Launching today
SB-GIMMEL
Pure math 17th-century mechanical ring inside a shader
4 followers
Pure math 17th-century mechanical ring inside a shader
4 followers
An interactive digital reconstruction of a rare 17th-century Gimmel wedding ring, engineered entirely within a single fragment shader using mathematical Raymarching. No 3D models (.glb/.obj), no image textures, no external canvas assets. Features fluid kinematic unfolding mechanics, anatomical D-shape comfort-fit gold profiles, and procedural Champlevé enamel monograms rendered at a rock-solid 60 FPS across all modern mobile and desktop viewports.
the fact that this is a whole wedding ring built from math with no models or textures is genuinely wild, and it runs smoothly on my phone. the little unfolding animation when you tap is such a satisfying detail.
@balkabakib78558 Thank you so much! Transforming pure mathematical formulas into smooth gold curves and fluid kinematics was the ultimate challenge here. I really wanted it to feel tactile and immediate on mobile, so hearing that the unfolding animation feels satisfying on your phone means the world to me. Appreciate the support! ⚜️
Hey everyone!
You have a unique chance to see behind the curtain of how this ring actually evolved. I still have a backup file from the early days, and back then, the hoop was just a harsh, industrial cylinder.
To turn it into a high-jewelry masterpiece, I had to completely alter the SDF math to create a genuine Comfort Fit (D-shape) profile. By replacing a flat cross-section with a soft trigonometric cosine distortion, the gold immediately started catching those narrow, liquid-like specular lights.
Here is the exact mathematical line that did the trick inside the map function:
float curveFactor = cos(clamp(p.z / thick, -1.0, 1.0) (3.14159 0.5));
dRing -= curveFactor * 0.012;
It natively softened the perimeter, completely killing the pixelated edges without a single heavy post-processing pass. Drop an upvote if you love clean math! ⚜️
A Quick Maker’s Self-Analysis: Shaders, Luxury, and the Limits of WebGL
Now that the launch is live, I wanted to share a raw, honest technical self-analysis of SB-GIMMEL. Building this wasn't just about code—it was a battle between old-world jewelry aesthetics and raw hardware math. Here is my breakdown of what worked, what I had to rethink, and where this architecture is heading next.
1. The Architectural Wins (Why it's fantastic):
The custom comfort-fit D-shape profile is an absolute victory. Getting gold to look heavy, cast, and authentic inside a single fragment shader required matching the soft blend fillets with a fine-tuned light rig. Shifting to an un-copyable procedural approach means there are no heavy .glb meshes to download, making the load times instant and the edge smoothness infinitely sharp on close zooms.
2. The Tech Pivots (What else I considered):
The war against aliasing was brutal. I initially built a heavy Supersampling (SSAA) rig, but it choked mobile GPUs by quadrupling the canvas calculations. I also tried an FXAA lens filter pass, but screen-space edge blurs ruined the ultra-sharp, premium specular highlights of the gold material. Shifting to an analytical, geometric solution by trigonometric distortion was the ultimate breakthrough—fixing the jaggies natively without performance hits.
3. Future Roadmap (What can be improved):
If I were to take this to a commercial scale, the next step is building an automated pipeline to dynamically generate a wider range of procedural gemstone cuts (like standard brilliants or emerald step-cuts) directly through code formulas.
This build proves that pure math can evoke the exact same emotional awe as a physical masterpiece from a museum. The technical foundation is rock-solid—now it's time to scale the vision into a decentralized digital luxury boutique.
Took it for a spin on my phone and the ring opening with a tap felt surprisingly satisfying, like a real hinge giving way. Hard to believe it's all math under the hood.