StoryGAN: A Sequential Conditional GAN for Story Visualization
This addresses the challenge of visualizing stories with consistent dynamic elements, which is incremental as it builds on GAN frameworks for a new task.
The authors tackled the problem of generating a sequence of images from a multi-sentence paragraph, focusing on global consistency across scenes and characters, and proposed StoryGAN, which outperformed state-of-the-art models in image quality and consistency metrics.
We propose a new task, called Story Visualization. Given a multi-sentence paragraph, the story is visualized by generating a sequence of images, one for each sentence. In contrast to video generation, story visualization focuses less on the continuity in generated images (frames), but more on the global consistency across dynamic scenes and characters -- a challenge that has not been addressed by any single-image or video generation methods. We therefore propose a new story-to-image-sequence generation model, StoryGAN, based on the sequential conditional GAN framework. Our model is unique in that it consists of a deep Context Encoder that dynamically tracks the story flow, and two discriminators at the story and image levels, to enhance the image quality and the consistency of the generated sequences. To evaluate the model, we modified existing datasets to create the CLEVR-SV and Pororo-SV datasets. Empirically, StoryGAN outperforms state-of-the-art models in image quality, contextual consistency metrics, and human evaluation.