PerfectFit 1.3.2

A correctness patch, driven almost entirely by one very thorough community report. Thanks to Guntram on the Discord, who ran the full procedure against OrcaSlicer 2.4.2 and documented every place our instructions did not match what is actually on screen — and to RThomasHyde for the Linux AppImage report.

⚖️ PerfectFit is now licensed under AGPL-3.0

Starting with this release, PerfectFit moves from its old custom non-commercial license (R3D-NC v1.0) to the GNU Affero General Public License v3 — the same license as OrcaSlicer, PrusaSlicer, and Slic3r.

For most people this is a loosening, not a restriction. The old license banned commercial use entirely; AGPL-3.0 permits it. You may now use, modify, sell, or host PerfectFit freely. The one obligation: if you distribute a modified version — or run one as a network service — you must make your source available under the same license.

The reason is compatibility. PerfectFit is built around the Orca-family slicers, and upcoming releases will integrate with them far more directly. AGPL section 7 forbids adding restrictions like "non-commercial" to a work that combines AGPL code, so the old license was a dead end for that roadmap.

Releases up to and including 1.3.0 remain available under their original terms.

Highlights

• ⚠️ Printer database was corrupted for 374 of 379 printers — now fixed. Please refresh your saved printers. A bug in the spreadsheet parser meant every blank cell stole the next cell's value, and 250 printers ended up with a max nozzle temperature of 27 °C or 69 °C. Because the app clamps the suggested temperature range to that limit and rejects anything above it, the temperature calibration step was effectively unusable for two-thirds of the printer database.

  Specs are copied into your printer profile when you add it, so fixing the database alone does not repair a printer you already saved. The app now offers to do it for you: open Printers and any affected profile shows a "↻ Updated specs available" callout. Reviewing it lists every change — Max nozzle temp: 27 °C → 300 °C — before anything is written, and refreshing keeps your printer name, notes, retraction range, and the profile itself, so projects using that printer stay linked.

  Don't delete and re-add the printer. Re-adding creates a new profile, and any project that referenced the old one loses its printer limits, range suggestions, and safety caps. (Your projects and calibration results live separately and are never deleted with a printer — but the link isn't restored by re-adding.) If you hand-tuned specs for modified hardware, cancel the refresh and edit the fields directly instead.

• The generator now rejects physically impossible values. validate:printers only checked that the JSON matched the workbook — which it faithfully did, which is why a 27 °C maximum nozzle temperature sailed through. Numeric specs are now range-checked when the database is generated, and anything implausible is stored as "not specified" and reported with its row number. It flagged 23 genuine data problems on first run, including a printer listed at 300 mm³/s of volumetric flow — all corrected in this release.

• Slicer menu paths corrected in both directions. Orca and Bambu Studio name the same tests differently, and several paths had drifted onto the wrong slicer's wording. Orca is Calibration → Flow ratio, Calibration → Retraction, and Calibration → Max flowrate at the top level; Bambu Studio is Calibration → Flow rate → Coarse/Fine, Retraction test, and More... → Max flowrate. The Bambu pressure advance path pointed at a menu entry that does not exist and now distinguishes the manual Pressure advance test from the machine's automatic Flow Dynamics wizard.

• Temperature values are handed over in the slicer's own field order — first layer, then other layers — matching how both slicers lay out the Nozzle line.

• The "Resonance avoidance" dialog is explained. Orca disables the setting for every calibration test by design; only one stock profile in Orca's catalog ships it enabled, so a small number of users saw an unexplained dialog. Transfer or Discard makes no difference, and the app now says why.

• Multi-filament printers get a warning about slot assignment. Orca's tests always target filament slot 1 and expose no extruder or slot picker anywhere, so the test can silently run with the wrong filament. This covers both toolchangers (Snapmaker U1, Prusa XL) and AMS/MMU machines (X1 Carbon, P1S, A1, MK4S) — an AMS bay and a separate toolhead are the same thing as far as the slicer is concerned. Both workarounds are spelled out.

• New prerequisite: confirm the selected printer preset matches the nozzle actually installed — the tests scale the model and layer height to the preset's nozzle diameter, so the wrong variant invalidates the result silently.

• Blank window on Linux/Wayland (AppImage) fixed — the app now disables WebKitGTK's DMABUF renderer, whose EGL initialisation fails on some Wayland setups.

Saved projects, printers, and backups all keep working — the only change to existing data is the opt-in spec refresh described above, which you review before it applies.

(1.3.1 was tagged but superseded before publication; 1.3.2 contains everything from it, so upgrading from 1.3.0 gets the lot.)

Downloads

• Windows: NSIS installer (.exe).
• macOS: universal Apple Silicon + Intel disk image (.dmg).
• Linux: Debian package (.deb) plus portable AppImage (.AppImage).

PerfectFit 1.3.0

Highlights

• Printer specification database: pick your printer from 379 known models across 64 manufacturers when adding a printer, and the machine specs — temperature limits, extruder type, max volumetric flow, build volume, supported nozzle sizes, heated chamber, firmware and more — fill in automatically. A searchable, manufacturer-grouped combobox handles the large list, and every value stays editable for modified or custom hardware. Bear with us as some specs on printers may not be accurate and exist only as placeholders while we gather all of the information needed on all printers currently in the database. Double check and change anything that is wrong / missing.
• “My printer is not listed” still gives you full manual entry — nothing about the previous workflow was removed. Manually configured printers are clearly labelled.
• Smarter safety guidance: the database’s heated-chamber and supported-nozzle data drive extra warnings (e.g. enclosure materials on a printer with no heated chamber), on top of the existing max-temperature and max-flow guardrails.
• Maintainable data pipeline: the database is edited in Printer_Database/Printer_Database.xlsx, regenerated with npm run generate:printers, and committed as src/data/printers.json. See “Updating the Printer Database” in the README.
• Local-time backups: slicer preset backup timestamps in Settings now display in your PC’s local time instead of UTC.

Existing saved printers and projects continue to work unchanged; the printer profile schema change is additive, and older backups migrate on import.

Downloads

• Windows: NSIS installer (.exe).
• macOS: universal Apple Silicon + Intel disk image (.dmg).
• Linux: Debian package (.deb) plus portable AppImage (.AppImage).

PerfectFit 1.2.0

A community-driven release — everything in it started as feedback from confuzled on the community Discord. Full notes: docs/RELEASE_NOTES_1.2.0.md in the repository.

Highlights

• Back up your slicer profiles BEFORE calibrating: a first-run prompt, a per-project prompt before any profile edits are suggested, and an on-demand button in Settings — all checksummed and restorable file-by-file from Settings → Slicer profile backups.
• New step — Flow Ratio Re-check (after Pressure Advance): PA changes what you judged in the flow test; the re-check confirms or corrects the saved ratio with one fine plate.
• New step — Shrinkage / Dimensional Accuracy: ap.engineering's free Printables calibration plate (spreadsheet scale-error entry or direct measurements), Vector3D's paid CaliFlower MK2, or any large measured object. Results flow into reports, cards, and generated profiles (filament_shrink).
• Honest drying advice: "fresh from a sealed bag" no longer counts as dry — PETG/TPU/PCTG often arrive wet from the factory; the checklist and material warnings now say so, with drying temperatures.
• Bambu Studio Develop Mode described accurately: the Preferences checkbox really is labeled "Develop Mode", and the calibration tests live under the title-bar Calibration button next to the Redo arrow (like all Orca-based slicers). All Bambu menu paths corrected.
• Each test names the profile to modify, with detected-preset suggestions when creating a project (desktop app).

Existing projects gain the two new steps as "not started" — progress and scores are preserved; both steps are skippable.

Downloads

• Windows: NSIS installer (.exe).
• macOS: universal Apple Silicon + Intel disk image (.dmg).
• Linux: Debian package (.deb) plus portable AppImage (.AppImage).

PerfectFit 1.1.5

Patch release for the experimental Slicer Profile Installer. 1.1.4 fixed the
wizard so it recommends stock baselines — and thereby exposed the next bug:
profiles cloned from a stock preset were installed correctly but never
appeared in a signed-in Bambu Studio. Diagnosed by structurally comparing a
PerfectFit-installed preset against presets Bambu Studio 2.7.x itself writes
into the same account folder (H2S).

Fixes

Profiles cloned from stock presets now appear in Bambu Studio

Diagnosed by surveying every preset Bambu Studio 2.7.x itself had written
into a real signed-in account folder (70+) against the invisible
PerfectFit-installed ones, and matching the invariants exactly:

1. Stock-preset plumbing removed. Clones carried type,
   instantiation, and include — keys no Bambu-written user preset has.
   include is the worst: it references template files that don't resolve
   from a user folder. All three are stripped; everything they provided
   still flows through inherits.
2. filament_extruder_variant added. Every visible preset declares this
   legend mapping per-slot values to hardware (on an H2S:
   ["Direct Drive Standard","Direct Drive High Flow"]). Variant-aware
   Bambu Studio does not show user presets without it.
3. version stamped from the vendor manifest. Every visible preset
   carries the vendor library version from system/BBL.json (zero-stripped:
   02.07.00.08 → 2.7.0.8) — a value no preset inside the library
   declares. The native scan now reads vendor manifests to supply it.
4. Concrete inherits. Clones kept the stock leaf's own inherits
   (an abstract @base preset). Bambu saves user presets inheriting the
   concrete system preset by name; clones now do the same.
5. Fresh filament_id, always. The 1.1.3 fix only fired when the base
   declared an id — stock leaves inherit theirs, so clones of stock presets
   had none. Now always assigned; validation blocks missing/colliding ids.
6. Account user_id in the .info sidecar, stamped at install time
   (local, non-account installs stay empty), matching Bambu-written presets.

If a PerfectFit - … preset you installed earlier never showed up, reinstall
this build and re-run 🧵 Create Slicer Profile; installing under the same
name replaces the invisible file.

The wizard no longer calls the H2S a dual-nozzle printer

Bambu filament presets index per-slot arrays by (tool × hotend variant). On
single-nozzle printers with interchangeable hotends (H2S, P1S, …) the two
slots are the Standard and High Flow hotends — not two nozzles. The
"Multi-tool profile" step (now "Per-tool / per-hotend values") explains both
meanings and labels the slots, so calibrated values land in the slot matching
the hotend you actually calibrated with. Pick slot 1 for the Standard hotend,
slot 2 for High Flow.

Unchanged

Everything from 1.1.4 stands: the startup-hang fix (stale service worker),
recursive stock-preset scanning, inherited-metadata resolution, and the
stock-only baseline recommendations.

PerfectFit 1.1.4

Patch release for the experimental Slicer Profile Installer. 1.1.3 promised
that baseline suggestions would be stock profiles compatible with the selected
printer, but on real Bambu Studio installs the fix was starved of data and the
wizard still fell back to user presets (reported with a Bambu Lab H2S). 1.1.4
fixes the three underlying defects, verified against a real Bambu Studio 2.7.x
install with an H2S.

Fixes

App no longer hangs on "Loading PerfectFit…" after an update

Updating (or uninstalling and reinstalling) could leave the app stuck on the
static loading screen. Cause: the PWA service worker was also registered
inside the Tauri desktop webview and cached index.html cache-first. After
an update it kept serving the previous version's HTML, which references a
hashed JS bundle that no longer exists — so the app never started. WebView2
profile data survives uninstall, which is why reinstalling didn't help.

Fixed in three layers: the desktop app no longer registers a service worker
(and unregisters any left behind by older versions); on Windows it removes
stale service-worker and HTTP-cache directories from the WebView2 profile at
startup, before the webview loads, which repairs installs already stuck; and
the service worker used by real web/PWA deployments is now network-first for
HTML so it can never pin an old shell again. Calibration data (IndexedDB /
localStorage) is not touched by any of this.

Stock baselines are now found and correctly matched

Three related defects, all diagnosed on a live install:

1. The system-library scan was not recursive. Bambu Studio keeps some
   vendor presets in subdirectories (e.g.
   system/BBL/filament/{P1P, Polymaker, SUNLU}/ — 150 presets on the dev
   machine), and the native scan only read the top-level folder. System scans
   now recurse (depth-limited).
2. Printer-specific stock leaves could not be material-matched. Presets
   like Bambu ABS @BBL H2S declare compatible_printers but inherit
   filament_type/filament_vendor from abstract parents, so the scanner saw
   them with no material: they scored below user presets and "qualified" for
   every material. The scanner now resolves inherited metadata through the
   system inheritance chain. The same resolution fills compatible_printers
   for user delta presets, which previously looked compatible with every
   printer and polluted fallback suggestions.
3. Recommendations now require an affirmative material match. Presets
   whose material remains unknown are no longer recommendable; they stay
   available under Advanced selection.

The scan is no longer silent when it fails

Wizard step 2 now shows a scan summary
(Scanned N preset(s): X stock · Y user · …) with an explicit warning when
zero stock presets arrive from the scan. If stock baselines are ever missing
again, the wizard says so instead of quietly suggesting user presets.

Changes

• The feature is now called "Create Slicer Profile" everywhere — the
  project-page button, the wizard page title, and the re-run button on the
  generated-profiles card all use the same name, matching the release notes
  and documentation (previously three different labels; see
  #10).
• Create Slicer Profile is reachable from the dashboard. Project cards
  show a 🧵 Create Slicer Profile button as soon as the project has at least
  one calibrated value — no need to open the project first.

Notes for existing users

• Reinstall this build, then re-run 🧵 Create Slicer Profile (dashboard
  card or project page). With a Bambu printer selected, step 2 should now
  recommend stock Bambu <material> @BBL <printer> (or Generic) baselines and
  show the scan-summary line.
• The 1.1.3 guidance for removing a cloud-stuck preset from a signed-in Bambu
  account still applies; see docs/RELEASE_NOTES_1.1.3.md.

Unchanged

Everything else from 1.1.0/1.1.3 stands: slicer detection, clone-and-patch
generation, validation, export, backup/restore, and verified direct install
for Orca Slicer, Bambu Studio, Snapmaker Orca, ElegooSlicer, and Flash Studio
on Windows.

Changelog
1.1.4 - Unreleased

Small discoverability release prompted by #10: the 1.1.3 notes told users to "re-run Create Slicer Profile", but the feature went by three different names in the app and had no entry point on the dashboard, so it couldn't be found by that name.

Changed
The profile feature is now called "Create Slicer Profile" everywhere. The project-page button (previously "Create slicer profile"), the wizard page title (previously "Create and Install Filament Profile"), and the re-run button on the generated-profiles card (previously "Open profile wizard", now "Re-run Create Slicer Profile") all use the same name, matching the release notes and documentation.
Create Slicer Profile is now reachable from the dashboard. Project cards show a 🧵 Create Slicer Profile button as soon as the project has at least one calibrated value — no need to open the project first.
Clarified the 1.1.3 "Notes for existing users" in the changelog and release notes: the profile is regenerated in PerfectFit (project page → 🧵 Create Slicer Profile), not via Bambu Studio's "Create New" dialog, which does not know about PerfectFit calibration data.
1.1.3 - 2026-07-20

Patch release fixing two profile-installer bugs found while using the 1.1.0 build with Bambu Studio. See docs/RELEASE_NOTES_1.1.3.md.

Fixed
Installed Bambu profiles now appear in the slicer. In 1.1.0 a profile installed for a signed-in Bambu account was written correctly but never showed up in the filament list. Cause: when signed in, Bambu Studio dedupes filament presets by filament_id, so a clone that kept its parent's filament_id was hidden behind the cloud-synced parent it was cloned from. Confirmed directly in Bambu Studio 2.7.x — a copy with a fresh filament_id appears immediately; the colliding one never does. Fix: generated presets now get a fresh unique filament_id, and the .info base_id chains to the stock/system ancestor instead of a parent user preset's cloud id.
Baseline suggestions are now stock profiles compatible with the selected printer. In 1.1.0 the "select a base profile" step suggested the user's own custom presets (some flagged as incompatible with the printer). It now recommends only stock (system) profiles — brand-name or generic — for the calibrated material that are compatible with the selected printer. User and incompatible-printer presets remain available under Advanced selection.
Notes for existing users
Reinstall this build for the fixes to take effect (the fix applies to newly generated profiles).
A profile installed by 1.1.0 into a signed-in Bambu account is stuck in Bambu's cloud with the colliding id; editing local files won't unhide it. Remove it in Bambu Studio: select the preset, open it for editing (the edit/pencil icon opens the Filament settings dialog), and click the small 'X' (delete) icon in the upper-right of that edit dialog — this removes it from your cloud sync. Then regenerate the profile in PerfectFit (not Bambu Studio): open your calibration project from the PerfectFit dashboard and click 🧵 Create Slicer Profile on the project page, then follow the wizard through to install/export. Your calibration data is preserved in PerfectFit, so no re-calibration is needed.
1.1.0 - 2026-07-19

See docs/RELEASE_NOTES_1.1.0.md for the full release notes.

Added
Linux desktop release packaging via .deb and AppImage artifacts.
Experimental slicer profile generation and direct install workflows for supported Orca-family slicers.
Bambu Studio Developer mode guidance for manual calibration tests with Bambu printers selected.
Regression tests covering Bambu Developer mode instructions, coarse/fine Flow Rate wording, and VFA mention.
Changed
Release workflow now builds Windows, macOS, and Linux artifacts into draft GitHub releases.
README and research notes now document Bambu Developer mode availability for Retraction, Max Flow Rate, and VFA.
Known limitations
Linux packages are generated, but Linux native slicer detection/install behavior is not yet verified.
macOS native slicer detection/install behavior remains export-oriented pending real-machine verification.

PerfectFit 1.1.0

PerfectFit 1.1.0 adds Linux desktop packages alongside the existing Windows and macOS release assets, and rolls up the slicer profile installer work, Bambu Studio Developer mode calibration guidance, documentation improvements, and project-health updates completed since 1.0.0.

Downloads

• Windows: NSIS installer (.exe).
• macOS: universal Apple Silicon + Intel disk image (.dmg).
• Linux: Debian package (.deb) plus portable AppImage (.AppImage) for broad distro coverage.

Highlights since 1.0.0

• Added experimental slicer profile generation and direct install workflows for Orca Slicer, Bambu Studio, Snapmaker Orca, ElegooSlicer, and Flash Studio Desktop. ***These are experimental and I've been having mixed results.
• Added transactional backups, restore support, diff previews, profile compatibility scoring, and validation so generated profiles only patch values backed by completed calibration steps.
• Verified Windows direct-install flows against real slicer account directories for Orca Slicer and Bambu Studio, including Bambu dual-nozzle array preservation.
• Updated Bambu Studio coached wizard pages to recommend Developer mode as the best way to access manual calibration tests while a Bambu printer remains selected.
• Documented Bambu Developer mode availability for Temperature, Flow Rate coarse/fine (no YOLO), Pressure Advance / Flow Dynamics, Retraction, Max Flow Rate, and VFA.
• Added Linux release packaging via .deb and AppImage artifacts.
• Added issue templates, contribution guidance, security policy, code of conduct, pull request template, and expanded automated coverage.

Known limitations

• Linux native slicer profile detection/install behavior is not yet claimed as verified; Linux packages are provided so users can run the desktop app and use browser-equivalent/export workflows while native Linux slicer integration is validated.
• macOS native slicer path detection remains documented but not fully real-machine verified.
• Cloud-synchronized Orca/Bambu account directories may later sync, duplicate, re-id, or remove local preset files; PerfectFit surfaces this caveat and keeps backups for installed profiles.
• VFA is mentioned where Bambu Developer mode exposes it, but PerfectFit does not currently have a dedicated VFA scoring step.