Generative Vision Atlas

Research line · ascendant

Representation-aligned latents

Keep the autoencoder, but apply pressure from a vision foundation model so its latent (or the generator's hidden states) become semantically structured.

What defines membership

The VAE is worth keeping; what needs fixing is the absence of semantic structure, and an alignment loss can supply it.

How the line developed

Read top to bottom: what came before, the idea itself, the evidence for it, what improved, and where it breaks.

The idea

REPA · 2024-10landmark

Aligning a diffusion transformer's hidden states to frozen DINOv2 features speeds training 17.5x with no change to the latent or architecture.

Evidence

MAETok · 2025-02core

Reframes the target as latent geometry: fewer Gaussian-mixture modes and more discriminative features, no variational form required.

Improvement

VA-VAE / LightningDiT · 2025-01core

Moves the alignment pressure into VAE training itself, and names the reconstruction-generation dilemma explicitly.

REPA-E · 2025-04core

Makes VAE and diffusion transformer trainable end to end through the alignment loss, which naive joint training destroys.

REG · 2025-07core

Entangles a semantic token into the denoising target rather than aligning hidden states as a side loss.

What it gets right

  • Large speedups with minimal disruption to an existing, well-understood stack
  • Directly measurable: convergence-rate gains are large and reproducible
  • Compatible with the entire VAE-latent ecosystem

Where it is weak

  • Treats the symptom rather than the cause — the latent is still a reconstruction code
  • Adds a dependency on a frozen encoder anyway, which invites the question RAE asks

Reported numbers

As published, with the guidance method, budget, and model size that produced them. Seethe comparison page for why these cannot be ranked naively.

PaperBenchmarkValueGuidanceBudgetParams
MAETokImageNet 256x256 gFID2.31none4M steps675M
MAETokImageNet 256x256 gFID1.67cfg=2.0, 250 steps4M steps675M generator + 176M tokenizer
REPAImageNet 256x256 gFID5.9none4M iters675M
REPAImageNet 256x256 gFID1.42guidance-interval800 epochs (~4M iters)675M
REPA-EImageNet 256x256 gFID1.69none800 epochs675M
REPA-EImageNet 256x256 gFID1.12CFG (scale not stated)800 epochs675M
VA-VAE / LightningDiTImageNet 256x256 gFID2.17none800 epochs675M
VA-VAE / LightningDiTImageNet 256x256 gFID1.35cfg-interval + timestep shift (scale not stated)800 epochs675M
MAETokImageNet 256x256 rFID0.48
VA-VAE / LightningDiTImageNet 256x256 rFID0.28

Competing answers

Open problems it has not solved

Also in this line

REPA spatial-structure study