Half a Million Views Later… BumpMesh Is Way Better — CNC Kitchen

Why this is in the vault

BumpMesh is a browser-based, local-first, AGPL-licensed tool for applying physical textures to 3D print geometry — zero server uploads, fully open source. This update video is a case study in community-powered product iteration: 500K views on launch triggered hundreds of PRs, issues, and feature contributions that shipped in ~3 months. The community flywheel pattern and the "local processing in the browser" architecture are both directly relevant to how RDCO thinks about tool design, open-source governance, and product velocity. Stefan's mesh-remeshing work (equilateral triangle quality goal) is also a clean example of computational geometry craft applied to a consumer problem.

Episode summary

Stefan reviews the major improvements shipped to BumpMesh since the original viral video (~500K views in ~3 months). The new version is faster (5× export speed), handles larger models, ships with community-contributed quality features, and has grown into a genuine open-source project with internationalization, mesh repair, remeshing, texture baking, and 3MF export.

Key arguments / segments

[00:00:01] Intro — BumpMesh recap: browser-based, no server upload, AGPL open source. ~500K views on launch video; community drove hundreds of PRs and issues.
[00:00:33] Community contributions framing — respecting open-source maintainer burden; Stefan accepted "many reasonable suggestions."
[00:01:06] Privacy + open-source model — everything processed locally in-browser, nothing leaves your machine; can also self-host. Addresses questions from large companies.
[00:02:05] Open Sauce 2026 sponsorship segment — San Mateo County Event Center, 100+ science creators.
[00:02:37] New UI overview — changelog modal on launch, save/load project or settings only, expanded language support (Spanish → Japanese/Korean), undo/redo, reset-all and per-slider reset via double-click, fine-tune via scroll-wheel on value field, perspective/orthographic toggle.
[00:03:08] Mesh repair on import — detects broken meshes (common with Tinkercad exports); warns user, points to slicer or online repair tools. Orientation improvements. Custom STEP importer built from scratch for post-processible mesh quality.
[00:04:47] Community feature: smooth transition — Andrew Sings contributed blended boundary between textured area and original surface. Cylindrical mapping: snap-to-seam-free scaling, proper projection cylinder placement.
[00:05:30] Remeshing (equilateral triangle goal) — Stefan spent several days on this; makes all triangles roughly equal-size and equilateral before texture displacement. On by default. Eliminates sliver/needle triangle artifacts. Still labeled beta.
[00:06:21] Beta features / overhang fix — prevent downward texture displacement (XY-only displacement avoids tiny islands under 45° overhangs). Texture baking: apply current texture and continue stacking without re-import.
[00:07:04] Export improvements — resolution slider direction reversed (more = finer, less = coarser; was backwards before). Recommendation system for export resolution + triangle count. 3MF export added (compressed archive → much smaller files vs STL).
[00:07:44] Performance — 5× faster export vs April release; memory optimizations enable significantly larger models and finer details. Flagged as one of the best overall improvements.
[00:08:21] Closing / roadmap teaser — multi-texture in one go, multicolor textures (Eric's contribution), more coming. Heartfelt thanks to the open-source community for tips, code, ideas, reports.

Mapping against Ray Data Co

Strength: medium

Three angles worth holding:

Community flywheel as a product strategy — BumpMesh went from 0 to 100s of community contributions in 90 days via AGPL + viral distribution. The pattern (launch opinionated open-source tool, let community surface edge cases and add feature breadth) is a viable acquisition and product R&D mechanism worth modeling for any RDCO tool that reaches a general technical audience.
Local-first / browser-native architecture — Stefan's explicit answer to the "where does my data go?" corporate concern is to process everything in-browser, zero server. This is directly relevant to any RDCO data tool where client data sensitivity is a blocker — the trust primitive is architecture, not policy.
Computational geometry craft — The remeshing work (equilateral triangle goal before texture displacement) is a good framing for "quality infrastructure you don't see but that eliminates downstream artifacts." Useful conceptual vocabulary when explaining invisible quality work to non-technical stakeholders.

Not directly actionable for phData or current client work. File as background on Stefan's open-source governance model and digital manufacturing tool design patterns.

Sponsorship

Open Sauce 2026 (science/maker fair at San Mateo County Event Center, ~2:00–2:37). Appears as a mid-roll community promo segment.

[[06-reference/2026-06-20-cnc-kitchen-laser-3d-printing-solution]] — prior CNC Kitchen vault entry, xTool laser review
[[01-projects/digital-manufacturing-discovery/2026-04-19-channel-shortlist]] — channel shortlist that positioned CNC Kitchen as a primary depth source for operational 3D printing knowledge