基于深度学习的创意绘画生成方法

Creative Painting Generation Method Based on Deep Learning

创意绘画在影视动漫、艺术创作、创意娱乐以及媒体广告等行业都有广泛的应用,如何快速制作高质量的绘画作品一直以来都是行业应用中的关键问题。风格化模拟是创意绘画制作中的一类关键技术,其目标是让计算机生成的图片在视觉特性、语义内涵等方面与艺术家创作的绘画非常相似。本文结合深度学习技术进行创意绘画自动生成,利用CycleGAN方法挖掘艺术家绘画技法特征并进行绘画风格模拟,进而输出新的绘画作品。

Creative painting is widely used in industries such as film and television animation,art creation,creative entertainment and media advertising.How to quickly produce high-quality painting works has always been one of the key issues in the industry.As a key technology in creative painting production,style simulation aims to make computer-generated pictures very similar to the paintings created by artists in terms of visual characteristics and semantic connotations.This paper combines deep learning technology to automatically generate creative paintings,uses CycleGAN to get the characteristics of artists’painting and simulates painting styles to output new paintings.