Haunts

12 nightmare/dreamscapes, created in 3D and processed post-render to emulate an impressionist aesthetic. Inspired by the paintings of Zdzisław Beksiński.

Asset Development

Look variant


 


Torso Build

The torso was built procedurally by taking an input posed rig and extracting a spine and sternum curve to draw rib cage geometry from.

By running through a for-loop to clone inwards, and manuplating the position of points based on the mirrored curveu attribute, the whole torso shape is controlled parametrically.

The arms were drawn by running a shortest-path curve generator over points scattered on the arms in the input posed geo.

Walking Man by Alberto Giacometti - reference

Inspiration / reference

Asset build steps

 

Head

The head was built by deforming a skull SDF with mirrored noises.

Basic skull model

Vellum shrink wrapping for watertight mesh

Various noises applied to the SDF for head shape, wrinkle, eye and teeth deformation

 

Asset Posing

By creating the asset based on a posed biped rig, each pose would have unique characteristics.

 

Environment

Various environments were constructed in Houdini.

Reference location, Kew Gardens

Viewport

Render

Composite /

Paint processing


Render

Composite

Paint processing

 

Beksinski Structures

One of the hallmarks of Zdzisław Beksiński’s work is the bone scaffolding interior supporting his structures.

Reference

Render/composite

Paint processing

 

Paint Processing

The second half of the project consists of impressionist paint emulation. This was achieved in Nuke using a variety of vector distortion, blurring and layer stacking.

Advection of noise field, driven by underlying render and fBm vectors

Accumulation of layers of “strokes”

Paint with hight dynamic range

Vector and noise advection

 

Canvas Texturing

I used layered sine waves to draw the canvas threads.

You can use sine waves for other styles of weave as well

The sine wave setup can be scaled infinitely, to any resolution, and distorted with noise.

Scaled down setup

Previous
Previous

Digital Neurons

Next
Next

Euclidean Stars