Beyond the Brush: Fully-automated Crafting of Realistic Inpainted Images

The generation of partially manipulated images is rapidly becoming a significant threat to the public’s trust in online content. The proliferation of diffusion model-based tools that enable easy inpainting operations has significantly lowered the barrier to accessing these techniques. In this context, the multimedia forensics community finds itself at a disadvantage compared to attackers, as developing new localization techniques often requires the creation of large datasets, a resource-intensive process due to the necessary human effort. In this paper, we introduce BtB, a technique that automates the generation of large datasets containing convincing inpainted images. Moreover, we provide a collection of 22,167 high-quality inpainted pictures obtained by manipulating images from state-of-the-art datasets. In addition to the generation pipeline, which can be executed without any human intervention, we also present an evaluation of the quality of produced images from both quantitative and qualitative perspectives. Our results demonstrate that our technique yields higher quality images compared to simpler automated generation mechanisms.