# LatentEdit: Adaptive Latent Control for Consistent Semantic Editing

## TL;DR

LatentEdit is a lightweight, adaptive latent fusion framework for diffusion-based image editing that achieves high-quality, controllable edits with improved efficiency and without modifying the underlying model.

## Contribution

It introduces a novel adaptive latent fusion method that enhances semantic editing control and efficiency without internal model modifications or complex mechanisms.

## Key findings

- Outperforms state-of-the-art methods in fidelity and editability.
- Achieves comparable results with fewer steps (8-15).
- Inversion-free variant significantly improves real-time deployment efficiency.

## Abstract

Diffusion-based Image Editing has achieved significant success in recent years. However, it remains challenging to achieve high-quality image editing while maintaining the background similarity without sacrificing speed or memory efficiency. In this work, we introduce LatentEdit, an adaptive latent fusion framework that dynamically combines the current latent code with a reference latent code inverted from the source image. By selectively preserving source features in high-similarity, semantically important regions while generating target content in other regions guided by the target prompt, LatentEdit enables fine-grained, controllable editing. Critically, the method requires no internal model modifications or complex attention mechanisms, offering a lightweight, plug-and-play solution compatible with both UNet-based and DiT-based architectures. Extensive experiments on the PIE-Bench dataset demonstrate that our proposed LatentEdit achieves an optimal balance between fidelity and editability, outperforming the state-of-the-art method even in 8-15 steps. Additionally, its inversion-free variant further halves the number of neural function evaluations and eliminates the need for storing any intermediate variables, substantially enhancing real-time deployment efficiency.