Physically-Based Modeling Techniques for Interactive Digital Painting

By William Valentine Baxter III

This dissertation is the single most important advance in digital art since the first paint application by Shoup. In contrast with earlier academic work, like Curtis and Salesin on water colour painting, or Cockshott on oil paint simulation or any of the other research papers published and collected in volumes like Non-Photorealistic Rendering (Gooch and Gooch), Bill Baxter has not just investigated his topic and written some text, but he has created a real, usable, interactive application that has been tested by actual artists. The other researchers have never reached that stage — well, maybe we should count Raph Levien’s Wet Dream, which is derived from Curtis et al.

In this dissertation, Baxter describes the following advances in the art of painting on a computer:

  • A three-dimensional brush model, based on a simulation of the physical properties of real brushes.
  • Not one, but three successful models of paint as an artistic substance.
  • Rendering paint and mixing paint in a way that simulates the way paint actually works, instead of just fudging it with RGB.
  • A way to use a force-feedback stylus to give the user the impression that he is holding a real brush.

As I said in an earlier blog entry, no other paint application needs to apply. Corel Painter is junk compared to the possibilities of a rich application model like this. Art Rage is a toy, and a very weak one, too. Twistedbrush is a joke… I could go on, say something scathing about the very idea to use a vector application to create an artistic image for instance, but I will stop here.

Of course, once you’ve got IMPaSTo (irregular capitalization rules), you’re not done yet. Something similarly rich and physical for watercolors, pencils, chalk, pen & ink, charcoal, brush & ink needs to be done, and preferably in such a way that you can mix materials and surfaces; I tend to combine oils with fine details in India ink.

This dissertation is, apart from presenting a truly great vista of what painting with a computer could be — and should be — remarkable in that it is also very clearly organized and very well-written. I still couldn’t grasp the math (what does an upside-down capital delta mean?), but that’s to be expected of someone who never had any decent maths, even at secondary school. But the principles were clear, the way Baxter uses — and I suspect sometimes abuses — OpenGL and shader routines to simulate and render paint was an eye-opener.

So it’s a pity the code isn’t going to be open-sourced — I can understand it, it has the potential of becoming a commercial hit — but I’d have loved to peek at the actual source…