Meta Platforms Technologies, LLC

Facebook Technologies, LLC

Rohit Rao Padebettu

Steven Paul Lansel

Todd Douglas Keeler

Behnam Bastani

Haomiao Jiang

Grant Kaijuin Yang

In particular embodiments, the disclosure provides a method comprising: rendering, on a graphics processing unit (GPU), a low-resolution image associated with a scene, the low-resolution image having a resolution that is lower than a target resolution; transmitting a version of the low-resolution image to a neural accelerator; processing, on the neural accelerator, the version of the low-resolution image using a trained machine-learning model, thereby outputting a plurality of control parameters; transmitting the control parameters from the neural accelerator to the GPU; processing, on the GPU, the low-resolution image and the control parameters to construct a high-resolution image having the target resolution, wherein the GPU is programmed to determine a plurality of pixel weights for performing an interpolation using the control parameters; and outputting the high-resolution image.

Split machine-learning assisted graphics rendering pipeline

In one embodiment, a method includes receiving a source image and its associated parameters from each of multiple image sources, associating each of the source images with a layer in a range of layers based on the parameters associated with the source images, the range of layers specifying a composition layering order of the source images, generating a corresponding customized distortion mesh for each particular source image in the source images based on the parameters associated with the particular source image and at least a portion of the parameters associated with each of the source images that is associated with any layer preceding a layer associated with the particular source image, modifying each of the source images using the corresponding customized distortion mesh, generate a composite image using the modified source images, and display the composite image as a frame in a video.

Generating a composite image

In one embodiment, a method includes receiving a source image and its associated parameters from each of multiple image sources, associating each of the source images with a layer in a range of layers based on the parameters associated with the source images, the range of layers specifying a composition layering order of the source images, generating a corresponding customized distortion mesh for each particular source image in the source images based on the parameters associated with the particular source image and at least a portion of the parameters associated with each of the source images that is associated with any layer preceding a layer associated with the particular source image, modifying each of the source images using the corresponding customized distortion mesh, generating a composite image using the modified source images, and displaying the composite image as a frame in a video.

Growing community of inventors

Cupertino, CA, United States of America

Rohit Rao Padebettu