Decoder. plain-English AI glossary

Avatar Generation

▲ Rising

Creating a digital character or persona - either from text description or by personalizing a template.

Think of it like

Character design: an artist sketches a character; the model generates a 3D or 2D avatar automatically.

Example

"A professional woman, brown hair, glasses, smiling" - generates a realistic avatar. Or: input a photo -> generates anime-style version of that person.

How it actually works

2D (image, video) vs. 3D (mesh, point cloud). Text-to-avatar: generate image/model from prompt. Photo-to-avatar: style transfer or 3D reconstruction. Challenges: identity preservation, realism, animation (can the avatar move naturally?). Methods: GANs, diffusion models, NeRF for 3D. Real-time avatars (gaming) vs. high-quality offline (cinema).

For product teams

Powers gaming, metaverse, virtual events, and personalized digital representations.

For engineers

2D: text-to-image + style transfer. 3D: text-to-3D (NeRF, diffusion) or photogrammetry. Models: DreamFusion.

Read anything AI without the jargon

Look up any term in plain English, or save terms as you read with the free Chrome extension.

Open DecoderAdd to Chrome