Maybe I’m missing something here, or using the wrong model…but this is the result of an attempt to convert an STL part to NURBS using GPT via Rhino MCP. Original mesh in grey on the left.
The way people on youtube are speaking you’d think we’d crossed some great hurdle…seems not.
STL for anyone interested:
mtron.stl (517.4 KB)
Hi Ftzuk,
There are hurdles to overcome. I’m working on a more deterministic AI workflow but am not quite far along enough to share this week, but hopefully next.
How does this file look?
The workflow creates editable parameters that can be driven by the ai, Rhino or GH.
Stl-Rhino.3dm (1.3 MB)
I would like to get this level in less than 5k in tokens total.
Measuring the mesh (profile, slots, groove) 10k
Writing the program and the first run 10k
Checking against the STL and three fix rounds (groove gap mirrored, half slot ,groove length) ~35k
