Abstract: Methods, systems, and non-transitory computer readable storage media are disclosed for generating digital chain pull paintings in digital images. The disclosed system generate, utilizing a neural network, a plurality of matrices over an ambient space for a plurality of polygons of a three-dimensional mesh based on a plurality of features of the plurality of polygons associated with the three-dimensional mesh. The disclosed system determines a gradient field based on the plurality of matrices of the plurality of polygons. The disclosed system generates a mapping for the three-dimensional mesh based on the gradient field and a differential operator corresponding to the three-dimensional mesh.

Abstract: A computing system receives a query for a three-dimensional representation of a target object. The query comprises input in the form of text describing the target object, a two-dimensional image of the target object, or a three-dimensional model of the target object. The computing system encodes the input using a machine learning model to generate an encoded representation of the input. The computing system searches a search space using nearest neighbors to identify a three-dimensional representation of the target object. The search space comprises encoded representations of multiple views of a plurality of sample three-dimensional object representations. The computing system outputs the identified three-dimensional representation of the target object.

Abstract: The present disclosure is directed toward systems, methods, and non-transitory computer readable media that provide a differentiable tiling system that generates aesthetically plausible, periodic, and tile-able non-square imagery using machine learning and a text-guided, fully automatic generative approach. Namely, given a textual description of the object and a symmetry pattern of the 2D plane, the system produces a textured 2D mesh which visually resembles the textual description, adheres to the geometric rules which ensure it can be used to tile the plane, and contains only the foreground object. Indeed, the disclosed systems generate a plausible textured 2D triangular mesh that visually matches the textual input and optimizes both the texture and the shape of the mesh and satisfy an overlap condition and a tile-able condition. Using the described methods, the differentiable tiling system generates the mesh such that the edges and the vertices align between repeatable instances of the mesh.

Abstract: Implementations of systems and methods for determining viewpoints suitable for performing one or more digital operations on a three-dimensional object are disclosed. Accordingly, a set of candidate viewpoints is established. The subset of candidate viewpoints provides views of an outer surface of a three-dimensional object and those views provide overlapping surface data. A subset of activated viewpoints is determined from the set of candidate viewpoints, the subset of activated viewpoints providing less of the overlapping surface data. The subset of activated viewpoints is used to perform one or more digital operation on the three-dimensional object.

Abstract: Systems and methods for extracting 3D shapes from unstructured and unannotated datasets are described. Embodiments are configured to obtain a first image and a second image, where the first image depicts an object and the second image includes a corresponding object of a same object category as the object. Embodiments are further configured to generate, using an image encoder, image features for portions of the first image and for portions of the second image; identify a keypoint correspondence between a first keypoint in the first image and a second keypoint in the second image by clustering the image features corresponding to the portions of the first image and the portions of the second image; and generate, using an occupancy network, a 3D model of the object based on the keypoint correspondence.

Abstract: The present disclosure relates to systems, non-transitory computer-readable media, and methods that tune a 3D-object-reconstruction-machine-learning model to reconstruct 3D models of objects from real images using real images as training data. For instance, the disclosed systems can determine a depth map for a real two-dimensional (2D) image and then reconstruct a 3D model of a digital object in the real 2D image based on the depth map. By using a depth map for a real 2D image, the disclosed systems can generate reconstructed 3D models that better conform to the shape of digital objects in real images than existing systems and use such reconstructed 3D models to generate more realistic looking visual effects (e.g., shadows, relighting).

Abstract: Methods, systems, and non-transitory computer readable storage media are disclosed for generating digital chain pull paintings in digital images. The disclosed system digitally animates a chain pull painting from a digital drawing path by determining a plurality of digital bead points along the digital drawing path. In response to a movement of one of the digital bead points from a first position to a second position (e.g., based on a pull input performed at a selected digital bead point), the disclosed system determines updated positions of one or more digital bead points along the path. The disclosed system also generates one or more strokes in the digital image from previous positions of the digital bead points to the updated positions of the digital bead points.

Abstract: Certain aspects and features of this disclosure relate to modeling shapes using differentiable, signed distance functions. 3D modeling software can edit a 3D model represented using the differentiable, signed distance functions while displaying the model in a manner that is computing resource efficient and fast. Further, such 3D modeling software can automatically create such an editable 3D model from a reference representation that can be obtained in various ways and stored in a variety of formats. For example, a real-world object can be scanned using LiDAR and a reference representation can be produced from the LiDAR data. Candidate procedural models from a library of curated procedural models are optimized to obtain the best procedural model for editing. A selected procedural model provides an editable, reconstructed shape based on the reference representation of the object.

Abstract: The present disclosure relates to systems, non-transitory computer-readable media, and methods that tune a 3D-object-reconstruction-machine-learning model to reconstruct 3D models of objects from real images using real images as training data. For instance, the disclosed systems can determine a depth map for a real two-dimensional (2D) image and then reconstruct a 3D model of a digital object in the real 2D image based on the depth map. By using a depth map for a real 2D image, the disclosed systems can generate reconstructed 3D models that better conform to the shape of digital objects in real images than existing systems and use such reconstructed 3D models to generate more realistic looking visual effects (e.g., shadows, relighting).

Abstract: A training system includes: a neural network model configured to determine three-dimensional coordinates of joints, respectively, representing poses of animals in images, where the neural network model is trained using a first training dataset including: images including animals; and coordinates of joints of the animals in the images, respectively; and a training module configured to, after the training of the neural network model using the first training dataset, train the neural network model using a second training dataset including motion capture data, where the motion capture data does not include images of animals and includes measured coordinates at points, respectively, on animals.