Research map / Image generation / Text-to-image diffusion
Create Your World: Lifelong Text-to-Image Diffusion
Gan Sun, Wenqi Liang, Jiahua Dong, Jun Li, Zhengming Ding, Yang Cong
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 changed | Where | Measured | Effect |
|---|---|---|---|
| use L2DM instead of SVDiff+LWF | SVDiff · Lifelong single-concept generation | IA avg: 79.1 -> 80.4 | Improved |
| use L2DM instead of SVDiff+LWF | SVDiff · Lifelong single-concept generation | TA avg: 28.3 -> 30.1 | Improved |
| use L2DM instead of Dreambooth+FT | Dreambooth · Lifelong single-concept generation | IA avg: 74.1 -> 80.4 | Improved |
| use L2DM instead of Dreambooth+FT | Dreambooth · Lifelong single-concept generation | TA avg: 23.8 -> 30.1 | Improved |
| use L2DM instead of SVDiff+LWF | SVDiff · Lifelong multi-concept generation | IA avg: 70.6 -> 75.1 | Improved |
| use L2DM instead of SVDiff+EWC | SVDiff · Lifelong multi-concept generation | IA avg: 72.8 -> 75.1 | Improved |
| use L2DM instead of SVDiff+LWF | SVDiff · Lifelong single-concept task forgetting | TFR-IA: 3.0 -> 1.2 | Improved |
| use L2DM instead of Dreambooth+FT | Dreambooth · Lifelong single-concept task forgetting | TFR-IA: 8.3 -> 1.2 | Improved |
| add TAME and ECD forgetting regularization modules (ablation) | L2DM · Lifelong single-concept generation | image alignment on earlier concepts | Improved |
| add Elastic Concept Distillation (ECD) to model with TAME (ablation) | L2DM · Lifelong single-concept generation | color detail of target concept | Mixed |
| add concept attention artist (CAA) (ablation) | L2DM · Lifelong multi-concept generation | text alignment | Improved |
| add orthogonal attention artist (OAA) (ablation) | L2DM · Lifelong multi-concept generation | attribute binding | Improved |
Closest papers by meaning
- How to Continually Adapt Text-to-Image Diffusion Models for Flexible Customization? (NeurIPS 2024 · similarity 0.91)
- ConceptGuard: Continual Personalized Text-to-Image Generation with Forgetting and Confusion Mitigation (CVPR 2025 · similarity 0.89)
- DreamDistribution: Learning Prompt Distribution for Diverse In-distribution Generation (ICLR 2025 · similarity 0.88)
- Concept Weaver: Enabling Multi-Concept Fusion in Text-to-Image Models (CVPR 2024 · similarity 0.88)
- Prompt-Free Diffusion: Taking "Text" out of Text-to-Image Diffusion Models (CVPR 2024 · similarity 0.88)
- Mastering Text-to-Image Diffusion: Recaptioning, Planning, and Generating with Multimodal LLMs (ICML 2024 · similarity 0.87)
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.
