AI Paper Atlas

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

ImageGem: In-the-wild Generative Image Interaction Dataset for Generative Model Personalization

Yuanhe Guo, Linxi Xie, Zhuoran Chen, Kangrui Yu, Ryan Po, Guandao Yang, Gordon Wetzstein, Hongyi Wen

ICCV 2025 · Image generation › Text-to-image diffusion · approach: Preference optimization for diffusion · 1 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
fine-tune with ImageGem subset instead of Pick-a-Pic subsetSD1.5 DiffusionDPO · DiffusionDPO OOD prompts (Cars)HPSv2: 0.2690 -> 0.2741Improved
fine-tune with ImageGem Dogs instead of Pick-a-pic DogsSD1.5 DiffusionDPO · DiffusionDPO OOD prompts (Dogs)HPSv2: 0.2739 -> 0.2789Improved
fine-tune with ImageGem Scenery Small instead of Pick-a-pic ScenerySD1.5 DiffusionDPO · DiffusionDPO OOD prompts (Scenery)Pick Score: 0.1936 -> 0.1949Improved
scale up scenery subset from small to large (ablation)SD1.5 DiffusionDPO · DiffusionDPO OOD prompts (Scenery)HPSv2: 0.2730 -> 0.2747Improved
scale up cars subset from small to large (ablation)SD1.5 DiffusionDPO · DiffusionDPO OOD prompts (Cars)CLIP Score: 0.3745 -> 0.3710Hurt
DiffusionDPO fine-tuning on scenery vs original SD1.5SD1.5 · DiffusionDPO OOD prompts (Scenery)CLIP Score: 0.3446 -> 0.3427Hurt
two-tower user embedding instead of ItemKNNImage retrieval · Images-1M retrievalRecall@10000: 0.4705 -> 0.5157Improved
SASRec self-attention over user sequences instead of ItemKNNModel retrieval · Models-200K rankingRecall@10: 0.0773 -> 0.1839Improved
VLM ranking instead of SASRecModel recommendation · Model recommendation (20 users)NDCG@5: 0.2795 -> 0.4981Improved
VLM ranking instead of ItemKNNImage recommendation · Image recommendation (20 users)NDCG@5: 0.7065 -> 0.6745Hurt
SVD top-1 rank reduction of user LoRA (ablation)SDXL LoRA · SVD preliminary (10178 SDXL LoRAs)CLIP similarity to base: 0.7563 -> 0.8114No clear effect
build W2W space from FF layers only (ablation)W2W editing · ani-real editingqualitative editing qualityHurt

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