Abstract: A method of controlling power usage in an image signal processor includes measuring an actual percentage of flat pixels in a digital image, adjusting a percentage threshold of flat pixels in the digital image, based on the measured actual percentage of flat pixels in the digital image; and processing the digital image by the image signal processor based on the percentage threshold of flat pixels. The percentage threshold of flat pixels is adjusted to reduce power used by the image signal processor in processing the digital image.

Abstract: Techniques related to providing high perceptual quality video from highly compressed and decompressed reconstructed video are discussed. Such techniques include applying a pretrained decompression upsampling portion of a generative adversarial network to the decompressed reconstructed video to upsample and improve the perceptual quality of the decompressed reconstructed video to generate output video.

Abstract: Techniques related to providing high perceptual quality video from highly compressed and decompressed reconstructed video are discussed. Such techniques include applying a pretrained decompression upsampling portion of a generative adversarial network to the decompressed reconstructed video to upsample and improve the perceptual quality of the decompressed reconstructed video to generate output video.

Abstract: An apparatus and method for occlusion data compression. For example, one embodiment of a graphics processing apparatus comprises: vertex generation circuitry/logic to generate vertices of objects in a 3D space including occlusion binary vectors for the vertices, the objects captured by a set of M cameras; sorting circuitry/logic to sort the vertices in accordance with coordinates of the vertices; pre-compression circuitry/logic to transform the occlusion binary vectors of the vertices by logically combining adjacent bit fields in the sorted order to generate converted bit strings having a larger number of binary zero values than the occlusion binary vectors; compression circuitry/logic to compress the converted bit strings using run length encoding (RLE) compression to generate compressed bit fields; and a memory and/or storage device to store the compressed bit fields.

Abstract: An embodiment of a semiconductor package apparatus may include technology to process a single stream of video frames which includes frame information from two or more video sources, and process a current frame based on information from a prior frame in the single stream of video frames which precedes the current frame by a number of frames based on a number of video sources represented in the single stream of video frames. Other embodiments are disclosed and claimed.

Abstract: An apparatus and method for occlusion data compression. For example, one embodiment of a graphics processing apparatus comprises: vertex generation circuitry/logic to generate vertices of objects in a 3D space including occlusion binary vectors for the vertices, the objects captured by a set of M cameras; sorting circuitry/logic to sort the vertices in accordance with coordinates of the vertices; pre-compression circuitry/logic to transform the occlusion binary vectors of the vertices by logically combining adjacent bit fields in the sorted order to generate converted bit strings having a larger number of binary zero values than the occlusion binary vectors; compression circuitry/logic to compress the converted bit strings using run length encoding (RLE) compression to generate compressed bit fields; and a memory and/or storage device to store the compressed bit fields.

Abstract: An embodiment of a semiconductor package apparatus may include technology to process a single stream of video frames which includes frame information from two or more video sources, and process a current frame based on information from a prior frame in the single stream of video frames which precedes the current frame by a number of frames based on a number of video sources represented in the single stream of video frames. Other embodiments are disclosed and claimed.

Abstract: The invention facilitates the use of low resolution estimation to identify an optimal sequence of frame types for encoding a stream of video data. In one embodiment a series of video frames are downscaled and divided into superblocks. Low resolution motion estimation metrics are calculated for the superblocks and compared to corresponding superblocks of a previous frame. A series of video frame types is selected based on the comparison metrics and used to encode the video frames.

Abstract: The invention facilitates the normalization of activity normalization parameters to be used to select the proper quantization value for encoded video data that directly influences the Human Visual System (HVS).

Abstract: The invention facilitates the use of low resolution estimation to identify an optimal sequence of frame types for encoding a stream of video data. In one embodiment a series of video frames are downscaled and divided into superblocks. Low resolution motion estimation metrics are calculated for the superblocks and compared to corresponding superblocks of a previous frame. A series of video frame types is selected based on the comparison metrics and used to encode the video frames.

Abstract: The invention facilitates the normalization of activity normalization parameters to be used to select the proper quantization value for encoded video data that directly influences the Human Visual System (HVS).