Batch AO + PBR Bakes
Run repeatable AO, normal, curvature, ID, and other agreed bakes across one asset or a folder; catch overlaps, padding, cage, tangent, and seam failures.

Your high-value work is deciding what the asset should be. I can take the repetitive production loop around that decision: audit, convert, bake, validate, document, package, and prove the handoff.

I reviewed the indexed work evidence available to me before writing this: John Deere dump-truck modeling, CAD part decisions, UV work, triangle-budget reduction, texture revisions, Unity checks, and Somero equipment updates. The pattern is clear: detailed 3D judgment surrounded by repetitive verification and packaging.
These are concrete handoffs, not generic AI suggestions. You provide the source files and acceptance notes. I return editable outputs, machine-readable checks, and visual proof. Nothing replaces your art direction.
Run repeatable AO, normal, curvature, ID, and other agreed bakes across one asset or a folder; catch overlaps, padding, cage, tangent, and seam failures.
Measure overlap, bounds, texel density, padding, mirrored/stacked regions, and waste; then render checker views so fixes are obvious before texturing.
List every mesh, duplicate, hidden part, material, transform, and bounding box; group likely keep/remove/rebuild candidates for your decision.
Find dense offenders, generate conservative reduction candidates, compare silhouettes, and preserve articulation boundaries and important close-up detail.
Batch rename, resize, convert, channel-pack, remove accidental alpha, verify color space, and flag mismatched dimensions or missing maps.
For scale, extension, decal, orientation, or texture corrections, compare before/after geometry and materials so unrelated details do not drift.
Validate scale, axes, pivots, hierarchy, materials, texture links, collision expectations, naming, and prefab-facing structure before the developer gets it.
Open every submitted model and produce geometry, materials, textures, dimensions, bones, missing links, naming drift, and renderability checks.
Turn screenshots and notes into a compact review page: issue, exact location, recommended correction, decision needed, and approval state.
Build the boring but critical delivery bundle: source, output, previews, manifest, changelog, checksums, and concise Slack-ready completion note.
Start with one real, low-risk asset. Keep the original untouched. Give me the result and constraints; I will inspect first, work on a copy, and return proof.