1 / 23

Learning and Producing 2D Artistic Shadings of 3D Models

Learning and Producing 2D Artistic Shadings of 3D Models. Sehoon Ha John Turgeson Karthik Raveendran. Motivation. Application: In- betweening. Pipeline. Feature Vector. Diffuse component Distance from camera Mean curvature Distance to closest edge

snow
Download Presentation

Learning and Producing 2D Artistic Shadings of 3D Models

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript

  1. Learning and Producing 2D Artistic Shadings of 3D Models SehoonHa John Turgeson Karthik Raveendran

  2. Motivation

  3. Application: In-betweening

  4. Pipeline

  5. Feature Vector • Diffuse component • Distance from camera • Mean curvature • Distance to closest edge • Is the closest edge a silhouette edge?

  6. Classification Vector • Intensity • Brush Size • Brush direction • Brush Type • Solid • Line – uses mouse velocity to determine hatching • Diagonals (forward, backword, cross) • Stripes (horizontal, vertical) • Cross Image from: http://bradalbright.blogspot.com/

  7. Learning Algorithms • Neural Network (3 layers, 13-15 hidden neurons) • kNN (5 neighbors)

  8. Synthesizing New Images

  9. Results: Bunny Reference

  10. Bunny Synthesized

  11. Applied to other shapes

  12. Applied to other shapes

  13. Capturing different styles

  14. Style comparison

  15. Style Comparison

  16. Neural networks vs. kNN

  17. Neural networks vs. kNN

  18. kNN: Number of neighbors

  19. Accuracy on training data

  20. The not-so-good results

  21. The not-so-good results

  22. Conclusion • Successful synthesis of images after learning the style of an artist • Neural networks produce more consistent results compared kNN

  23. Questions? • Thank you for coming to the last day of talks!

More Related