Track papers across labs, topics, and trends.
Search papers, labs, and topics across Lattice.
EditCrafter is a novel high-resolution image editing method that utilizes pretrained text-to-image diffusion models without the need for tuning. By employing tiled inversion and a noise-damped manifold-constrained classifier-free guidance (NDCFG++), it effectively preserves the identity of high-resolution images while allowing for flexible editing across various resolutions. The results demonstrate that EditCrafter can achieve high-quality edits at resolutions beyond those used during training, overcoming limitations faced by existing methods.
EditCrafter enables high-resolution image editing without tuning, achieving impressive results that defy the constraints of traditional diffusion models.
We propose EditCrafter, a high-resolution image editing method that operates without tuning, leveraging pretrained text-to-image (T2I) diffusion models to process images at resolutions significantly exceeding those used during training. Leveraging the generative priors of large-scale T2I diffusion models enables the development of a wide array of novel generation and editing applications. Although numerous image editing methods have been proposed based on diffusion models and exhibit high-quality editing results, they are difficult to apply to images with arbitrary aspect ratios or higher resolutions since they only work at the training resolutions (512x512 or 1024x1024). Naively applying patch-wise editing fails with unrealistic object structures and repetition. To address these challenges, we introduce EditCrafter, a simple yet effective editing pipeline. EditCrafter operates by first performing tiled inversion, which preserves the original identity of the input high-resolution image. We further propose a noise-damped manifold-constrained classifier-free guidance (NDCFG++) that is tailored for high resolution image editing from the inverted latent. Our experiments show that the our EditCrafter can achieve impressive editing results across various resolutions without fine-tuning and optimization.