AI Paper Atlas

Research map / Image generation / Autoregressive and latent generation

Binary Latent Diffusion

Ze Wang, Jiang Wang, Zicheng Liu, Qiang Qiu

CVPR 2023 · Image generation › Autoregressive and latent generation · approach: Latent diffusion with learned tokenizers · 18 citations (OpenAlex)

Read the paper

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
Use binary latent diffusion with 64 steps instead of the strongest reported baselineImage generation · LSUN Churches 256×256FID: 3.86 -> 4.36Hurt
Use binary latent diffusion with 64 steps instead of the strongest reported baselineImage generation · FFHQ 256×256FID: 4.16 -> 5.85Hurt
Use binary latent diffusion with 64 steps instead of the strongest reported baselineImage generation · LSUN Bedrooms 256×256FID: 2.35 -> 3.85Hurt
Use binary latent diffusion with 64 steps instead of the strongest reported baselineImage generation · CelebA-HQ 256×256FID: 5.11 -> 6.24Hurt
Use binary latent diffusion instead of absorbing diffusion to model binary latentsBinary latent sampler · LSUN Churches 256×256FID: 5.44 -> 4.36Improved
Use binary latent diffusion instead of absorbing diffusion to model binary latentsBinary latent sampler · FFHQ 256×256FID: 7.64 -> 5.85Improved
Represent image patches with binary latents rather than VQ codesImage autoencoder · LSUN Bedrooms validation, 256×256, 16×16 latent gridPSNR: 20.08 ± 1.84 -> 24.02 ± 2.11Improved
Represent image patches with binary latents rather than VQ codesImage autoencoder · LSUN Bedrooms validation, 256×256, 16×16 latent gridSSIM: 0.62 ± 0.09 -> 0.75 ± 0.06Improved
Use binary latent diffusion sampling instead of LDMImage generation · Sampling speedseconds per sample: 15.68 -> 0.20Improved
Set λ to 0.1 instead of 0 in the combined objective (ablation)Binary Latent Diffusion · LSUN Churches 256×256FID: 5.07 -> 4.36Improved
Set λ to 1.0 instead of 0.1 in the combined objective (ablation)Binary Latent Diffusion · LSUN Churches 256×256FID: 4.36 -> 5.12Hurt
Predict the binary residual zᵗ ⊕ z⁰ rather than z⁰ (ablation)Binary Latent Diffusion · LSUN Churches 256×256FID: 4.72 -> 4.36Improved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 0 of 2 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy