AI Paper Atlas

Research map / Image generation / Text-to-image diffusion

SEGA: Instructing Text-to-Image Models using Semantic Guidance

Manuel Brack, Felix Friedrich, Dominik Hintersdorf, Lukas Struppek, Patrick Schramowski, Kristian Kersting

NeurIPS 2023 · Image generation › Text-to-image diffusion · approach: Text-driven synthesis and semantic guidance · 12 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

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
apply SEGA edit guidance to generated faces (positive guidance)Stable Diffusion v1.5 · Face attribute editssuccess rate (%): 95.0Improved
apply four simultaneous SEGA edits instead of oneSEGA · Four simultaneous face editssuccess rate: 90.7 (all 4 attributes, Combination 1); 75.6 (Combination 2)No clear effect
edit generated faces with SEGAStable Diffusion v1.5 · FID vs FFHQFID: 117.73 -> 59.86Improved
add SEGA guidance away from inappropriate conceptsStable Diffusion v1.5 · I2P overallprobability of inappropriate content: 0.38 -> 0.11Improved
add SEGA guidance away from inappropriate conceptsDeepFloyd-IF · I2P overallprobability of inappropriate content: 0.38 -> 0.15Improved
use SEGA instead of Composable Diffusion for multi-conditioning editsuser study · Multi-Conditioningsuccess rate (%): 35.00 -> 80.00Improved
use SEGA instead of Composable Diffusion for minor changesuser study · Minor Changessuccess rate (%): 72.00 -> 90.91Improved
use SEGA instead of Composable Diffusion for style transferuser study · Style Transfersuccess rate (%): 100.0 -> 90.00No clear effect
use SEGA instead of Composable Diffusion for concept removaluser study · Concept Removalsuccess rate (%): 35.00 -> 30.00No clear effect
use SEGA instead of Prompt2Prompt/Disentanglementuser study · Overall successsuccess rate (%): 43.25 -> 72.72 (vs Prompt2Prompt); 41.35 -> 72.72 (vs Disentanglement)Improved
use SEGA instead of Composable Diffusion for faithfulness to original compositionuser study · pairwise similarity to originaluser preference (%): 83.33% vs 13.33%Improved

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