7 results about "Supersampling" patented technology

The invention discloses a motion vector refinement for oversampling of time domain spread. A residual network is used to predict a set of residual motion vectors that provides additional motion data for portions of frames for which motion vectors are not provided, such as animation textures, mirrored / reflective objects, and / or moving objects without motion information.

A plant three-dimensional reconstruction method for a weak texture image and scale distortion comprises the following steps: step 1, recovering an initial Gaussian point cloud and camera parameters through an SfM method by using a multi-view RGB image; 2, differential rendering is executed in training iteration, a predicted image is compared with a real image, and a loss function containing various constraints is calculated to serve as a basis for gradient solving and parameter updating; 3, calculating a pixel weighted average gradient and an artifact suppression coefficient of each Gaussian in a visible view angle; 4, updating Gaussian parameters through back propagation, and adaptively triggering Gaussian cloning, splitting or deleting operation based on the pixel gradient, the covariance matrix and transparency; and 5, dynamically adjusting the Gaussian projection matrix according to the focal length and the resolution of the camera during rendering so as to maintain scale consistency, and performing anti-aliasing rendering in combination with pixel-level super-sampling. According to the method, the reconstruction precision and the structure reduction capability of the virtual plant in the weak texture region are remarkably improved.

Joint denoising and supersampling of graphics data is described. An example of a graphics processor includes multiple processing resources, including a least a first processing resource including a pipeline to perform a supersampling operation; and the pipeline including circuitry to jointly perform denoising and supersampling of received ray tracing input data, the circuitry including first circuitry to receive input data associated with an input block for a neural network, second circuitry to perform operations associated with a feature extraction and kernel prediction network of the neural network, and third circuitry to perform operations associated with a filtering block of the neural network.

The application discloses a global illumination neural rendering method and system based on neural field space-time domain joint super sampling, which comprises the following steps: encoding input scene dynamic information to generate global illumination information of each object; obtaining current frame position related information and converting to object neural field independent coordinate space to form object position data; performing space-time joint super sampling through a global illumination transformation network combined with historical features; performing low-resolution aggregation and global illumination decoding on the super sampling samples to obtain global illumination features of each object; combining the global illumination features of each object with current frame geometry related information to generate fusion features; encoding and decoding the fusion features in a low-resolution space and performing end upsampling to output rendering results and update historical features. The application realizes global illumination neural rendering by adopting a neural field space-time domain joint super sampling method, and has the advantages of high efficiency, good details and strong time sequence stability.

Temporally amortized supersampling in graphics processing is described. An example of an apparatus includes a computer memory to store data for processing, including graphics data for a graphical application, and one or more processors including a graphical processing unit (GPU). The GPU includes a network to perform spatiotemporal upscaling filter kernel prediction for supersampling for the graphical application, the network including an input processing stage, a neural network stage, and an output filtering stage.

The application discloses a kind of extraction methods of corner point in checkerboard calibration object image.The application first obtains the rough position of checkerboard calibration object corner point;Then the rough position of corner point is converted to the image coordinate system of supersampling and the image window is intercepted at the rough position of corner point;Then the response value of window center area pixel is calculated, and the response graph is generated;Response graph is blurred to make it smooth, and the area near the corner point in the blurred response graph is fitted as elliptic paraboloid;Then the subpixel extreme position of bright spot is calculated according to the elliptic paraboloid parameter;Finally, the accurate position of corner point is obtained after coordinate conversion.Compared with the corner point boundary line-based checkerboard calibration object corner point position extraction method, the method proposed by the application can achieve higher corner point position extraction accuracy, thereby improving the accuracy of camera calibration result, and is beneficial to the accurate implementation of computer vision application.

Generating synthesized data includes capturing one or more frames of a scene at a first frame rate by one or more cameras of a wearable device, determining body position parameters for the frames, and obtaining geometry data for the scene in accordance with the one or more frames. The frames, body position parameters, and geometry data are applied to a trained network which predicts one or more additional frames. With respect to virtual data, generating a synthesized frame includes determining current body position parameters in accordance with the one or more frames, predicting a future gaze position based on the current body position parameters, and rendering, at a first resolution, a gaze region of a frame in accordance with the future gaze position. A peripheral region is predicted for the frame at a second resolution, and the combined regions form a frame that is used to drive a display.