TeRA generates photorealistic 3D avatars from text in 12 seconds by training a latent diffusion model on a compact distilled latent space from a pretrained human reconstruction model.
Structured 3d latents for scalable and versatile 3d gen- eration
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
TeRA: Rethinking Text-guided Realistic 3D Avatar Generation
TeRA generates photorealistic 3D avatars from text in 12 seconds by training a latent diffusion model on a compact distilled latent space from a pretrained human reconstruction model.