Skip to content

Doing style transfer with linguistic features using OpenAI's CLIP.

License

Notifications You must be signed in to change notification settings

Zasder3/CLIP-Style-Transfer

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

7 Commits
 
 
 
 
 
 

Repository files navigation

Learning Artistic Style From Language Features

Open In Colab

Doing style transfer with grounding OpenAI's CLIP.

portrait a van gogh starry night style painting of a modern house

Typical Style Transfer

Under normal circumstances, one would use a method like that of Gatys et al. to transfer the style of one image to the content of another. This only encompasses one photo and has no true concept of space.

CLIP-based Style Transfer

Using a similar technique to that of Gatys et al. we trade Gram Matrices for CLIP activations. Given your image and a phrase, CLIP can transfer (somewhat successfully) the artist or painting's style that you specify!

Special Features of this Method

  1. There is an understanding of spatial relationships. 🌌
  2. Ghost faces and self-portraits. 👻
  3. Abstract/general style and motif comprehension. 🎨

About

Doing style transfer with linguistic features using OpenAI's CLIP.

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published