ebsynth is a versatile tool for by-example synthesis of images. It can be used for a variety of image synthesis tasks, including guided texture synthesis, artistic style transfer, content-aware inpainting and super-resolution.
The focus of ebsynth is on preserving the fidelity of the source material. Unlike other recent approaches, ebsynthdoesn’t rely on neural networks. Instead, it uses a state-of-the-art implementation of non-parametric texture synthesis algorithms. Thanks to its patch-based nature, ebsynth produces crisp results, which preserve all the fine detail present in the original image.