In&amp;Out : Diverse Image Outpainting via GAN Inversion

Cheng, Yen-Chi; Lin, Chieh Hubert; Lee, Hsin-Ying; Ren, Jinchang; Tulyakov, Sergey; Yang, Ming–Hsuan

doi:10.48550/arxiv.2104.00675

Cited by 7 publications

(17 citation statements)

References 39 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We extend MaskGIT to this problem by tokenizing the masked image and interpreting the inpainting mask as the initial mask in our iterative decoding. We then composite the output image by linearly blending it with the input based on the masking boundary following [8]. To match the training of our baselines, we train MaskGIT on the 512ˆ512 center-cropped images from the Places2 [58] dataset.…”

Section: Image Editing Applicationsmentioning

confidence: 99%

See 1 more Smart Citation

MaskGIT: Masked Generative Image Transformer

Chang¹,

Zhang²,

J³

et al. 2022

Preprint

View full text Add to dashboard Cite

Figure 1. Example generation by MaskGIT on image synthesis and manipulation tasks. We show that MaskGIT is a flexible model that can generate high-quality samples on (a) class-conditional synthesis, (b) class-conditional image manipulation, e.g. replacing selected objects in the bounding box with ones from the given classes, and (c) image extrapolation. Examples shown here have resolutions 512ˆ512, 512ˆ512, and 512ˆ2560 in the three columns, respectively. Zoom in to see the details.

show abstract

Section: Image Editing Applicationsmentioning

confidence: 99%

“…We compare against common GAN-based baselines, including Boundless [43], In&Out [8], InfinityGAN [31], and CoModGAN [57] on extrapolating rightward with a 50% ratio. We evaluate on the image set generously provided by the authors of InfinityGAN [31] and In&Out [8].…”

Section: Image Editing Applicationsmentioning

confidence: 99%

MaskGIT: Masked Generative Image Transformer

Chang¹,

Zhang²,

J³

et al. 2022

Preprint

View full text Add to dashboard Cite

show abstract

“…Our Resize Source < l a t e x i t s h a 1 _ b a s e 6 4 = " T C L G 2 A Z q l a a 5 R y x z g I R / d 0 p I Panorama generation Panorama generation aims to generate a sequence of continuous images in an unconditional setting [18,20,21] or conditioning on given images [17]. These methods perform generation conditioning on a coordinate system.…”

Section: Naïve Resize Naïve Resizementioning

confidence: 99%

“…Furthermore, various conventional conditional image generation tasks can be achieved with the help of the inversion techniques. For example, image-to-image translation can be done by injecting encoded features to StyleGANs [14,15], and image inpainting and outpainting can realized by locating the appropriate codes in the latent space [16,17,18]. However, most methods either are designed in a task-specific manner or require additional architectures.…”

Section: Introductionmentioning

confidence: 99%

StyleGAN of All Trades: Image Manipulation with Only Pretrained StyleGAN

Chong¹,

Lee²,

Forsyth³

2021

Preprint

Self Cite

View full text Add to dashboard Cite

Recently, StyleGAN has enabled various image manipulation and editing tasks thanks to the high-quality generation and the disentangled latent space. However, additional architectures or task-specific training paradigms are usually required for different tasks. In this work, we take a deeper look at the spatial properties of StyleGAN. We show that with a pretrained StyleGAN along with some operations, without any additional architecture, we can perform comparably to the state-ofthe-art methods on various tasks, including image blending, panorama generation, generation from a single image, controllable and local multimodal image to image translation, and attributes transfer. The proposed method is simple, effective, efficient, and applicable to any existing pretrained StyleGAN model.Preprint. Under review.

show abstract

“…Semantic image synthesis [6,28,37] approaches leverage the input semantic segmentation map that contains the structural specification and generate correlated results by filling categorical content.Image outpainting [1,18,30,40] is related to image inpainting [19,41] and shares similar issues that the generator tends to copy-andparaphrase the conditional input or create mottled textural samples, leading to repetitive results especially when the outpainted region is large. InOut [4] proposes to outpaint image with GANs inversion and yield results with higher diversity. We show that with InfinityGAN as the deep image prior along with [4], we obtain the state-of-the-art outpaint- 𝐿 𝑎𝑟 (Equation 5)…”

Section: Related Workmentioning

confidence: 99%

InfinityGAN: Towards Infinite-Pixel Image Synthesis

Lin¹,

Lee²,

Cheng³

et al. 2021

Preprint

Self Cite

View full text Add to dashboard Cite

Training Patch Size (101x101) Figure 1: Synthesizing infinite-resolution images from finite-resolution inputs. A 1024×2048 image composed of 242 patches, independently synthesized by InfinityGAN with spatial fusion of two styles. The generator is trained on 101×101 patches (e.g., marked in top-left) sampled from 197×197 real images. Note that training and inference (of any resolution) are performed on a single GTX TITAN X GPU. Zoom-in for better experience.

show abstract

In&Out : Diverse Image Outpainting via GAN Inversion

Cited by 7 publications

References 39 publications

MaskGIT: Masked Generative Image Transformer

MaskGIT: Masked Generative Image Transformer

StyleGAN of All Trades: Image Manipulation with Only Pretrained StyleGAN

InfinityGAN: Towards Infinite-Pixel Image Synthesis

Contact Info

Product

Resources

About