Where to Look Is Not How to Fix: Pre-Denoising Diagnostics and Modality-Dependent Control in Diffusion Composition
cs.CV
Submitted: 2026-10-02
Updated: 2026-10-02
Sources
- Understanding intermediate layers using linear classifier probes
- Training-Free Structured Diffusion Guidance for Compositional Text-to-Image Synthesis
- Progressive Compositionality in Text-to-Image Generative Models
- Prompt-to-Prompt Image Editing with Cross Attention Control
- Unlocking the Potential of Text-to-Image Diffusion with PAC-Bayesian Theory
- Divide & Bind Your Attention for Improved Generative Semantic Nursing
- Attention Overlap Is Responsible for The Entity Missing Problem in Text-to-image Diffusion Models!
- Diffusion Beats Autoregressive: An Evaluation of Compositional Generation in Text-to-Image Models
- SDXL: Improving Latent Diffusion Models for High-Resolution Image Synthesis
- Hierarchical Text-Conditional Image Generation with CLIP Latents
- Improving Compositional Attribute Binding in Text-to-Image Generative Models via Enhanced Text Embeddings
Related papers
- Loss Knows Best: Detecting Annotation Errors in Videos via Loss Trajectories
- AnchorWeave: World-Consistent Video Generation with Retrieved Local Spatial Memories
- Benchmarking the Robustness of Foundation Models for Mammography under Domain Shift
- MambaX-Net: Dual-Input Mamba-Enhanced Cross-Attention Network for Longitudinal MRI Segmentation
- TeleOCR: Navigating Document Parsing Across Digital and Camera-Captured Documents
- A Survey on Efficient Vision-Language-Action Models