ForgeFX research brief · CAD → Unity / FBX · current through July 2026

Make Halliburton SolidWorks CAD usable in a Unity Windows desktop app.

The fastest low-risk path is not “find one magic exporter.” It is a two-lane pipeline: use Unity Asset Transformer/Pixyz as the primary CAD-to-game pipeline, keep CAD Exchanger or SimLab CADVRter as quick fallback converters, then run a small asset and the real Halliburton assembly through the same measurable smoke test before artists spend time cleaning it by hand.

1Primary

Unity Asset Transformer / Pixyz

Best fit for native .sldasm/.sldprt, hierarchy retention, tessellation control, decimation, LODs, occlusion removal, merge/bake workflows, and Unity handoff.

SolidWorks 97–2025 in Toolkit docsFBX / glTF / USD exportUnity-native workflow

2Fast fallback

CAD Exchanger or SimLab CADVRter

Use when we need a simple desktop converter quickly, or when Pixyz licensing/setup blocks the first day. CAD Exchanger is stronger as an SDK/automation path; SimLab is stronger as drag-and-drop trial workflow.

SLDASM→FBXWin/MacTrial-friendly

3Art cleanup

Blender + Simplygon/RapidPipeline-style optimization where needed

Use DCC tools after conversion for UV/material cleanup, texture authoring, naming, pivots, collision proxies, and final prefab packaging. Do not start with Blender alone for native SolidWorks unless using a paid importer.

UVs/materialsArtist controlPrefab-ready

Recommendation

Pick Unity Asset Transformer/Pixyz as the main CAD ingest/optimization pipeline, run CAD Exchanger and SimLab CADVRter as same-day fallback trials, and make Blender the preferred art-finishing/QA stage after CAD has been converted to FBX/GLB. ForgeFX likes Blender, and that is the right instinct for cleanup, inspection renders, material tuning, UV fixes, collision proxies, and batch FBX→GLB work. The caveat: stock Blender is not the first tool for native .sldasm/.sldprt; those proprietary assemblies need a CAD-aware front-end first.

Feature comparison

Tool / service	Native SolidWorks?	Unity/FBX path	Optimization features	Automation	Why use it	Risks
Unity Asset Transformer / Pixyz Studio, SDK, Toolkit	Yes; Toolkit docs list SolidWorks `.sldasm`/`.sldprt` 97–2025.	Export FBX/OBJ/glTF/USD, import via Unity native importers, glTFast, or Pixyz Toolkit.	Strong: tessellation, decimation, LOD generation, remove occluded geometry, retopologize/proxy, merge, bake materials.	Strong via SDK/Python; Toolkit inside Unity.	Best all-around ForgeFX fit for large CAD and repeatable Unity delivery.	Licensing/setup; source version support must match Halliburton file version.
CAD Exchanger Lab, CLI/SDK	Yes; claims SOLIDWORKS 2004–2025 and SLDASM→FBX.	Direct SLDASM/SLDPRT to FBX; Unity then imports FBX.	Moderate: conversion/viewing, less Unity-specific game optimization than Pixyz.	Strong SDK in C#, C++, Java, JS, Python; desktop app for manual tests.	Good second lane, especially if we need automation or cross-platform conversion service.	Need verify material fidelity, hierarchy, tessellation controls on our asset.
SimLab CADVRter / Composer	Yes; lists SLDPRT/SLDASM import.	Exports FBX/GLB/OBJ/USDZ and other formats; Unity imports FBX/GLB.	Moderate: tessellation controls, scene optimization language, materials/textures support.	Moderate; simpler GUI-first workflow.	Fastest non-engineer trial; likely useful for quick first look and one-off conversion.	May be less controllable/repeatable for huge assemblies; verify batch/CLI before pipeline commitment.
Okino PolyTrans\|CAD	Yes; long-running CAD/DCC translator for SolidWorks, STEP, IGES, etc.	Exports to FBX and DCC/VR/AR pipelines including Unity-oriented workflows.	Strong: polygon reduction, scene optimization, batch conversion, huge assembly focus.	Moderate/strong, but vendor-guided evaluation model.	Proven fallback for nasty enterprise CAD conversion problems.	Less simple; buying/evaluation flow is heavier than CAD Exchanger/SimLab.
Autodesk VRED → Pixyz / FBX	Indirect; depends on VRED/CAD imports.	VRED .vpb can go through Pixyz; FBX often used for downstream.	Strong for visualization scenes; variants/materials/animations can matter.	Moderate; requires VRED license locally for Pixyz VPB import.	Use only if Halliburton or ForgeFX already has VRED-prepped scenes.	Overkill for raw SLDASM/SLDPRT; material loss and license dependency.
Blender + SimLab importer	Yes with paid plugin; vanilla Blender cannot natively import SLDASM/SLDPRT.	Import SolidWorks into Blender, clean materials/UVs, export FBX/GLB.	Strong manual art cleanup; not industrial CAD optimization by default.	Moderate via Blender scripts after import.	Good art-finishing stage, especially UV/material/collision cleanup.	Plugin dependency; huge assemblies may be painful; no native SolidWorks in stock Blender.
SOLIDWORKS export → STEP/VRML/STL/PLY → DCC	Yes if opened in SOLIDWORKS	Export neutral format, convert/import to FBX/Unity.	Weak/moderate; depends on export settings and downstream tools.	Low unless SolidWorks macro pipeline is built.	Fallback if native converters fail but SOLIDWORKS access exists.	Can lose hierarchy/materials/metadata; 3DEXPERIENCE export constraints may apply.
FreeCAD / OpenCascade tools	No reliable native SLDASM/SLDPRT	Useful for STEP/IGES; then OBJ/mesh export through DCC.	Limited for game-ready optimization.	Scriptable.	Useful free helper for neutral CAD, not the first move here.	Native SolidWorks proprietary format is the blocker; large assembly workflow likely brittle.

Mini smoke-test plan

Test A — small sample asset

Pick a known small SolidWorks assembly with 5–20 parts, visible material colors, nested assembly hierarchy, and at least one curved/high-tessellation part.
Run the same source through Pixyz, CAD Exchanger, SimLab CADVRter, and Blender+SimLab importer if available.
Export FBX and GLB where supported; import both into a clean Unity 6 Windows desktop test project.
Score each output: import success, scale/orientation, part hierarchy, pivots, materials, mesh normals, triangle count, draw calls, missing parts, and Unity frame time.
Create one prefab per candidate with standard material remap, collider proxy, and camera orbit scene; capture screenshot and profiler numbers.

Pass/fail gates

Imports without manual repair.
Hierarchy is still usable for selecting interactive parts.
No obvious missing geometry or inverted normals.
Can hit 60 fps on target Windows dev workstation for the small asset.

Test B — Halliburton large asset

Copy the SharePoint folder locally, preserving every .sldasm/.sldprt relative path. Assemblies will fail if referenced part files are missing.
Inventory file count, total size, top-level assemblies, largest parts, and missing external references before conversion.
Run only the top-level assembly through the same 2–3 best tools from Test A; record conversion time, peak memory, output size, crash/log errors, and missing reference reports.
In Pixyz, create three quality profiles: hero/interactive, mid/background, proxy/impostor. Test decimation, remove-occluded, merge-by-material/static-group, LOD generation, and material baking.
In Unity, test Windows desktop scene load, orbit camera, selection outline on candidate interactive parts, material readability, and profiler frame time.

Large-asset gates

Top-level assembly opens with no missing critical parts.
Optimized Unity scene stays within target load time and memory budget.
Interactive subassemblies remain separate; non-interactive internals can be merged/culled/proxied.
Artists can identify where manual UV/material work is actually worth doing.

Concrete next three steps

Set up the conversion bench today: install/request trials for Unity Asset Transformer/Pixyz, CAD Exchanger Lab, and SimLab CADVRter on one Windows workstation with Unity 6, Blender, and a clean Unity desktop smoke-test project.
Run the small sample through all candidates: capture one comparison table with conversion time, triangle count, draw calls, material/hierarchy quality, plus a Blender inspection/render pass and Unity screenshot/profiler evidence.
Promote the winner to the Halliburton folder: copy the full SharePoint assembly locally, identify the top-level .sldasm, run the winning conversion profile, pass the FBX/GLB through Blender for inspection/material/UV/collision cleanup, and split results into interactive hero parts, static optimized parts, and deprioritized/internal parts for Brandon/Jesto review.

External source coverage

Pixyz is now Unity Asset Transformer; current downloads include SDK 2026.4, Studio 2026.4, Toolkit 3.3.
Unity Asset Transformer SDK export docs: FBX, glTF/GLB, USD, Unity-native importer and glTFast workflow options.
Unity Asset Transformer Toolkit supported formats: SolidWorks .sldasm/.sldprt, STEP, IGES, FBX, glTF and more.
Unity optimization guidelines: decimation, occlusion removal, retopology/proxy meshes, LODs, merge/bake for draw-call reduction.
CAD Exchanger SolidWorks→Unity and SLDASM→FBX: CAD Exchanger Lab/SDK, SolidWorks 2004–2025, FBX 7.2/7.4/7.5 export.
SimLab CADVRter: local CAD conversion, SLDPRT/SLDASM import, FBX/GLB export, tessellation controls, materials/textures.
Okino PolyTrans|CAD: CAD/DCC translator updated July 1, 2026, large assembly conversion, optimization, polygon reduction, batch conversion.
SOLIDWORKS 2026 export docs: export paths to STEP, IGES, STL, VRML, etc.; useful fallback but not a direct game-ready pipeline.
SimLab SolidWorks to Blender importer: native import via plugin; vanilla Blender does not natively import SLDPRT/SLDASM.

Internal source coverage

Used: this Slack thread only. Devin reports native SolidWorks .sldasm/.sldprt files are uploading to the Halliburton Cementing Services SharePoint “Elite phase 1 batch” folder; Dave’s note asks for AI-assisted conversion, texture/UV help, and game-usable geometry; Brandon plans model review.

Not directly inspected: the SharePoint files themselves, Jira/GitHub project case details, and channel history outside the inline thread context. The smoke test intentionally includes first-step inventory and missing-reference checks because the actual uploaded assembly tree has not been verified inside this report.

Confidence: high on tool shortlist and pipeline shape; medium on final winner until the small sample and Halliburton assembly are actually converted and profiled.