Skip to content

DATA_FORMAT_COMPARISON.md

Latest commit

 

History

History
131 lines (102 loc) · 3.5 KB

DATA_FORMAT_COMPARISON.md

File metadata and controls

131 lines (102 loc) · 3.5 KB

XPD Data Format Comparison

Summary

The XPD reader successfully handles two different spline data layouts based on primSize:

Format 1: Full Grooming Data (primSize = 51)

Files: box.xpd Use case: Complete grooming/hair data with detailed guide and styling information

Data Layout (51 floats):

  1. primitiveId (1)
  2. surfaceUV (2)
  3. CVs positions (15 = 5 CVs × 3)
  4. guideId (1)
  5. guideWeight (1)
  6. guideType (1)
  7. guideDirection (3)
  8. guidePrimRef (1)
  9. guideUV (2)
  10. surfaceNormal (3)
  11. cvParameters (15 = 5 CVs × 3)
  12. widthScale (4)
  13. additionalAttributes (2)

Total: 51 floats

Format 2: Basic Spline Data (primSize = 25)

Files: sample.xpd, spline_1_taper_1.xpd Use case: Simple spline data without detailed styling

Data Layout (25 floats):

  1. primitiveId (1)
  2. surfaceUV (2)
  3. CVs positions (15 = 5 CVs × 3)
  4. guideId (1)
  5. guideWeight (1)
  6. guideType (1)
  7. guideDirection (3)
  8. guidePrimRef (1)

Total: 25 floats

Validation Results

✅ box.xpd

  • primSize: 51
  • numFaces: 6
  • Total primitives: 144 (24+25+25+25+24+21)
  • Parser status: ✅ All fields parsed correctly
  • Output: box.json (351KB)

✅ sample.xpd

  • primSize: 25
  • numFaces: 444 (sparse, ~100 primitives total)
  • Total primitives: ~100 (scattered across faces)
  • Parser status: ✅ Correctly handles smaller format
  • Output: samples/sample.json

✅ spline_1_taper_1.xpd

  • primSize: 25
  • numFaces: 13
  • Total primitives: 1 (on face 12)
  • Parser status: ✅ Correctly handles smaller format
  • Output: samples/spline_1_taper_1.json

Parser Behavior

The parser uses conditional logic to handle different data sizes:

// Parse guide data based on XGen spline format
if (remaining >= 7) {
  // Parse guide id, weight, type, direction, primRef
  ...
}

// Only parse additional fields if they exist
if (remaining >= 9) {
  // Parse guide UV
  ...
}

if (idx + 2 < offset + floats_per_prim) {
  // Parse surface normal
  ...
}

// etc.

This means the parser gracefully handles:

  • Full format (51 floats): Parses all available fields
  • Basic format (25 floats): Stops after guide data, doesn't try to read non-existent fields
  • Future formats: Will work with any primSize as long as the basic fields are present

Common Fields

All formats share these core fields:

Primitive Identification

  • primitiveId: Unique spline ID
  • surfaceUV: [u, v] attachment point on surface

Geometry

  • cvs: Array of 5 control vertices with positions [x, y, z]

Guide Reference

  • guideId: Which guide curve this follows
  • guideWeight: Influence (typically 0.1 = 10%)
  • guideType: Type of guide (0 = standard)
  • guideDirection: Tangent vector at attachment
  • guidePrimRef: Guide primitive reference index

Extended Fields (primSize = 51 only)

  • guideUV: UV coordinates on guide surface
  • surfaceNormal: Normal vector at attachment point
  • cvParameters: Per-CV parametric data (t-values 0.0-1.0)
  • widthScale: Width/taper control (base, tip, params)
  • additionalAttributes: Extra styling data

Conclusion

✅ Parser is robust and handles all tested formats correctly

The conditional parsing logic ensures compatibility with:

  • Different primSize values (25, 51, potentially others)
  • Different primVersion values (all tested files use version 3)
  • Sparse data (many faces with 0 primitives)
  • Various mesh topologies (6 faces to 444 faces)