petal variation explores soft organic surface construction through layered volume and diffuse light interaction. The primary challenge was maintaining facial structure while introducing overlapping material density. Organic materials tend to exaggerate softness, requiring constraint reinforcement to preserve silhouette clarity.

The feather system introduces directional texture layering and increased structural detail. Compared to the petal variation, this material required tighter control over density and flow to prevent visual clutter. Motion refinement focused on subtle drift without destabilizing facial features.

The crystal variation shifts from organic layering to rigid refractive geometry. Transparent materials introduced distortion risk, requiring repeated constraint reinforcement to preserve facial proportions and structural integrity. Light behavior and reflection balance became the primary refinement focus.

Selected renders were animated with subtle looped motion including:

Controlled breathing

Petal expansion and contraction

Feather drift

Light shimmer

Motion refinement focused on:

Reducing visible loop seams

Slowing cadence for realism

Minimizing exaggerated AI-driven motion artifacts

Preserving facial integrity during animation

Subtle movement proved more effective than dramatic motion.

Generative systems exaggerate expressive elements unless tightly constrained.

Anatomical consistency must be reinforced repeatedly in prompts.

Clean background separation improves animation stability.

Less motion produces more believable results.

Surface material transformations are easier to control than structural transformations.

This project reflects my broader approach to working with generative systems:

Define constraints clearly

Iterate deliberately

Control output behavior

Refine until stable

Whether designing business systems or AI-driven visuals, my work emphasizes structure within experimentation.