Abstract: A machine learning based image processing architecture and associated applications are disclosed herein. In some embodiments, a machine learning framework is trained to learn low level image attributes such as object/scene types, geometries, placements, materials and textures, camera characteristics, lighting characteristics, contrast, noise statistics, etc. Thereafter, the machine learning framework may be employed to detect such attributes in other images and process the images at the attribute level.

Abstract: Two-dimensional compositing that preserves the curvatures of non-flat surfaces is disclosed. In some embodiments, a mapping is associated with a two-dimensional rendering that maps a potentially variable portion of the two-dimensional rendering to a canvas. The mapping is generated from a three-dimensional model of the potentially variable portion of the two-dimensional rendering. The potentially variable portion of the two-dimensional rendering is dynamically modified according to the mapping to reflect content comprising the canvas or edits received with respect to the canvas.

Abstract: A content generation platform is generally described herein. More specifically, interactive image generation and techniques and features thereof are disclosed herein. One or more sets of images of a scene are captured in an imaging studio. The captured one or more sets of images of the scene are processed using one or more machine learning based networks to generate an interactive image of the scene comprising a plurality of interactive features. One or more of the plurality of interactive features of the generated interactive image may be modified or edited according to user preferences.

Abstract: Techniques for tagging virtualized content are disclosed. In some embodiments, a modeled three-dimensional scene of objects representing abstracted source content is generated and analyzed to determine a contextual characteristic of the scene that is based on a plurality of objects comprising the scene. The modeled scene is tagged with a tag specifying the determined contextual characteristic.

Abstract: Techniques for machine learning based image attribute determination are disclosed. In some embodiments, one or more unknown attributes of a received image comprising a prescribed environment are determined using a machine learning based framework. The machine learning based framework is at least in part trained on image data sets comprising a model environment that substantially simulates the prescribed environment.

Abstract: Techniques for rendering a scene are disclosed. In some embodiments, a scene definition is modified to generate a modified scene definition based at least in part on a current state of a local database at a device. The modified scene definition is constrained to include references to objects locally available in the local database, and at least one object referenced in the scene definition is substituted with a different object in the modified scene definition that is locally available in the local database. The scene is rendered at the device according to the modified scene definition.

Abstract: Techniques for generating aligned, augmented reality views are disclosed. In some embodiments, an estimate of a floorplan of a three-dimensional space is first generated based on one or more received views of the three-dimensional space from a camera located in the three-dimensional space. A camera pose of the camera in the three-dimensional space is next determined based on generating a corrected floorplan from the estimate of the floorplan based on foreknowledge associated with the floorplan. Display of a rendering having the determined camera pose is facilitated in an actual view of the camera, resulting in an aligned, augmented reality view.

Abstract: Techniques for delivering virtualized content are disclosed. In some embodiments, elements in source content are mapped to a set of corresponding database objects, and the set of database objects is modified to generate a virtualized version of the source content. A specification of the virtualized version of the source content is provided to an output device in response to a request from the output device for the source content.

Abstract: A machine learning based image processing and generation framework is disclosed. In some embodiments, depth values of an object or asset in a received input image are at least in part determined using a machine learning based framework that is constrained to a known prescribed environment. Determined depth values facilitate generation of other views of the object or asset.

Abstract: Techniques for generating an image are disclosed. In some embodiments, a received input image is transformed to generate an output image using a machine learning based framework that is trained on a constrained set of images. The generated output image comprises an attribute learned by the machine learning based framework from the set of images.

Abstract: Techniques for generating an arbitrary view or perspective of an ensemble scene are disclosed. In some embodiments, in response to a received request for a prescribed perspective of an ensemble scene comprising a plurality of assets, an output image of the ensemble scene for the requested prescribed perspective is generated based at least in part on combining at least a portion of an existing image of each of at least a subset of the plurality of assets.

Abstract: Various techniques associated with automatic mesh generation are disclosed. One or more center curves of an outline of an object or figure are first determined. Next, for each of a plurality of points of each of the one or more center curves, a pair of rays is cast from a center curve in opposite directions, wherein the rays collide with opposite sides of the outline, and a collision pair is generated that comprises a line connecting collision points of the pair of rays on opposite sides of the outline. A mesh model of the object or figure is generated by mapping each of a set of collision pairs to polygons used to define the mesh model.

Abstract: Techniques for generating a prescribed perspective of a scene are disclosed. In some embodiments, a specification of a scene comprising a plurality of objects is specified using object views that are substantially free of perspective distortions, and a prescribed perspective of the scene is at least in part generated by populating the prescribed perspective of the scene with pixels from existing images of the plurality of objects.

Abstract: A machine learning based image processing architecture and associated applications are disclosed herein. In some embodiments, a machine learning framework is trained to learn low level image attributes such as object/scene types, geometries, placements, materials and textures, camera characteristics, lighting characteristics, contrast, noise statistics, etc. Thereafter, the machine learning framework may be employed to detect such attributes in other images and process the images at the attribute level.

Abstract: A machine learning based image processing architecture and associated applications are disclosed herein. In some embodiments, a machine learning framework is trained to learn low level image attributes such as object/scene types, geometries, placements, materials and textures, camera characteristics, lighting characteristics, contrast, noise statistics, etc. Thereafter, the machine learning framework may be employed to detect such attributes in other images and process the images at the attribute level.

Abstract: Improved techniques for generating video content are disclosed. In some embodiments, it is determined whether a frame comprising a scene is an independent frame or a predictive frame. In the event that the frame is an independent frame, the frame is encoded independently. In the event that the frame is a predictive frame, block differences and motion vectors are encoded, wherein motion vectors are determined from a known three-dimensional model and time evolution of the scene.

Abstract: Adaptive sampling of pixels is disclosed. In some embodiments, convergence of pixels of a rendering is evaluated by removing texture from the rendering to eliminate non-convergence due to texture complexity and determining whether pixels values satisfy convergence criteria associated with pixel sampling. Those pixels of the rendering that have not converged are sampled with one or more additional samples. The completed rendering is output. Pixels of the completed rendering satisfy the convergence criteria and are not uniformly sampled.

Abstract: Techniques for virtualizing content are disclosed. Elements comprising source content are virtualized by mapping each to and representing each with a corresponding database object. A specification of the corresponding database objects is provided for rendering the source content instead of any original pixel information of the source content so that a virtualized version of the source content is rendered.

Abstract: Techniques for generating an ensemble image are disclosed. In some embodiments, images associated with a plurality of independent scenes are translated to a prescribed origin, translated images associated with each independent scene are transformed into a prescribed perspective; and pixels of an image array having the prescribed perspective that is associated with each independent scene is populated with corresponding pixels from the transformed translated images associated with that independent scene. An ensemble image comprising the prescribed perspective is at least in part generated by combining at least some pixels of image arrays associated with the plurality of independent scenes.

Abstract: Two-dimensional compositing that preserves the curvatures of non-flat surfaces is disclosed. In some embodiments, a mapping is associated with a two-dimensional rendering that maps a potentially variable portion of the two-dimensional rendering to a canvas. The mapping is generated from a three-dimensional model of the potentially variable portion of the two-dimensional rendering. The potentially variable portion of the two-dimensional rendering is dynamically modified according to the mapping to reflect content comprising the canvas or edits received with respect to the canvas.