top of page

ERRATUS

Character Modeling | 3D Environments | Rigging & Animation | NPR Shaders

Fig 1: An assortment of panels from ERRATUS (dialogue removed)

ERRATUS was created at USC in the Spring of 2024 for ART312: Comics Project; notably, not a single panel of the piece was drawn by hand, instead generated entirely within 3D graphics software -- leveraging a plethora of NPR render techniques to craft an efficient and effective workflow for comic-style rendering.

Fig 2: Full PDF of ERRATUS
(apologies for the file-size)

Breakdown

Coming first in the semester-long process of rendering ERRATUS was to sculpt, re-topologize, and rig/weight-paint its cast of main characters; this was, admittedly, the least interesting bit, though it does offer a chance to showcase some creative character designs, and formidable 3D-modeling skills. 

image_edited.jpg

Fig 3: Hollis(past), Hollis(present), and Lonny, from left to right.

lineart_demo.gif

After likewise modeling an assortment of environments and props for our characters to play with, the real work could begin: that of assembling several node-trees to make Blender's EEVEE engine look semi-convincingly like a comic.

While much of the line-art relied on simple, inverted-hull techniques, the main characters were treated to a custom node-group for procedurally generating scratchy, hatched outlines. Parameters like the line thickness, hatch density, or outline offset were exposed such that they could be tweaked to fit each panel's particular needs. 

painted_demo.gif

In alleviating the characteristic smoothness of 3D models, another node-group was produced to sample said models' normals over a Voronoi map; then, by distorting the Voronoi with a savvy choice of noise textures, the distribution of light over a given model's surface was made to appear painterly, chaotic, or otherwise hand-drawn. After layering on a number of shadows, accents, or highlights, the final shader effectively conveyed a comic-style aesthetic.

Fig. 4: Procedural NPR node groups

In combination with a few more technical work-arounds, such as splitting each scene's lighting into red, green, or blue channels, the remaining work involved posing each character, and fine-tuning each node-group's parameters, for every panel of the ensuing comic. In practice, it felt much like playing with and photographing action figures, which made for an absolute blast.

The end-result falls just short of my highest hopes -- especially where the actual story's concerned -- and so I do intend to re-use this workflow for making comics in the future.

its3d_1.gif
its3d_2.gif

Fig. 5: Yes, it's actually 3D 

shaderLayers.gif

Fig. 6: Shader layers from the bottom-up 

bottom of page