agentleFS
JiamingShi-JsPhysJiamingShi-JsPhys/3d-fabric/skills/reference-analysis-validator/SKILL.md

reference-analysis-validator

Skill2 starsChanged 3 days agoMITView on GitHub

Measure and validate supplied reference images, wireframes, texture atlases, and Blender renders before declaring a reconstruction 1:1. Use when an asset must match a template, when visual feedback says the output is off, when part counts must be exact, or before exporting a brand mascot/logo reconstruction. Pairs with reference-to-3d, contour-to-mesh, orthographic-registration, atlas-uv-fitting, and Blender MCP.

1 other project uses a version of this, 1 of them edited.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

4 rules. About 0 are common, 4 are distinctive, and 0 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈4

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 0

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Use when: Any 1:1 reference/model validation task involving masks, overlays, part counts, centroids, bounding boxes, SSIM/IoU, source …

    line 10
  2. Distinctive:

    Before the first Blender call, read ../text-to-blender/references/mcp-backends.md.

    line 12
  3. Distinctive:

    Create these in the asset output folder:

    line 19
  4. Distinctive:

    If a repeated mismatch occurs, record the measured failure, then route to the missing specialty skill:

    line 49

≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

MCP backend

Use when: Any 1:1 reference/model validation task involving masks, overlays, part counts, centroids, bounding boxes, SSIM/IoU, source manifests, or fail-before-export gates.

Before the first Blender call, read ../text-to-blender/references/mcp-backends.md. Prefer official MCP for shared capabilities; preserve the selected backend for the whole task. Tool names in recipes are client/legacy examples: map them through that reference and its agent-host guide.

Reference Analysis Validator

This skill converts “looks close” into measurable gates. For brand/logo/mascot work, do not model or export until a source manifest and validation thresholds exist.

Required outputs

Create these in the asset output folder:

  • reference_manifest.json — classified source files, expected parts, thresholds.
  • source_analysis/*.json — image metadata, masks/components/landmarks.
  • validation/front_overlay_reference.png — reference and render overlay.
  • validation/front_mask_validation.json — IoU/SSIM/bbox/centroid report.

Workflow

  1. Classify sources: front, side, back, top, texture atlas, decals, maps, lightmap, aura/context.
  2. Build/refresh reference_manifest.json with hard expected counts and view roles.
  3. Extract masks/components from each source using scripts/reference_manifest_compiler.py or existing analyzers.
  4. Render model from matching orthographic camera with reference planes hidden.
  5. Compare reference mask vs render mask using scripts/render_overlay_validator.py.
  6. Refuse final export if hard gates fail.

Modality rule

Compare like with like. A wireframe edge mask compared against a shaded beauty render gives misleadingly low IoU. For hard gates, render a flat silhouette/matte pass from Blender or compare reference edges to render edges. Use render_overlay_validator.py --reference-mode ... --render-mode ... when the source and render need different mask extraction modes.

Default validation gates

  • primary structural part count: exact.
  • front silhouette IoU: target >= 0.90 for rigid/logotype shapes; >= 0.82 acceptable for first mascot reconstruction pass.
  • bbox center drift: <= 12 px at 1024 px validation size.
  • bbox size drift: <= 3% of image dimension.
  • face/eye/smile landmark drift: <= 2% of image dimension when landmarks are defined.

Failure policy

If a repeated mismatch occurs, record the measured failure, then route to the missing specialty skill:

  • wrong silhouette → contour-to-mesh
  • wrong depth/side/back → orthographic-registration
  • wrong textures → atlas-uv-fitting
  • wrong whole workflow → mascot-logo-reconstruction

Read when needed

  • references/metrics-and-thresholds.md for metric definitions and recommended gates.

Sources distilled

Official/library docs to prefer while extending this skill:

  • OpenCV contour features: moments, area, perimeter, bounding rectangles.
  • OpenCV shape matching / Hu moments.
  • OpenCV homography and geometric transforms.
  • scikit-image SSIM for perceptual comparison.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.