Diverse Single Image Generation with Controllable Global Structure
This work provides an incremental improvement in single image generation quality and diversity for researchers and practitioners working with generative models, particularly for images with strong global structures.
The paper addresses the challenge of generating diverse and realistic images from a single input image, particularly for subjects requiring global context like faces or animals. The authors achieve visually superior results compared to state-of-the-art methods, especially in global context generation, and demonstrate improved diversity as measured by the average standard deviation of pixels.
Image generation from a single image using generative adversarial networks is quite interesting due to the realism of generated images. However, recent approaches need improvement for such realistic and diverse image generation, when the global context of the image is important such as in face, animal, and architectural image generation. This is mainly due to the use of fewer convolutional layers for mainly capturing the patch statistics and, thereby, not being able to capture global statistics very well. We solve this problem by using attention blocks at selected scales and feeding a random Gaussian blurred image to the discriminator for training. Our results are visually better than the state-of-the-art particularly in generating images that require global context. The diversity of our image generation, measured using the average standard deviation of pixels, is also better.