Generative Probabilistic Image Colorization
This addresses the challenge of generating diverse and realistic colorizations for line-drawing images, which is useful for artists and designers, though it is incremental as it builds on existing diffusion models.
The paper tackles the ill-posed problem of colorizing line-drawing images by proposing a diffusion-based generative method that produces multiple candidate colorizations, avoiding mode collapse without extra constraints or re-training, and it performed well on color-conditional generation and practical tasks like inpainting.
We propose Generative Probabilistic Image Colorization, a diffusion-based generative process that trains a sequence of probabilistic models to reverse each step of noise corruption. Given a line-drawing image as input, our method suggests multiple candidate colorized images. Therefore, our method accounts for the ill-posed nature of the colorization problem. We conducted comprehensive experiments investigating the colorization of line-drawing images, report the influence of a score-based MCMC approach that corrects the marginal distribution of estimated samples, and further compare different combinations of models and the similarity of their generated images. Despite using only a relatively small training dataset, we experimentally develop a method to generate multiple diverse colorization candidates which avoids mode collapse and does not require any additional constraints, losses, or re-training with alternative training conditions. Our proposed approach performed well not only on color-conditional image generation tasks using biased initial values, but also on some practical image completion and inpainting tasks.