Filters








3,624 Hits in 5.3 sec

High resolution neural texture synthesis with long range constraints [article]

Nicolas Gonthier and Yann Gousseau and Saïd Ladjal
2020 arXiv   pre-print
However, neural synthesis methods still struggle to reproduce large scale structures, especially with high resolution textures.  ...  Experiments show the interest of the multi-scale scheme for high resolution textures and the interest of combining it with additional constraints for regular textures.  ...  These high resolution (1024 × 1024) textures have been chosen to include both structured and irregular textures. Some of them display strong long-range dependency.  ... 
arXiv:2008.01808v1 fatcat:whoc577dongh3aatn46zbog73e

Deferred Neural Rendering: Image Synthesis using Neural Textures [article]

Justus Thies and Michael Zollhöfer and Matthias Nießner
2019 arXiv   pre-print
To address this challenging problem, we introduce Deferred Neural Rendering, a new paradigm for image synthesis that combines the traditional graphics pipeline with learnable components.  ...  Similar to traditional textures, neural textures are stored as maps on top of 3D mesh proxies; however, the high-dimensional feature maps contain significantly more information, which can be interpreted  ...  We used neural textures with a resolution of 512 × 512 with 4 hierarchy level, containing 16 features per texel and a U-Net with 5 layers as a neural renderer. e quality of our image synthesis approach  ... 
arXiv:1904.12356v1 fatcat:r3vfiidi4fdgpnbr6phlvhsuuu

Super-resolution Using Constrained Deep Texture Synthesis [article]

Libin Sun, James Hays
2017 arXiv   pre-print
We build on recent success in deep learning based texture synthesis and show that this rich feature space can facilitate successful transfer and synthesis of high frequency image details to improve the  ...  visual quality of super-resolution results on a wide variety of natural textures and images.  ...  This feature space constraint has been shown to excel at representing natural image textures for texture synthesis, style transfer, and super-resolution.  ... 
arXiv:1701.07604v1 fatcat:56x37wqlurclfcy5rruwuwireu

Combining Markov Random Fields and Convolutional Neural Networks for Image Synthesis

Chuan Li, Michael Wand
2016 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR)  
Unlike standard MRF-based texture synthesis, the combined system can both match and adapt local features with considerable variability, yielding results far out of reach of classic generative MRF methods  ...  We apply the method to both photographic and non-photo-realistic (artwork) synthesis tasks.  ...  Long range correlations have also been modeled by spatial LTSM neural networks; results so far are still limited to semiregular textures [25] .  ... 
doi:10.1109/cvpr.2016.272 dblp:conf/cvpr/LiW16 fatcat:yikv7hc7lvearbinlowfd7lrby

Combining Markov Random Fields and Convolutional Neural Networks for Image Synthesis [article]

Chuan Li, Michael Wand
2016 arXiv   pre-print
Unlike standard MRF-based texture synthesis, the combined system can both match and adapt local features with considerable variability, yielding results far out of reach of classic generative MRF methods  ...  We apply the method to both photographic and non-photo-realistic (artwork) synthesis tasks.  ...  Long range correlations have also been modeled by spatial LTSM neural networks; results so far are still limited to semiregular textures [25] .  ... 
arXiv:1601.04589v1 fatcat:mbcmayq2k5atxobuamldmm6g2q

A Sliced Wasserstein Loss for Neural Texture Synthesis [article]

Eric Heitz and Kenneth Vanhoey and Thomas Chambon and Laurent Belcour
2021 arXiv   pre-print
It is theoretically proven,practical, simple to implement, and achieves results that are visually superior for texture synthesis by optimization or training generative neural networks.  ...  We address the problem of computing a textural loss based on the statistics extracted from the feature activations of a convolutional neural network optimized for object recognition (e.g. VGG-19).  ...  Neural texture synthesis with spatial constraints. The problem defined in Section 2 aims at capturing the stationary statistics that define a texture.  ... 
arXiv:2006.07229v4 fatcat:mtlqbrhgzjg5thdod367y23zq4

A survey of exemplar-based texture synthesis

Lara Raad, Axel Davy, Agnès Desolneux, Jean-Michel Morel
2018 Annals of Mathematical Sciences and Applications  
This article accounts for the very rapid and impressive recent apparition of new texture synthesis methods with striking results. We shall retrace their theoretical roots.  ...  They produce impressive synthesis on various kinds of textures.  ...  Long range structures are missed and the method tends to homogenize the output texture. Figure 4 shows two synthesis results.  ... 
doi:10.4310/amsa.2018.v3.n1.a4 fatcat:atf736rbi5fitka6lotklfhp2m

A survey of exemplar-based texture synthesis [article]

Lara Raad, Axel Davy, Agnès Desolneux, Jean-Michel Morel
2017 arXiv   pre-print
They produce impressive synthesis on various kinds of textures.  ...  Exemplar-based texture synthesis is the process of generating, from an input sample, new texture images of arbitrary size and which are perceptually equivalent to the sample.  ...  Long range structures are missed and the method tends to homogenize the output texture. Figure 4 shows two synthesis results.  ... 
arXiv:1707.07184v2 fatcat:j5ba4r6perfbpdpzy7gnx6ihyu

Fast and Scalable Earth Texture Synthesis using Spatially Assembled Generative Adversarial Neural Networks [article]

Sung Eun Kim, Hongkyu Yoon, Jonghyun Lee
2020 arXiv   pre-print
However, the texture synthesis with the GANs framework is often limited by the computational cost and scalability of the output texture size.  ...  Recently, generative adversarial neural networks (GANs) have demonstrated a potential of synthesizing input textural images and creating equiprobable geomaterial images.  ...  with statistical properties (e.g., the long-range connectivity) even using a single TI.  ... 
arXiv:2011.06776v1 fatcat:ov2owt2huvebnnqsyt64bd5jj4

Non-Parametric Neural Style Transfer [article]

Nicholas Kolkin
2021 arXiv   pre-print
Finally I will describe a framework inspired by both modern neural style transfer algorithms and traditional patch-based synthesis approaches which is fast, general, and offers state-of-the-art visual  ...  begin by proposing novel definitions of style and content based on optimal transport and self-similarity, and demonstrating how a style transfer algorithm based on these definitions generates outputs with  ...  Textures that do not contain any complex long range structures can be fairly well modeled by parametric texture synthesis models such as [92] (top row).  ... 
arXiv:2108.12847v1 fatcat:v3mtlfi45vccjnh4r2pzuf4ntu

TextureGAN: Controlling Deep Image Synthesis with Texture Patches [article]

Wenqi Xian, Patsorn Sangkloy, Varun Agrawal, Amit Raj, Jingwan Lu, Chen Fang, Fisher Yu, James Hays
2018 arXiv   pre-print
Our generative network learns to synthesize objects consistent with these texture suggestions.  ...  In this paper, we investigate deep image synthesis guided by sketch, color, and texture.  ...  By training TextureGAN with local texture constraints, we demonstrate its effectiveness on sketch and texture-based image synthesis.  ... 
arXiv:1706.02823v3 fatcat:yzwuihw2r5he5hss3jnyos3kma

Learning and synthesizing mpeg-4 compatible 3-d face animation from video sequence

Wen Gao, Yiqiang Chen, Rui Wang, Shiguang Shan, Dalong Jiang
2003 IEEE transactions on circuits and systems for video technology (Print)  
two-dimensional (2-D) visual feature matrix to the representation in 3-D MPEG-4 face animation parameter space, in assistance with the computer vision method.  ...  In face tracking, to reduce the complexity of the tracking process, a novel coarse-to-fine strategy combined with a Kalman filter is proposed for localizing key facial landmarks in each image of the video  ...  Wang for performing the training data and assisting with processing and analyzing the data.  ... 
doi:10.1109/tcsvt.2003.817629 fatcat:ugij6nxvivgthcse53ncpbpg4a

TextureGAN: Controlling Deep Image Synthesis with Texture Patches

Wenqi Xian, Patsorn Sangkloy, Varun Agrawal, Amit Raj, Jingwan Lu, Chen Fang, Fisher Yu, James Hays
2018 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition  
Top row: Sketch with texture patch overlaid. Bottom row: Results from TextureGAN. Abstract In this paper, we investigate deep image synthesis guided by sketch, color, and texture.  ...  Our generative network learns to synthesize objects consistent with these texture suggestions.  ...  By training TextureGAN with local texture constraints, we demonstrate its effectiveness on sketch and texture-based image synthesis.  ... 
doi:10.1109/cvpr.2018.00882 dblp:conf/cvpr/XianSARLFYH18 fatcat:rdlfbz6a3rcs3m2ijm2a2dmvjm

Neural Lumigraph Rendering [article]

Petr Kellnhofer, Lars Jebe, Andrew Jones, Ryan Spicer, Kari Pulli, Gordon Wetzstein
2021 arXiv   pre-print
State-of-the-art (SOTA) neural volume rendering approaches, however, are slow to train and require minutes of inference (i.e., rendering) time for high image resolutions.  ...  texture information.  ...  Neural Volumes We use the original code shared by the authors [31] . We train our models for at least 150K iterations.  ... 
arXiv:2103.11571v1 fatcat:ltqjdj6it5brtgeilws2gipbnm

Semantic Style Transfer and Turning Two-Bit Doodles into Fine Artworks [article]

Alex J. Champandard
2016 arXiv   pre-print
Convolutional neural networks (CNNs) have proven highly effective at image synthesis and style transfer.  ...  Applications include semantic style transfer and turning doodles with few colors into masterful paintings!  ...  As long as you visit us in Vienna on July 18-20, it's better this way!  ... 
arXiv:1603.01768v1 fatcat:oxm4ktz6mjfhlpa44sren6pl4q
« Previous Showing results 1 — 15 out of 3,624 results