A Multi-Implicit Neural Representation for Fonts

Fonts are ubiquitous across documents and come in a variety of styles. They are either represented in a native vector format or rasterized to produce fixed resolution images. In the first case, the non-standard representation prevents benefiting from latest network architectures for neural representations; while, in the latter case, the rasterized representation, when encoded via networks, results in loss of data fidelity, as font-specific discontinuities like edges and corners are difficult to represent using neural networks. Based on the observation that complex fonts can be represented by a superposition of a set of simpler occupancy functions, we introduce \textit{multi-implicits} to represent fonts as a permutation-invariant set of learned implict functions, without losing features (e.g., edges and corners). However, while multi-implicits locally preserve font features, obtaining supervision in the form of ground truth multi-channel signals is a problem in itself. Instead, we propose how to train such a representation with only local supervision, while the proposed neural architecture directly finds globally consistent multi-implicits for font families. We extensively evaluate the proposed representation for various tasks including reconstruction, interpolation, and synthesis to demonstrate clear advantages with existing alternatives. Additionally, the representation naturally enables glyph completion, wherein a single characteristic font is used to synthesize a whole font family in the target style.

[1]  Karthik Ramani,et al.  SurfNet: Generating 3D Shape Surfaces Using Deep Residual Networks , 2017, 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).

[2]  Thomas Funkhouser,et al.  Local Deep Implicit Functions for 3D Shape , 2019, 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR).

[3]  Douglas Eck,et al.  A Learned Representation for Scalable Vector Graphics , 2019, 2019 IEEE/CVF International Conference on Computer Vision (ICCV).

[4]  Hao Zhang,et al.  Learning Implicit Fields for Generative Shape Modeling , 2018, 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR).

[5]  Robert F. Sproull,et al.  Filtering edges for gray-scale displays , 1981, SIGGRAPH '81.

[6]  Charles T. Loop,et al.  Resolution independent curve rendering using programmable graphics hardware , 2005, ACM Trans. Graph..

[7]  Xiao Liu,et al.  STGAN: A Unified Selective Transfer Network for Arbitrary Image Attribute Editing , 2019, 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR).

[8]  Max Welling,et al.  Auto-Encoding Variational Bayes , 2013, ICLR.

[9]  Mathieu Aubry,et al.  A Papier-Mache Approach to Learning 3D Surface Generation , 2018, 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition.

[10]  Jung-Woo Ha,et al.  StarGAN: Unified Generative Adversarial Networks for Multi-domain Image-to-Image Translation , 2017, 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition.

[11]  Simon Lucey,et al.  SDF-SRN: Learning Signed Distance 3D Object Reconstruction from Static Images , 2020, NeurIPS.

[12]  Pratul P. Srinivasan,et al.  NeRF , 2020, ECCV.

[13]  S. Zagatti On viscosity solutions of Hamilton-Jacobi equations , 2008 .

[14]  Charles T. Loop,et al.  Neural Geometric Level of Detail: Real-time Rendering with Implicit 3D Shapes , 2021, 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR).

[15]  Gordon Wetzstein,et al.  Implicit Neural Representations with Periodic Activation Functions , 2020, NeurIPS.

[16]  Tzu-Mao Li,et al.  Differentiable vector graphics rasterization for editing and learning , 2020, ACM Trans. Graph..

[17]  Trevor Darrell,et al.  Multi-content GAN for Few-Shot Font Style Transfer , 2017, 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition.

[18]  Antonio Elias Fabris,et al.  Antialiasing of curves by discrete pre-filtering , 1997, SIGGRAPH.

[19]  Ivan Simecek,et al.  Improved Corners with Multi‐Channel Signed Distance Fields , 2018, Comput. Graph. Forum.

[20]  Alexandre Alahi,et al.  DeepSVG: A Hierarchical Generative Network for Vector Graphics Animation , 2020, NeurIPS.

[21]  Diego F. Nehab,et al.  Random-access rendering of general vector graphics , 2008, ACM SIGGRAPH Conference and Exhibition on Computer Graphics and Interactive Techniques in Asia.

[22]  Zhouhui Lian,et al.  Attribute2Font , 2020, ACM Trans. Graph..

[23]  Chris Green,et al.  Improved alpha-tested magnification for vector textures and special effects , 2007, SIGGRAPH Courses.

[24]  Niloy J. Mitra,et al.  Im2Vec: Synthesizing Vector Graphics without Vector Supervision , 2021, 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW).