3D Art / 2026 / Work in progress
Before fighting the topology, I had to fight the references. Some images that looked like clean side views had a small amount of camera yaw. It was enough to expose front and rear faces that shouldn't exist in a true orthographic view, which made some proportions misleading. I rebuilt the reference setup around cleaner views and used the real overall length as the main scale anchor inside Blender.
The blockout came together quickly because I relied on booleans and bevels to find the shapes without worrying too early about beautiful topology. I paid for that speed later. Turning it into a predictable high poly meant going piece by piece: cleaning geometry where shading actually cared, adding support loops only where they earned their place, and learning that an ngon isn't automatically a problem just because it exists.
One small topology change produced a visible shading line across an otherwise flat section. Throwing more geometry at it wasn't the answer. The problem was how the surface was being resolved. I changed the split by connecting an additional vertex and the artifact disappeared. Tiny problem, surprisingly useful lesson.
The low poly isn't a blind reduction of the high poly. Flat surfaces can be brutally simple. Curved silhouettes and mechanical intersections are where the geometry earns its budget. The handle, for example, currently sits at 2,918 triangles without its screw and base, or 3,386 with them included. I'm treating those as working numbers rather than arbitrary targets.
The modeling and low poly live in Blender. Once the geometry is locked, the next steps are UVs, baking and normal-map checks in Marmoset Toolbag, followed by PBR texturing in Substance Painter and a final presentation back in Marmoset. That's the plan, not a finished-stage checklist: the image on this page still shows the model exactly where it is today.
