Research map / Image generation / Text-to-image diffusion
Breaking the Lock-in: Diversifying Text-to-Image Generation via Representation Modulation
Dahee Kwon, Haeun Lee, Jaesik Choi
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 |
|---|---|---|---|
| apply DAVE DC attenuation in early denoising steps | SD3.5 · ImageNet | Vendi: 1.71 -> 2.33 | Improved |
| apply DAVE DC attenuation in early denoising steps | SD3.5 · MS-COCO | FID: 36.38 -> 29.56 | Improved |
| apply DAVE in the in-batch setting versus SPELL, DiverseFlow, Oscar, PG | SD3.5 · MS-COCO in-batch | Recall: 0.255 -> 0.441 | Improved |
| sample intervention block randomly from pool L instead of one fixed block (ablation) | DAVE · ImageNet | Vendi: 2.33 -> 2.50 (SD3.5) | Improved |
| lengthen the intervention window (larger temporal cutoff tau) (ablation) | DAVE · SD3 | structural variation vs image quality | Mixed |
| lower attenuation strength alpha (ablation) | DAVE · SD3 | LPIPS | Mixed |
| amplify DC component (alpha > 1) (ablation) | DAVE · SD3 | structural change | No clear effect |
| calibrate block pool L on a different dataset (ablation) | DAVE · SD3.5 MS-COCO/ImageNet | Vendi: 2.0275 -> 1.8840 | No clear effect |
| apply DAVE on a distilled model | SD3.5-Large-Turbo · Distilled sampling | Vendi: 1.383 -> 1.643 | Improved |
| combine DAVE with higher CFG scales | SD3 · CFG 7 | Vendi: 3.385 -> 3.828 | Improved |
| apply DAVE at CFG 3 | SD3 · CFG 3 | Vendi: 2.540 -> 2.307 | Hurt |
| apply DAVE to complex prompts | SD3 · PartiPrompts | Vendi: 1.511 -> 2.006 (Comp.) | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- CADS — CADS: Unleashing the Diversity of Diffusion Models through Condition-Annealed Sampling (ICLR 2024) · better on 2 of 2 shared results
- DiverseFlow — DiverseFlow: Sample-Efficient Diverse Mode Coverage in Flows (CVPR 2025) · better on 1 of 1 shared result
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Meissonic: Revitalizing Masked Generative Transformers for Efficient High-Resolution Text-to-Image Synthesis (ICLR 2025 · similarity 0.88)
- Rare Text Semantics Were Always There in Your Diffusion Transformer (NeurIPS 2025 · similarity 0.87)
- Rethinking Training for De-biasing Text-to-Image Generation: Unlocking the Potential of Stable Diffusion (CVPR 2025 · similarity 0.87)
- On the Scalability of Diffusion-based Text-to-Image Generation (CVPR 2024 · similarity 0.87)
- Exploring the Deep Fusion of Large Language Models and Diffusion Transformers for Text-to-Image Synthesis (CVPR 2025 · similarity 0.87)
- Towards Transformer-Based Aligned Generation with Self-Coherence Guidance (CVPR 2025 · 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.
