Work / Hardware & Making

CAD & 3D printing

Scaled models to the build limits of three printers and ported projects across printers and slicers, TPU included.

Status
explored
Role
Solo: print engineering
Timeline
May 2025 →
  • Bambu Studio
  • Anycubic Slicer Next
  • OrcaSlicer
  • PLA and TPU

TL;DR

  • I scaled two third-party models for three printers each, and all six saved variants sit within 4.3 mm of their printer’s limit.
  • I’ve saved print-ready projects for 4 printer models, worked in 3 slicers, and ported projects between the Bambu and Anycubic ecosystems, including one on flexible TPU.

The problem

A model that fits one printer rarely fits another. Scale it up and it overflows the build volume; scale it down and thin features drop below what a nozzle can lay. Every printer, slicer and filament combination is its own set of limits.

What I built

  • Scaling to the envelope. A 1.96-million-triangle statue scan comes out 179.2 mm tall for a Bambu Lab A1 mini (180 mm build height), 255.7 mm for an A1 (256 mm) and 324.2 mm for an H2D (325 mm): each the largest 0.01 scale step that fits, with tree supports on all three. A glider’s diagonal wing plate spans 177.4 mm on the A1 mini’s 180 mm bed, 254.5 mm on the A1’s 256 mm bed, and 295.7 mm inside the 300 mm strip that both of the H2D’s nozzles can reach.
  • Multi-printer projects. Print-ready projects for the Bambu Lab A1 mini, A1 and H2D and the Anycubic Kobra 3 V2, working across Bambu Studio, Anycubic Slicer Next and OrcaSlicer.
  • Porting. Two projects made for Bambu printers moved onto the Kobra 3 V2 in Anycubic Slicer Next, one of them a handbag on an Anycubic TPU profile.
  • Variants as projects. Each scale, printer and infill variant is its own saved project: 11 print-ready projects in all.
  • Infill comparison. On a bonsai model I compared infill patterns head to head, saving gyroid, crosshatch and 3D-honeycomb variants across 46 slicing runs in 78 minutes.

Key decisions

  • Decision: scale to the binding dimension of each machine. Why: the statue is limited by height, the glider by its diagonal wing plate, and each needs a different constraint. Trade-off: uniform scaling thins every feature by the same factor.
  • Decision: one saved project per variant. Why: a scale or infill experiment can be compared and reprinted without re-deriving its settings. Trade-off: more files to keep straight.

The hard part

Fitting a model to a machine without breaking it. Uniform scaling fills the envelope but thins every feature by the same factor: at 0.69× the glider’s 0.4 mm tail drops to about 0.28 mm, one or two 0.20 mm layers.

The statue builds sit at the other edge. Each takes the largest 0.01 step that fits, and one more step overflows every machine (1.37× gives 180.6 mm against a 180 mm build height). The H2D adds a third constraint: its two nozzles have offset reach, so only a 300 mm strip of its 350 mm plate is reachable by both, and the glider’s 295.7 mm wing sits inside that strip.

Results

  • Six variants across two models and three printers, each within 4.3 mm of its limit; the tightest clears by 0.3 mm.
  • 11 saved print-ready projects across 4 printer models and 2 filament materials, with work in 3 slicers.

What I’d do next

  • Original designs of my own, and photos of finished prints next to their slicer settings.
  • More experiments and tinkering: Lab

Verified numbers

MetricValueSource
Scaled variants, each within 4.3 mm of its printer’s build limit (2 models × 3 printers)6local: Metadata/project_settings.config
Tightest fit, a 255.7 mm build under a 256 mm limit0.3 mm clearancelocal: Downloads, statue-scan slicer project (A1, full size)
Printer models with saved print-ready projects4local: Metadata/project_settings.config
Slicing runs in one infill-comparison session (19 pattern changes)46 in 78 minlocal: BambuStudio/log/debug_Tue_Mar_17_*.log.0
m.mittal
Home
Work
Research
Lab
About
Experience
Now
Résumé
Uses
Colophon
Contact
AccountWardin-build
Redlbuilt
The Councilresearch
BEMAresearch
EXIT LIQUIDITYin-build
NEXUSarchived
Warrant + Forgebuilt
Concordarchived
BlueCollarPalin-build
Chispenbuilt
Warrant Portalbuilt
JobAppin-build
Night/Dayexplored
BLACKSITE: NULLin-build
SlugBitesbuilt
Riptidebuilt
AeroBitesarchived
Internbuilt
CAD & 3D printingexplored
ForgeCouncilexplored
ADDE: adversarial due-diligence engineexplored
Model routing in practiceexplored
AI due-diligence market mapexplored
Quorumarchived
Colossus Wakeexplored
FitFindrbuilt
Project Omniexplored
Oblivionexplored
MeetWisebuilt
EyeOSexplored
LinkLeap AIexplored
Up-Toexplored
Offline speech-to-notes (Java)explored
AI-assisted game production pipelineexplored
COLLAPSEarchived
THE TRIALSarchived
EcoNodearchived
ESP32-S3 / LoRa benchexplored
CleanPlaybuilt
Rezonyrbuilt
Habit Tracker Telegram Botbuilt
Job Board Aggregatorin-build
Visual Hand Trackbuilt
CLI Task Trackerbuilt
All work44 entries
Can’t Hallucinate, Can Still Be Wrong: Calibration of a Typed-Decision Model Under Input Noisepaper
Disagreement as Signal: A Hybrid Multi-Agent and Council Architecture for Error Detection in LLM Systemspaper
Killing Good Ideasessay
Uncorrelated Failure Modesessay
ADDE: adversarial due-diligence engineexplored
AI due-diligence market mapexplored
AI-assisted game production pipelineexplored
Aura Chatexplored
Colossus Wakeexplored
EcoNodearchived
ESP32-S3 / LoRa benchexplored
EyeOSexplored
FitFindrbuilt
ForgeCouncilexplored
LinkLeap AIexplored
MeetWisebuilt
Model routing in practiceexplored
Offline speech-to-notes (Java)explored
Oblivionexplored
Print benchexplored
Project Omniexplored
Provably-fair outcome enginebuilt
Quorumarchived
Up-Toexplored
Copy hello@mitanshm.com
Switch theme
Play motion
RésuméPDF
Open GitHub ↗mit37
Open LinkedIn ↗
Toggle layout gridh
Toggle single-key shortcuts/ h