Research map / Image generation / Text-to-image diffusion
Accelerated Image Editing via Consistency-Aware Source Token Pruning
Jeongsol Kim, Jong Chul Ye, Yufei Wang, Jian Wang
Read the paperCheck your idea against 137,217 papers
AI-extracted summary. Written by AI from the paper's abstract only. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Proposes TokenDrop, an efficient image-editing framework that prunes source image tokens in large multimodal diffusion transformers to cut attention overhead while preserving editing consistency.
Problem
Text-based image editing in large multimodal transformers concatenates source image tokens with text and noise tokens, incurring heavy attention computation overhead.
Approach
Shifts computational burden away from dense source-image token processing by pruning source tokens in a consistency-aware way, rather than processing all source tokens at every layer.
Closest papers by meaning
- Improving Editability in Image Generation with Layer-wise Memory (CVPR 2025 · similarity 0.88)
- Highly Compressed Tokenizer Can Generate Without Training (ICML 2025 · similarity 0.88)
- NEP: Autoregressive Image Editing via Next Editing Token Prediction (NeurIPS 2025 · similarity 0.87)
- Towards Efficient Diffusion-Based Image Editing with Instant Attention Masks (AAAI 2024 · similarity 0.87)
- Invertible Consistency Distillation for Text-Guided Image Editing in Around 7 Steps (NeurIPS 2024 · similarity 0.87)
- Inversion-Free Image Editing with Language-Guided Diffusion Models (CVPR 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.
