ABSTRACT:
Nowadays, Manga (comics in Japanese) and cartoon productions have
grown to become well-established industries. However, traditional
productions cannot keep up with the growth due to labor-intensive and
complex drawing procedures. Some manga artists start to incorporate
computer programs into their workflow, even though most existing
computing techniques still requires large amount of user input and
adjustments. So, how can we intelligently harness computational power
and really free artists from those boring chores?
In this talk, I will discuss some of my recent work in computational
manga. Specifically, I will describe 1) Manga colorization, a
technique that aim to unify the B/W and color manga productions; 2)
Structure-aware halftoning, an advanced technique for producing
detailed manga background; 3) Manga screening, a framework to automate
the manga screening process. Some future challenges and opportunities
in computational manga will also be included and discussed.
grey level continuity <=> pattern continuity
In existing literature, people use grey level continuity to process traditional images, however, in comics, there is no grey level continuity, to segment it, we cannot employ that method. Instead, we can observe the pattern used in comics. Pattern continuity is used.
Leak proofing colorization vs. Stroke preserving colorization
To colour the unbounded regions in comics.
Halftoning
To create comic scenes from images, represent colour with Black-White patterns. Mapping colour between pattern. Construct the relevance relationships in each area, colour and pattern. Mapping the relationship structure of each area, finally, construct the mapping between colour and pattern. Mapping colour and pattern changes to mapping difference of different colours and difference of different patterns.
Character Creation
In her talk, she envisioned that in the future they maybe do some research on given an initial character in comics and then automatically create the different poses, emotions, frames from different views, etc.. Her view is something like our work now.
Speaker's site:http://appsrv.cse.cuhk.edu.hk/~ygqu/