Unity Technologies Sf

Weta Digital Limited

Erik B Edlund

Dejan Momcilovic

Tobias B Schmidt

Josh Hardgrave

Embodiments allow live action images from an image capture device to be composited with computer generated images in real-time or near real-time. The two types of images (live action and computer generated) are composited accurately by using a depth map. In an embodiment, the depth map includes a 'depth value' for each pixel in the live action image. In an embodiment, steps of one or more of feature extraction, matching, filtering or refinement can be implemented, at least in part, with an artificial intelligence (AI) computing approach using a deep neural network with training. A combination of computer-generated ('synthetic') and live-action (&#x201c;recorded&#x201d;) training data is created and used to train the network so that it can improve the accuracy or usefulness of a depth map so that compositing can be improved.

System for image compositing including training with synthetic data

Compositing is provided in which visual elements from different sources, including live action objects and computer graphic (CG) merged in a constant feed. Representative output images are produced during a live action shoot. The compositing system uses supplementary data, such as depth data of the live action objects for integration with CG items and light marker detection data for device calibration and performance capture. Varying capture times (e.g., exposure times) and processing times are tracked to align with corresponding incoming images and data.

Correlation of multiple-source image data

Embodiments allow camera effects, such as imaging noise, to be included in a generation of a synthetic data set for use in training an artificial intelligence model to produce an image depth map. The image depth map can then be employed to assist in compositing live action images from an image capture device with computer generated images in real-time or near real-time. The two types of images (live action and computer generated) are composited accurately by using a depth map. In an embodiment, the depth map includes a 'depth value' for each pixel in the live action image. In an embodiment, steps of one or more of feature extraction, matching, filtering or refinement can be implemented, at least in part, with an artificial intelligence (AI) computing approach using a deep neural network with training.

Growing community of inventors

Wellington, New Zealand

Erik B Edlund