Tuesday, May 25, 2010

Next steps to think about

1. Solve problems from stroke-based NPR? Worth? or not?
2. To try to apply it to mosaic. It seems very interesting now.
3. Scratchboard effects seem working, but there are some big issues. How to obtain big regions without loosing the details? And a perception question: How to represent tone and structure in lines with only black and white.

1 comment:

  1. I think it's worth trying to address some of the problems of the ETF in this application, including sharp corners, and noise and swirls in regions of low gradient magnitude. A partial solution can involve a second field, not necessarily at right angles to the first.

    I'm less worried about big regions, because we can use graph cuts to take care of that. As for tone and structure in lines, this doesn't sound like a perceptual question as much as an artistic question: we are mainly missing line width, and I think with a precomputed high-density set of lines, we can use tone quite effectively to control width.

    ReplyDelete