Nick Hardeman

@nick-hardeman.bsky.social

I like to draw, I like to paint, I like to code. Minister of Interactive Art at Design I/O. Occasionally wood working in the garage. https://nickhardeman.com https://www.design-io.com

Code/Art Fest 2025 is around the corner, and we need YOUR help 🎉 Sign up to volunteer and help make this the best Code/Art Fest yet! Whether you're assisting with workshops, greeting guests, or supporting our student showcase, your time makes a difference! ✨ Register: bit.ly/4hDfqWe

Bild

Running the output of the edge detection shader through the Zhang-Suen thinning algorithm ( b & w ), then ChatGPT giving a go at the CPU lines; Each line is a color. Unsatisfied with the result. Going to think of it as way finding for a lost creature. Compression is unkind to thin lined images :/

Lines depicting an edge detection algorithm where red is difference in normals, g is depth and b is color. Output in black and white as a line drawing with the goal of being 1 pixel wide. On the right is multiple color outlines showing many different lines composing a line drawing of a stalk of flowers.

Using a shader to detect edges from a G Buffer. The left shows a combination of normals (R), depth (G) and color(B) differences using a Sobel kernel or similar. Now I would like to get the lines from the image into a vector of lines on the CPU. Not looking for real time. Any suggestions?

Render of computer generated flower objects. On the left side is a line style output depicting the edges generated from a shader, with red showing difference in normals, g in depth and b in color. The right side shows the flowers as meshes.

I first started using CA glue for its wood working applications as a quick clamp. Using an accelerator it cures instantly. It's very handy to have around, most recent glue ups: ☑️ Pretzel to hair clip ☑️ Purse Latch ☑️ Magnet to wood ☑️ Small Figurine ☑️ Earring backing ☑️ Fingers