3D Toon Render to 2D "Painting" - Syntharan Cityscapes (2023)

Results from an experiment/demo livestream on a workflow designed more for "sketch to render" using Vizcom AI's custom diffusion-assisted design tool. These 3D toon renders worked surprisingly well as inputs, considering the density variation. On closer inspection, there is some minimal foreground/background confusion, but not too bad for mood renders! I also did some work painting directly within Vizcom to do some "inpainting" generation, and later Photoshop to pull out some colors and fix some artifacts, but the post-processing was fairly minimal.

Diffusion render (result).

Diffusion render (result).

Input toon render, this one with manual outlining to try to differentiate foreground and background.  This proved not to be necessary.

Input toon render, this one with manual outlining to try to differentiate foreground and background. This proved not to be necessary.

Result

Result

Input

Input

Result

Result

Input

Input