Scribbler: Controlling Deep Image Synthesis with Sketch and Color
It addresses the utility issue for users in controlling deep image synthesis, enabling interactive editing and colorization, but is incremental as it builds on existing adversarial methods with new conditioning.
The paper tackles the problem of limited user control in deep image synthesis by proposing a deep adversarial architecture conditioned on sketched boundaries and sparse color strokes, which generates realistic images of cars, bedrooms, or faces that satisfy user constraints, with real-time feedback and comparisons showing more realistic, diverse, and controllable outputs than recent work.
Recently, there have been several promising methods to generate realistic imagery from deep convolutional networks. These methods sidestep the traditional computer graphics rendering pipeline and instead generate imagery at the pixel level by learning from large collections of photos (e.g. faces or bedrooms). However, these methods are of limited utility because it is difficult for a user to control what the network produces. In this paper, we propose a deep adversarial image synthesis architecture that is conditioned on sketched boundaries and sparse color strokes to generate realistic cars, bedrooms, or faces. We demonstrate a sketch based image synthesis system which allows users to 'scribble' over the sketch to indicate preferred color for objects. Our network can then generate convincing images that satisfy both the color and the sketch constraints of user. The network is feed-forward which allows users to see the effect of their edits in real time. We compare to recent work on sketch to image synthesis and show that our approach can generate more realistic, more diverse, and more controllable outputs. The architecture is also effective at user-guided colorization of grayscale images.