AI Paper Atlas

Research map / Image generation / Autoregressive and latent generation

Domain Expansion of Image Generators

Yotam Nitzan, Michaël Gharbi, Richard Zhang, Taesung Park, Jun-Yan Zhu, Daniel Cohen-Or, Eli Shechtman

CVPR 2023 · Image generation › Autoregressive and latent generation · approach: Generative latent space tooling · 13 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
Expand the generator with StyleGAN-NADA instead of using a separate adapted generatorStyleGAN2 · StyleGAN-NADA image comparisonUser preference: 41.2% -> 58.8%Improved
Expand the generator with MyStyle instead of using a separate adapted generatorStyleGAN2 · MyStyle image comparisonID: 0.80 ± 0.06 -> 0.76 ± 0.05No clear effect
Expand the generator with MyStyle instead of using a separate adapted generatorStyleGAN2 · MyStyle image comparisonDiversity: 3.08 ± 0.15 -> 3.14 ± 0.14No clear effect
Expand the generator with StyleGAN-NADA instead of using a separate adapted generatorStyleGAN2 · StyleGAN-NADA image comparisonDiversity: 2.42 ± 0.13 -> 2.42 ± 0.13No clear effect
Expand the source generator with new domains while preserving its base subspaceStyleGAN2 · FFHQFID: 2.77 -> 2.80No clear effect
Expand the source generator with new domains while preserving its base subspaceStyleGAN2 · AFHQ DogFID: 7.43 -> 7.51No clear effect
Expand the source generator with new domains while preserving its base subspaceStyleGAN2 · LSUN ChurchFID: 3.92 -> 3.76No clear effect
Expand the source generator with new domains while preserving its base subspaceStyleGAN2 · SD-ElephantFID: 2.30 -> 2.70No clear effect
Increase the expansion from 105 to 400 new domains (ablation)FFHQ StyleGAN2 · FFHQ base subspaceFID: 2.80 -> 2.83No clear effect
Use a larger distance between base and repurposed subspaces (ablation)FFHQ StyleGAN2 expansion · New-domain CLIP alignmentCLIP error: s=5; s=20; s=50Improved
Set the base-to-repurposed distance too large (ablation)FFHQ StyleGAN2 expansion · New-domain generationImage quality: s = 50Hurt
Set the base-to-repurposed distance too small (ablation)FFHQ StyleGAN2 expansion · New-domain generationTarget-domain representation: s ≤ 5Hurt

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