GIDE: Unlocking Diffusion LLMs for Precise Training-Free Image Editing

TL;DR

GIDE introduces a novel, training-free image editing framework for diffusion large language models, achieving high-fidelity, precise edits while preserving unedited regions, supported by a new benchmark and extensive experimental validation.

Contribution

GIDE presents a Discrete Noise Inversion technique and a multi-stage editing pipeline, enabling accurate, versatile, and training-free image editing with a new comprehensive benchmark.

Findings

Outperforms prior methods with 51.83% improvement in semantic correctness.

Achieves 50.39% higher perceptual quality.

Demonstrates broad applicability and photorealistic results.

Abstract

While Diffusion Large Language Models (DLLMs) have demonstrated remarkable capabilities in multi-modal generation, performing precise, training-free image editing remains an open challenge. Unlike continuous diffusion models, the discrete tokenization inherent in DLLMs hinders the application of standard noise inversion techniques, often leading to structural degradation during editing. In this paper, we introduce GIDE (Grounded Inversion for DLLM Image Editing), a unified framework designed to bridge this gap. GIDE incorporates a novel Discrete Noise Inversion mechanism that accurately captures latent noise patterns within the discrete token space, ensuring high-fidelity reconstruction. We then decompose the editing pipeline into grounding, inversion, and refinement stages. This design enables GIDE supporting various editing instructions (text, point and box) and operations while strictly preserving the unedited background. Furthermore, to overcome the limitations of existing single-step evaluation protocols, we introduce GIDE-Bench, a rigorous benchmark comprising 805 compositional editing scenarios guided by diverse multi-modal inputs. Extensive experiments on GIDE-Bench demonstrate that GIDE significantly outperforms prior training-free methods, improving Semantic Correctness by 51.83% and Perceptual Quality by 50.39%. Additional evaluations on ImgEdit-Bench confirm its broad applicability, demonstrating consistent gains over trained baselines and yielding photorealistic consistency on par with leading models.