PANDORA: Pixel-wise Attention Dissolution and Latent Guidance for Zero-Shot Object Removal
For image editing tasks, this work provides a practical zero-shot method that avoids the need for fine-tuning or prompt engineering, addressing scalability and artifact issues in object removal.
PANDORA proposes a zero-shot object removal framework for pre-trained diffusion models that eliminates objects by nullifying attention keys and guiding denoising, achieving superior visual fidelity and semantic plausibility without fine-tuning, prompts, or optimization.
Removing objects from natural images is challenging due to difficulty of synthesizing semantically coherent content while preserving background integrity. Existing methods often rely on fine-tuning, prompt engineering, or inference-time optimization, yet still suffer from texture inconsistency, rigid artifacts, weak foreground-background disentanglement, and poor scalability for multi-object removal. We propose a novel zero-shot object removal framework, namely PANDORA, that operates directly on pre-trained text-to-image diffusion models, requiring no fine-tuning, prompts, or optimization. We propose Pixel-wise Attention Dissolution to remove object by nullifying the most correlated attention keys for masked pixels, effectively eliminating the object from self-attention flow and allowing background context to dominate reconstruction. We further introduce Localized Attentional Disentanglement Guidance to steer denoising toward latent manifolds favorable to clean object removal. Together, these components enable precise, non-rigid, prompt-free, and scalable multi-object erasure in a single pass. Experiments demonstrate superior visual fidelity and semantic plausibility compared to state-of-the-art methods. The project page is available at https://vdkhoi20.github.io/PANDORA.