Adobe, Inc.

Alexandre S Parenteau

Lubomir D Bourdev

Cynthia W Lau

Stephan R Yhann

Dejan Markovic

Daniel J Clark

Hailin Jin

Shmuel Avidan

Manish S Kulkarni

Abhinav Darbari

Timothy H Roth

John Charles Nash

Kevin T Dale

A method, system, and computer-readable storage medium are disclosed for rendering knockout groups using a graphics processing unit (GPU). Input comprising at least one knockout group may be received at the GPU. The knockout group may comprise a plurality of objects that are ordered by paint order. Each object may be represented in the input by a plurality of vertices. The plurality of objects may be drawn into a frame buffer of the GPU such that a topmost object at each pixel in the frame buffer determines a color and an opacity of the pixel with respect to the other objects in the knockout group. In drawing the plurality of objects of the knockout group into the frame buffer of the GPU, program code may be executed by the GPU.

Fast rendering of knockout groups using a depth buffer of a graphics processing unit

Various embodiments of a system and methods for labeling images are described. An image labeling system may receive multiple images. A subset of the images may be labeled to identify image content or elements, such as faces. The system may display some of the labeled image elements in different portions of a display area. Unlabeled image elements may be displayed in the same display area. The display size and position of each unlabeled image element may be dependent on similarities between the unlabeled image element and the displayed, labeled image elements. The system may receive input which indicates that one or more of the displayed, unlabeled image elements should receive the same label as one of the displayed, labeled image elements. Similarities between image elements may be re-analyzed, dependent on the assigned labels, and the display of unlabeled images may be updated.

System and method for labeling a collection of images

A method and apparatus for tracking objects within a video frame sequence. In one embodiment, a method for tracking an object within a video frame sequence is disclosed. The method includes processing each jframe of the video frame sequence to determine a motion vector defining motion between a prior jframe and a current jframe. The method includes creating an object descriptor for an object being tracked followed by generating a document object model comprising motion information and the object descriptor.

Method and apparatus for tracking objects within a video frame sequence

Face detection is performed on a plurality of images to identify a plurality of faces. A subset of the plurality of faces is activated, including by loading into memory the subset of faces and/or data corresponding to the subset of faces. At least one of the plurality of faces is left un-activated. A distance for each pair of activated faces is calculated, wherein the distance is associated with a measure of a similarity between a first face and a second face. The activated faces are clustered into a plurality of groups based at least in part on the distances. A representative face is selected from each group resulting from the clustering and the representative faces are displayed.

Efficient and scalable face recognition in photo albums

A method, system, and computer-readable storage medium are disclosed for rendering an artwork comprising a first plurality of objects. At least one common attribute shared by a second plurality of objects in the artwork may be automatically identified. The second plurality of objects may comprise a subset of the first plurality of objects. Each object of the second plurality of objects may be automatically assigned to a group. It may be determined whether the group is located within an area to be rendered. The group may be culled if the group is not located within the area to be rendered.

Batch early engagement rules graphics processing

Transparency groups or other images may be rendered on graphics hardware using a GPU utilizing only a single frame buffer and without the need to switch contexts to another frame buffer. A single frame buffer may be allocated and the overall background image may be rendered to the frame buffer. In order to render a foreground image to be combined with the background image, a sub-image of the background image is copied from the frame buffer to a texture atlas. The foreground image may then be rendered to the portion of the frame buffer from which sub-image was copied. The foreground image may then be copied from the frame buffer into the texture atlas. Additionally, both the sub-image of the background image and the foreground image may be merged and copied from the texture atlas into the frame buffer.

Frame buffer based transparency group computation on a GPU without context switching

A method, system, and computer-readable storage medium are disclosed for rendering artwork using a graphics processing unit (GPU). The GPU may comprise a depth buffer and a stencil buffer. Artwork input comprising one or more dirty regions and one or more clip paths may be received at the GPU. A culling operation may be performed on the artwork input. The culling operation may comprise configuring the depth buffer to reject pixels located outside a union of the one or more dirty regions. A clipping operation may be performed on the artwork input. The clipping operation may comprise configuring the stencil buffer to reject pixels located outside an intersection of the one or more clip paths.

Clipping and culling artwork using a graphics processing unit

Shader programs that execute on graphics processing units (GPUs), such as vertex and pixel shaders may be generated by defining individual shader snippets. Each snippet may represent and/or specifies a particular shader operation. In one embodiment, each snippet may indicate a particular vertex shader operation, a particular pixel (or fragment) shader operation, or both. Various combinations of these snippets may then be combined to create more complex shader programs. A shader snippet framework may be configured to receive information specifying individual snippets, as well as the combinations of snippets representing various shader programs. The framework may define such shader programs using only the identifiers for the various snippets and thus a shader program, as described herein, may not include any actual code, but instead may refer to an ordered list of snippets. At runtime, the snippets are then instantiated and executed to execute the shader program.

System and method for implementing graphics processing unit shader programs using snippets

A new transparency group may be rendered and blended with other, previously rendered, transparency groups, by using the Porter-Duff algebra available on the GPU even though the transparency groups include pre-multiplied color and alpha information. Additionally, the number of copies of the back buffer (the image information for the previously rendered transparency groups) required to properly render, blend and combine the new transparency group into the image information of previously rendered transparency groups may be minimized.

Transparency group computation on a graphics processing unit

Computer-implemented methods and apparatus for processing a graphical element that has an associated original type, including blending at least part of the graphical element and at least part of one or more other graphical elements to produce a transformed graphical element. The transformed graphical element has an associated transformed type, and the transformed type is different than the original type. Information about the original type is stored, and the transformed graphical element, an adjacent graphical element, or both are processed using the stored information about the original type.

Processing illustrations using stored information

Computer-implemented methods and apparatus for converting a computer graphics illustration that includes a plurality of overlapping artwork elements from an input format to an output format that differs from the input format. The overlapping artwork elements are classified as transparent or non-transparent artwork elements. The transparent artwork elements are classified as being compatible or incompatible with the output format. The incompatible transparent artwork elements and any of the artwork elements that are at least partially overlapped by any of the incompatible transparent artwork elements are transformed into transformed artwork elements, without transforming any compatible artwork element that is not overlapped by an incompatible transparent artwork element. The transformed artwork elements are compatible with the output format.

Selectively transforming overlapping illustration artwork

A method, system, and computer-readable storage medium are disclosed for performing anti-aliasing operations on lines using a graphics processing unit (GPU). Input comprising a plurality of vertices may be received at a GPU. The plurality of vertices may represent one or more line segments. One or more anti-aliased line segments may be generated based on the input. In generating the one or more anti-aliased line segments, the intensity of a first edge and a second edge of each line segment may be modified based on application of an anti-aliasing function. A maximum intensity of a central portion may be maintained for at least one of the line segments, wherein the central portion is between the two edges and is greater than one pixel in thickness.

Growing community of inventors

San Jose, CA, United States of America

Alexandre S Parenteau