
This is a small gift for a friend
DragonSaviik, an experimental work of using style transfer based on deep neural networks (DNNs) to create art. I am a little surprised to find that Wolfram Mathematica can do it too, and its recent versions' updates likely oriented to machine learning. The frame fractals used on this one were created about two years ago when I had some passion in fractals.
I made two versions, and the dark version is my preferred one. Wish you like it.
You can find the other version here: https://www.furaffinity.net/view/39903886/
送给一个朋友的小礼物,用基于DNNs的风格迁移做些画的实验性作品。最近才发现Mathematica也可以搞这个,也不算太奇怪它最近几个版本应该都是集中升级机器学习能力的。图上用的框架分形是两年前比较迷分形的时候做的。
有两个版本,我比较喜欢暗版。希望您喜欢。
另一个版本在这:https://www.furaffinity.net/view/39903886/

I made two versions, and the dark version is my preferred one. Wish you like it.
You can find the other version here: https://www.furaffinity.net/view/39903886/
送给一个朋友的小礼物,用基于DNNs的风格迁移做些画的实验性作品。最近才发现Mathematica也可以搞这个,也不算太奇怪它最近几个版本应该都是集中升级机器学习能力的。图上用的框架分形是两年前比较迷分形的时候做的。
有两个版本,我比较喜欢暗版。希望您喜欢。
另一个版本在这:https://www.furaffinity.net/view/39903886/
Category Artwork (Digital) / Abstract
Species Western Dragon
Size 1247 x 1246px
File Size 2.42 MB
Listed in Folders
Yah, the style transfer pic is a different fractal. I have tried using the same pic on the right as the transfer base, but the output is not ideal. The style weight and the algorithm itself might contribute to it. I might try to adjust the code to do more like this in the future.
Comments