Research map / Image generation / Text-to-image diffusion
Intermediate Text Representation Guided Text-to-Image Generation for Enhancing One-and-Only Alignment
Soyoun Won, Aryan Yazdan Parast, Basim Azam, Jean Honorio, NAVEED AKHTAR
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 |
|---|---|---|---|
| add IR-guidance (intermediate text representation) | SD3.0 · OAO-AttackBench Landmark Material | VQAScore: 0.525 -> 0.716 | Improved |
| add IR-guidance | SD3.0 · OAO-AttackBench | VQAScore: up to a 19.1 percentage-point improvement | Improved |
| add IR-guidance | SD2.1 · OAO-AttackBench Artwork Style | VQAScore: 0.563 -> 0.622 | Improved |
| add IR-guidance | SD3.0 · OAO-AttackBench | CLIPScore | No clear effect |
| add IR-guidance | SD3.0 · OAO-AttackBench | HPSv2: 0.269 -> 0.273 | Improved |
| add IR-guidance | SD3.0 · Gecko(S) | VQAScore: 0.744 -> 0.748 | No clear effect |
| add IR-guidance | SD3.0 · OAO-AttackBench Celestial Objects | KID: 0.0576 -> 0.0483 | Improved |
| use additive injection instead of using c_h directly (None) (ablation) | IR-guided diffusion · OAO-AttackBench Landmark Material (SD3.0) | VQAScore: 0.630 -> 0.716 | Improved |
| use additive injection instead of Mean-Std (ablation) | IR-guided diffusion · OAO-AttackBench Landmark Material (SD3.0) | VQAScore: 0.617 -> 0.716 | Improved |
| use additive injection instead of Projection (ablation) | IR-guided diffusion · OAO-AttackBench Landmark Material (SD3.0) | VQAScore: 0.658 -> 0.716 | Improved |
| use adaptive scheduling instead of fixed schedule (ablation) | IR-guided diffusion · OAO-AttackBench and Whoops | VQAScore and CLIPScore | Improved |
| use small injection strength lambda in [0.1, 0.5] (ablation) | IR-guided diffusion · OAO-AttackBench and Whoops | VQAScore | Improved |
Closest papers by meaning
- Localizing and Editing Knowledge In Text-to-Image Generative Models (ICLR 2024 · similarity 0.89)
- Progressive Text-to-Image Diffusion with Soft Latent Direction (AAAI 2024 · similarity 0.89)
- The Hidden Language of Diffusion Models (ICLR 2024 · similarity 0.89)
- Information Theoretic Text-to-Image Alignment (ICLR 2025 · similarity 0.88)
- Aligning Text to Image in Diffusion Models is Easier Than You Think (NeurIPS 2025 · similarity 0.88)
- A Cat Is A Cat (Not A Dog!): Unraveling Information Mix-ups in Text-to-Image Encoders through Causal Analysis and Embedding Optimization (NeurIPS 2024 · similarity 0.88)
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.
