Noise Map Guidance: Inversion with Spatial Context for Real Image Editing

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Cho, Hansam, Lee, Jonghyun, Kim, Seoung Bum, Oh, Tae-Hyun, Jeong, Yonghyun
Natura: Preprint
Pubblicazione: 2024
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866911772728885248
author Cho, Hansam
Lee, Jonghyun
Kim, Seoung Bum
Oh, Tae-Hyun
Jeong, Yonghyun
author_facet Cho, Hansam
Lee, Jonghyun
Kim, Seoung Bum
Oh, Tae-Hyun
Jeong, Yonghyun
contents Text-guided diffusion models have become a popular tool in image synthesis, known for producing high-quality and diverse images. However, their application to editing real images often encounters hurdles primarily due to the text condition deteriorating the reconstruction quality and subsequently affecting editing fidelity. Null-text Inversion (NTI) has made strides in this area, but it fails to capture spatial context and requires computationally intensive per-timestep optimization. Addressing these challenges, we present Noise Map Guidance (NMG), an inversion method rich in a spatial context, tailored for real-image editing. Significantly, NMG achieves this without necessitating optimization, yet preserves the editing quality. Our empirical investigations highlight NMG's adaptability across various editing techniques and its robustness to variants of DDIM inversions.
format Preprint
id arxiv_https___arxiv_org_abs_2402_04625
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Noise Map Guidance: Inversion with Spatial Context for Real Image Editing
Cho, Hansam
Lee, Jonghyun
Kim, Seoung Bum
Oh, Tae-Hyun
Jeong, Yonghyun
Computer Vision and Pattern Recognition
Text-guided diffusion models have become a popular tool in image synthesis, known for producing high-quality and diverse images. However, their application to editing real images often encounters hurdles primarily due to the text condition deteriorating the reconstruction quality and subsequently affecting editing fidelity. Null-text Inversion (NTI) has made strides in this area, but it fails to capture spatial context and requires computationally intensive per-timestep optimization. Addressing these challenges, we present Noise Map Guidance (NMG), an inversion method rich in a spatial context, tailored for real-image editing. Significantly, NMG achieves this without necessitating optimization, yet preserves the editing quality. Our empirical investigations highlight NMG's adaptability across various editing techniques and its robustness to variants of DDIM inversions.
title Noise Map Guidance: Inversion with Spatial Context for Real Image Editing
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2402.04625