Generative Vision Atlas

Representation option

Frozen encoder + trained decoder

Keep a pretrained vision encoder entirely frozen as the latent space; train only a lightweight decoder to map that space back to pixels.

Introduced by

Used by (3)

RAE, RAEv2, Scale-RAE

Alternatives on this axis

Continuous tokens, Deep-compression latent, Hybrid semantic + detail codebook, Multi-scale (next-scale) tokens, Pixels, Semantic / foundation latent, VAE latent

← All Representation options