A neural-based method for stippled image rendering has been created. This method aims to replace the manual process of drawing a stippling, which is extremely laborious and time consuming. This method automatically creates these types of drawings quickly, efficiently and with high accuracy.
Stippling is an art form that is widely used to represent texture information of objects when documents are needed to be printed in black and white. The current process for creating a hand-made stippling takes hours to complete an individual drawing. This newly developed method uses neural-based techniques which allow content and style information to be extracted out of the input and target style images. Other semantic information of the input image is additionally extracted to refine the process, which creates directionally aware stippling dots. A much quicker and more accurate stippled image results.
· Very fast process
· Improves on prior art by taking advantage of deep learning
· Automatic synthesis of stippled images
· Results are more accurate and better represent the stippling style
Information content such as style and semantic information is extracted from input images using deep learning techniques. Using this information a directional map is computed and utilized during the stippling rendering process. The final image is produced very quickly and with high fidelity and accuracy.
The technology is patent pending. Further information on licensing opportunities is available on request.