CLI slicing 3MF fails with "nozzle_volume_type not found" + assertions (P2S 0.4)

[tylerreece22]

Bambu Studio Version

02.05.00.66 (built from source, arm64; same as release tag)

Where is the application from?

Built from source by yourself or others

OS version

macOS 26.1 (Build 25B78)

Additional system information

Architecture: arm64

Printer

Bambu Lab P2S

Printer Firmware Version

Unknown (CLI slicing only; printer not connected)

How to reproduce

1. Use BambuStudio CLI 02.05.00.66 (built from source) on macOS arm64.
2. Run:
   BambuStudio --allow-newer-file --slice 0 --export-3mf /tmp/out.3mf Millenium_Falcon_100%_infill.3mf
   (3MF was resaved from Bambu Studio 2.5.0.66; printer preset: Bambu Lab P2S 0.4 nozzle; process: 0.20mm Standard @bbl P2S; filament: Bambu PLA Basic @bbl P2S)
3. Same result with:
   --uptodate
   --load-settings ".../machine/Bambu Lab P2S 0.4 nozzle.json;.../process/0.20mm Standard @bbl P2S.json" --load-filaments ".../filament/Bambu PLA Basic @bbl P2S.json"

Actual results

CLI logs:

• 3207: nozzle_volume_type not found, when different_extruder or multiple extruder, new_printer_name Bambu Lab P2S 0.4 nozzle, extruder_count 1
  Then the CLI crashes with an assertion, e.g.
• Assertion failed: (std::abs(d2new - clip_dist_scaled2) < 1e-5 * clip_dist_scaled2), function smooth_outward, file MutablePolygon.cpp, line 330.
• Assertion failed: ((sign1.first == 0 && sign1.second == 0) == (sign2.first == 0 && sign2.second == 0)), function segments_intersect, file Geometry.hpp, line 149.
  No output 3MF is produced.

Expected results

CLI should slice the 3MF and export a sliced 3MF (or G-code) without crashing.

Project file & Debug log uploads

Attached:

• Project file zip: Millenium_Falcon_100_infill.zip
• Debug log zip: bambu-debug-log.zip

bambu-debug-log.zip

Checklist of files to include

• Log file
• Project file

[glacierck]

[2026-02-09 21:42:27.395972] [0x00000001f9a86240] [error] 3207: nozzle_volume_type not found, when different_extruder or multiple extruder, new_printer_name , extruder_count 1
[2026-02-09 21:42:27.395979] [0x00000001f9a86240] [warning] no filament colors found in projects

[tkoenig]

I hit a related crash in PrintObject.cpp — calc_estimated_filament_print_time() unconditionally dereferences filament_extruder_variant and filament_self_index options, which don't exist in older/simpler 3MF files. Adding null checks with fallback to empty vectors fixes it.

There's also a separate CLI segfault in PartPlate.cpp (see #4627). Full disclosure: the diagnosis and fix were 100% done by AI (Claude via pi). I have zero familiarity with this codebase, so please review carefully. Both fixes are on my fork: tkoenig#1

[rowbotik]

I hit what looks like the same root class of failures while driving
BambuStudio from CLI mode for BBL profiles.

Short version: the GUI appears to slice with fully resolved preset state, but
the CLI path accepts leaf profile JSONs that still depend on BBL inherits
chains. In CLI mode those leaf files do not appear to get normalized the same
way the GUI-normalized project does, so required derived fields can be missing
or internally inconsistent by the time slicing starts.

I worked around this externally by flattening the BBL profile inheritance chain
before invoking the CLI:

1. Index every bundled profile JSON under Resources/profiles/BBL/{machine,process,filament} by its name.
2. Walk each requested leaf profile's inherits chain recursively.
3. Deep-merge parent to child so child values win.
4. Set CLI-facing identity fields so validation still resolves against the
   original system preset:
   • machine: from: "User", inherits: <leaf machine name>
   • process/filament: from: "User", inherits: <original inherited preset name>
   • printer_settings_id, print_settings_id, filament_settings_id
5. Derive missing machine fields that the GUI-saved project has but the CLI
   leaf input may not, especially nozzle_volume_type.
6. Apply the relevant BBL/cli_config.json machine-limit overlay before
   writing the temporary flattened configs.

With that preprocessing in place, the same BambuStudio CLI path generated
valid sliced .3mf output for:

• H2S 0.4 nozzle
• H2D 0.4 nozzle
• X1C 0.4 nozzle
• P1S 0.4 nozzle

The output archives contained Metadata/plate_1.gcode with non-empty gcode in
each case. This also resolved the specific class of missing
nozzle_volume_type failures for the H2/P/X profiles I tested.

The practical implementation is here:

• src/slicer/profile-flatten.ts
• src/stl/stl-manipulator.ts behind BAMBU_CLI_FLATTEN=true
• tests: tests/slicer/profile-flatten.test.mjs

Repo branch:

• https://github.com/rowbotik/bambu-printer-mcp/tree/codex/collar-charm-h2-cleanup

This is probably not the ideal long-term fix because the CLI should not require
external callers to reconstruct BambuStudio's preset resolver. But it suggests
a small upstream direction: add an internal --auto-flatten / --resolve-presets
step, or run the existing GUI preset normalization code before CLI slicing when
--load-settings / --load-filaments point at BBL system profiles.

I can turn the workaround into a smaller upstream patch if maintainers can point
me at the preferred preset-resolution path for CLI mode.

Local evidence:

npm test
for m in h2s h2d x1c p1s; do
  node scripts/test-cli-slice.mjs --model "$m"
done

Current result:

• npm test: 19/19 passing
• H2S smoke: Metadata/plate_1.gcode present, 110,970 bytes
• H2D smoke: Metadata/plate_1.gcode present, 116,704 bytes
• X1C smoke: Metadata/plate_1.gcode present, 111,415 bytes
• P1S smoke: Metadata/plate_1.gcode present, 111,313 bytes