Deepfake

A closer look

A convincing synthetic portrait may look entirely ordinary. For an explanatory artwork, matched lighting, aligned facial planes and deliberately revealed seams can make the transformation legible. Glitches are optional metaphors, not a definition.

A technique with many possible appearances

Deepfake is a media technique rather than a historical art movement or a single recognizable palette. An artist might use it in a film, an identity study or a staged portrait. The central artistic question is what the replacement changes: whose expression we read, whose voice we imagine, or how a face carries a role. That question should guide the image before the effects begin.

The term became prominent in 2017, including through abusive face substitutions. Its history matters when a work borrows someone else's likeness. This gallery instead uses invented people. Every image is an AI-generated conceptual illustration; none documents an actual face-swap experiment or an event involving a real subject.

Separate identity from performance

The Centre for Data Ethics and Innovation's 2019 overview distinguishes face replacement, facial reenactment, generated faces and speech synthesis. Replacement changes apparent identity. Reenactment changes the performance of a face without necessarily substituting another identity. These distinctions give an art project more precise goals than a general request for a strange digital portrait.

In a still-image study, decide which visual variables remain stable. A repeated head angle with a changing expression suggests performance. A stable pose and lighting with a changing face suggests identity. Let clothing and background act as controls. If every element changes at once, the viewer cannot tell what the comparison is testing.

Make the transformation readable

Begin with coherent eyes, facial proportions and directional light. A small difference in shadow direction can make a composite feel like a pasted cutout. Hair crossing the cheek, translucent glasses and hands near the face introduce useful visual complexity, but each needs a clear spatial relationship.

For an explanatory illustration, a thin displaced portrait plane can expose the idea of substitution. For a theatrical image, repeated expression panels can suggest a performed identity. These are artistic devices. A floating mask or colored mesh does not reveal how all deepfake systems work, and a smooth face does not establish that an image is authentic.

Context belongs in the artwork

An invented performer allows a project to explore synthetic identity without borrowing a real person's biography. When a consenting performer is involved, explain what was changed and keep their role in the work visible. A clear caption can make the difference between an intelligible fiction and an unsupported claim about someone.

NIST's 2024 technical report treats provenance tracking, synthetic-content labelling and detection as distinct approaches to media transparency. For this gallery, the practical commitment is straightforward: preserve the generation record and describe the images as conceptual studies. A caption supplies context; it is not a forensic certificate.

Use restraint to keep the question visible

This subject works well for editorial illustration, speculative portraiture, installation proposals and stories about mediated identity. Its visual weakness is the easy substitution of neon fragments for an actual idea. Keep one decisive alteration and let the rest of the portrait remain convincing.

The showcase moves between a photographic portrait plane, a sequence of expressions and a sculptural mask. Each offers a different way to think about the relationship between a face and a person. Judge the result by whether the transformation clarifies that relationship, rather than by how many digital effects fit around the eyes.