Research map / Image generation / Text-to-image diffusion
Stable Diffusion is Unstable
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 ATM rather than random prompt perturbations | Stable Diffusion, short prompts · ImageNet-1K short prompts | success rate: 79.2 -> 91.1 | Helped |
| Use ATM rather than random prompt perturbations | Stable Diffusion, long prompts · ImageNet-1K long prompts | success rate: 41.4 -> 81.2 | Helped |
| Use 100 rather than 50 search steps (ablation) | ATM · ImageNet-1K long prompts | success rate: 68.7 -> 81.2 | Helped |
| Use 150 rather than 100 search steps (ablation) | ATM · ImageNet-1K long prompts | success rate: 81.2 -> 67.2 | Hurt |
| Add a fluency constraint (ablation) | ATM without constraints · ImageNet-1K long prompts | TS: 0.37 -> 0.79 | Helped |
| Add a fluency constraint (ablation) | ATM without constraints · ImageNet-1K long prompts | success rate: 91.3 -> 81.7 | Hurt |
| Add a BERTScore constraint (ablation) | ATM without constraints · ImageNet-1K long prompts | TS: 0.37 -> 0.51 | Helped |
| Add a BERTScore constraint (ablation) | ATM without constraints · ImageNet-1K long prompts | success rate: 91.3 -> 89.8 | Hurt |
| Use fluency and BERTScore constraints together rather than fluency alone (ablation) | ATM · ImageNet-1K long prompts | TS: 0.79 -> 0.84 | Helped |
| Replace DDIM with DPM-Solver (ablation) | ATM · ImageNet-1K long prompts | success rate: 81.2 -> 76.5 | Hurt |
| Use ATM rather than random perturbations | Stable Diffusion, short prompts · ImageNet-1K short prompts | TS: 0.69 -> 0.72 | Helped |
| Use ATM rather than random perturbations | Stable Diffusion, long prompts · ImageNet-1K long prompts | TS: 0.94 -> 0.84 | Hurt |
Closest papers by meaning
- On Discrete Prompt Optimization for Diffusion Models (ICML 2024 · similarity 0.86)
- Discovering Failure Modes of Text-guided Diffusion Models via Adversarial Search (ICLR 2024 · similarity 0.86)
- Token by Token, Compromised: Backdoor Vulnerabilities in Unified Autoregressive Models (NeurIPS 2026 · similarity 0.86)
- Intriguing Properties of Diffusion Models: An Empirical Study of the Natural Attack Capability in Text-to-Image Generative Models (CVPR 2024 · similarity 0.86)
- Masked Diffusion Models are Secretly Time-Agnostic Masked Models and Exploit Inaccurate Categorical Sampling (ICLR 2025 · similarity 0.85)
- Distraction is All You Need: Memory-Efficient Image Immunization against Diffusion-Based Image Editing (CVPR 2024 · similarity 0.85)
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.