PreciseEdit: Strong Yet Stable Semantic Transport for One-Step Image Editing

Key points

• However, existing one-step editing methods often construct a global transport field through simple vector operations between conditional predictions.1
• While such strong transport can effectively drive the image toward the target semantics, it may be released along unreliable directions or over non-edited regions, leading to structural distortions within the editing region and background drift elsewhere.1
• In this work, we revisit one-step image editing from the perspective of the tension between semantic strength and transport stability, and identify a key insight: strong semantic transport is not inherently unstable.1
• Motivated by this insight, we introduce PreciseEdit, a model-agnostic, training-free and inversion-free one-step editing framework that strengthens target-semantic alignment while preserving background fidelity and structural consistency.1

Sources (1)

• [1]PreciseEdit: Strong Yet Stable Semantic Transport for One-Step Image Editing
  arXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 16, 02:33 PM

  Recent one-step text-to-image (T2I) models enable efficient image synthesis, opening up new possibilities for interactive Web-based image editing.

  However, existing one-step editing methods often construct a global transport field through simple vector operations between conditional predictions.

Extractive summary: sentences quoted from the sources.