Abstract: A method of determining an authenticity of a normal map is disclosed. An input candidate normal map is received. A reconstructed candidate normal map is generated based on a performance of a mathematical differentiation on an integration of the input candidate normal map. A reconstruction error is determined based on a comparison of the input candidate normal map to the reconstructed candidate normal map. An authenticity of the input candidate normal map is determined based on the reconstruction error being within a configurable threshold.

Abstract: Herein is presented a light-weight, high-quality texture synthesis algorithm that generalizes to other applications. We utilize an optimal transport optimization process within a bottleneck layer of an auto-encoder, achieving quality and flexibility on par with expensive back-propagation based neural texture synthesis methods, but at interactive rates. In addition to superior synthesis quality, our statistically motivated approach generalizes better to other special case texture synthesis problems such as Style Transfer, Inverse-Texture Synthesis, Texture Mixing, Multi-Scale Texture Synthesis, Structured Image Hybrids and Texture Painting. We treat the texture synthesis problem as the optimal transport between Probably Density Function of the deep neural activation vectors of the image being synthesized and the exemplar texture.

Abstract: Systems and methods perform non-parametric texture synthesis of arbitrary shape and/or material data taken from an exemplar object in accordance with embodiments of the invention. Exemplar data is first analyzed. Based upon the analysis, new unique but similar data is synthesized in a myriad of ways.

Abstract: Herein is presented a light-weight, high-quality texture synthesis algorithm that generalizes to other applications. We utilize an optimal transport optimization process within a bottleneck layer of an auto-encoder, achieving quality and flexibility on par with expensive back-propagation based neural texture synthesis methods, but at interactive rates. In addition to superior synthesis quality, our statistically motivated approach generalizes better to other special case texture synthesis problems such as Style Transfer, Inverse-Texture Synthesis, Texture Mixing, Multi-Scale Texture Synthesis, Structured Image Hybrids and Texture Painting. We treat the texture synthesis problem as the optimal transport between Probably Density Function of the deep neural activation vectors of the image being synthesized and the exemplar texture.

Abstract: Systems and methods for performing non-parametric texture synthesis of arbitrary shape and/or material data taken from an exemplar object in accordance with embodiments of the invention are illustrated. Exemplar data is first analyzed. Based upon the analysis, new unique but similar data is synthesized in a myriad of ways.

Abstract: Systems and methods perform non-parametric texture synthesis of arbitrary shape and/or material data mimicking input exemplar data in accordance with embodiments of the invention. Exemplar data is first analyzed and appearance vectors are generated based on geometric information determined for the exemplar data. Feature vector maps are generated for locations of the exemplar data based on the geometric information and the appearance vectors. Based upon the feature vector maps, outputs can be synthesized.

Abstract: Systems and methods for training a generative ensemble network to generate image data in accordance with embodiments of the invention are illustrated. One embodiment includes a method for generating an image through an ensemble network architecture. The method includes steps for passing a set of one or more images through a single standard convolution layer that acts as a root node to produce a first output and passing the first output through a plurality of branches to produce a plurality of outputs for the plurality of branches. Each branch is a separate and independent network that receives input from the single standard convolution layer. The method further includes steps for passing the plurality of outputs through a supervisor layer that combines the plurality of outputs into a final solution.

Abstract: Systems and methods for providing convolutional neural network based image synthesis using localized loss functions is disclosed. A first image including desired content and a second image including a desired style are received. The images are analyzed to determine a local loss function. The first and second images are merged using the local loss function to generate an image that includes the desired content presented in the desired style. Similar processes can also be utilized to generate image hybrids and to perform on-model texture synthesis. In a number of embodiments, Condensed Feature Extraction Networks are also generated using a convolutional neural network previously trained to perform image classification, where the Condensed Feature Extraction Networks approximates intermediate neural activations of the convolutional neural network utilized during training.

Abstract: Systems and methods for performing non-parametric texture synthesis of arbitrary shape and/or material data taken from an exemplar object in accordance with embodiments of the invention are illustrated. Exemplar data is first analyzed. Based upon the analysis, new unique but similar data is synthesized in a myriad of ways.

Abstract: Systems and methods for providing convolutional neural network based image synthesis using localized loss functions is disclosed. A first image including desired content and a second image including a desired style are received. The images are analyzed to determine a local loss function. The first and second images are merged using the local loss function to generate an image that includes the desired content presented in the desired style. Similar processes can also be utilized to generate image hybrids and to perform on-model texture synthesis. In a number of embodiments, Condensed Feature Extraction Networks are also generated using a convolutional neural network previously trained to perform image classification, where the Condensed Feature Extraction Networks approximates intermediate neural activations of the convolutional neural network utilized during training.

Abstract: Systems and methods for providing convolutional neural network based image synthesis using localized loss functions is disclosed. A first image including desired content and a second image including a desired style are received. The images are analyzed to determine a local loss function. The first and second images are merged using the local loss function to generate an image that includes the desired content presented in the desired style. Similar processes can also be utilized to generate image hybrids and to perform on-model texture synthesis. In a number of embodiments, Condensed Feature Extraction Networks are also generated using a convolutional neural network previously trained to perform image classification, where the Condensed Feature Extraction Networks approximates intermediate neural activations of the convolutional neural network utilized during training.

Abstract: Systems and methods for providing convolutional neural network based image synthesis using localized loss functions is disclosed. A first image including desired content and a second image including a desired style are received. The images are analyzed to determine a local loss function. The first and second images are merged using the local loss function to generate an image that includes the desired content presented in the desired style. Similar processes can also be utilized to generate image hybrids and to perform on-model texture synthesis. In a number of embodiments, Condensed Feature Extraction Networks are also generated using a convolutional neural network previously trained to perform image classification, where the Condensed Feature Extraction Networks approximates intermediate neural activations of the convolutional neural network utilized during training.