Development of an Image Generation System based on a Depth Map

Image processing and generation has always been an important part of broadcasting. There is color correction, creating test images, keying and other objectives in the process of video production. The task of generating new images is time consuming. Technology can provide wide opportunities for data operating due to progress in neural network development, which provide ample opportunities for processing and generating images. This research is focused on working with generative adversarial network (GAN) trained on a specific dataset based on the depth maps. The end result is an AI-generated map based on a heightmap by using lidar or ToF camera datasets.