Tutorial / 14 January 2021

Using A.I Image Enhancing with Neural-Style


There is a secret to high quality outputs and its A.I Base enlarging with either Let's Enhance or Topaz A.I. (Topaz A.I is really good, just need a decent GPU for fast processing)


Take a peek under the hood to see what makes these results so stunning, and its linked to the style image containing the right resolutions and that is 7200px, my max default is 7500px. You also lose a bit of resolution when combining multiple styles and you can push higher resolutions when using less styles or only one.


The Folder with the 5 styles is called "X28E" I try to use a systematic approach, but eventually its just extending to new names like X28D,X28E,X28F...etc when trying new styles or similar styles.


For best results its good to match the same resolution as the final output, anything higher is good, like 10000px or 19000px.
  


I use google for processing, it helps keep it on the cloud and on my personal drive. 


This is how it forms.


A1 Generation - 512px



  

A3 Generation - 1024px
  

A5 Generation - 1400px
 

A8-Final Generation - 7200px
  

Feel free to experiment with different fractals, images or photographs, there is no limit to what you can use, some of the most amazing results come from trying new things you would least expect to provide good results, and like always the community is always open to sharing ideas about how good results are produced.