JavaScript converter from VRML 1.0 to VRML97 file format, based on Wings 3D conversion logic.
Go to file
Glen Whitney 6892001b8d feat: support Viewpoints, translating PerspectiveCamera (#21)
Reviewed-on: #21
Co-authored-by: Glen Whitney <glen@studioinfinity.org>
Co-committed-by: Glen Whitney <glen@studioinfinity.org>
2024-02-18 19:57:29 +00:00
etc chore: Prepare package for publication (#8) 2023-09-05 01:12:55 +00:00
src feat: support Viewpoints, translating PerspectiveCamera (#21) 2024-02-18 19:57:29 +00:00
tools chore: Prepare package for publication (#8) 2023-09-05 01:12:55 +00:00
.gitignore feat: install moo and use in toy example (#2) 2023-09-01 21:32:25 +00:00
LICENSE Initial commit 2023-08-31 07:05:29 +00:00
README.md feat: support Viewpoints, translating PerspectiveCamera (#21) 2024-02-18 19:57:29 +00:00
package.json5 feat: support Viewpoints, translating PerspectiveCamera (#21) 2024-02-18 19:57:29 +00:00
pnpm-lock.yaml feat: support DEF and USE syntax to share subtrees (#18) 2024-02-14 02:55:39 +00:00
tsconfig.json feat: install moo and use in toy example (#2) 2023-09-01 21:32:25 +00:00

README.md

vrml1to97

JavaScript converter from VRML 1.0 to VRML97 file format.

This converter was originally a JavaScript reimplementation of the algorithm of the "token rearranger" found in the Wings 3D x3d importer (which was written in Erlang).

However, from version 0.3, it adds support for a significantly larger subset of VRML 1.0. Overall, it recognizes and converts

  • Title, SceneInfo, BackgroundColor, and View "Info" nodes [this node type was removed in VRML97].
  • All Light nodes
  • PerspectiveCamera nodes (converted into Viewpoint nodes)
  • Grouping nodes including Separator, Group, Switch, and WWWAnchor
  • Interprets a Switch named "Cameras" (by a DEF) as a list of Viewpoints
  • ShapeHints
  • Transformation nodes including Translation, Rotation, Scale, and Transform
  • All shape nodes including Cube, Cone, Cylinder, Sphere, IndexedFaceSet, IndexedLineSet, PointSet, Coordinate3, and Normal
  • All material nodes including Material, TextureCoordinate2, and Texture2
  • DEF and USE constructs to share subtrees

Usage

From an es6 module under Node (for example)

import {convert} from 'vrml1to97'
const vrml1spec = '# VRML 1.0 ....'
const vrml97spec = convert(vrml1spec)

or from a script in a webpage

(async () => {
   const vrml1to97 = await import('./dist/vrml1to97/index.js')
   const vrml1spec = '# VRML 1.0 ....'
   const vrml97spec = vrml1to97.convert(vrml1spec)
})()

or from the command line via node

npx vrml1to97 < old.wrl > shinynew.wrl

API

Currently this package exports just two functions:

  • convert(vrml1: string, source?: string): string

    The main function, takes in VRML 1 syntax (note that it does not actually check that its input is VRML 1, so its behavior is undefined if given anything but valid VRML 1 syntax). Returns VRML 97 syntax for the same scene, as nearly as it can translate. Note that not all of VRML 1 is recognized (see above for a list of constructs that should be handled), and some of the translations may be somewhat approximate.

    If the optional second argument source is supplied, convert adds a comment indicating that the original vrml1 came from the specified source, before conversion to VRML97.

  • tree97(vrml1: string): Tree

    Takes the same input as convert, but rather than returning a string, it returns a very rough syntax tree for the converted VRML97 scene. The returned Tree data structure is an object whose keys are property names and whose values are lists of either strings or sub-Trees. Note however that this syntax tree does not represent a complete parse; since VRML 1 and VRML97 are close, whole blocks of syntax are left as strings rather than broken into node names and property values. If there is a need to generate XML syntax output, for example, the code would need to be modified to break the tree down further to be ready for conversion to XML.

Conversion notes

One sort of geometry common to VRML 1 and VRML97 is the IndexedFaceSet. These entities are often used to render solids. Indeed, the default in VRML97 is to assume they do represent a solid, with the normals pointing outward. Unusual visual effects ensue if the normals are not properly directed (basically, you see through the "front" of the solid and see the "backs" of the faces on the opposite side).

As a result, unless the input VRML 1 explicitly includes an explicit vertexOrdering of CLOCKWISE or COUNTERCLOCKWISE, the default solid treatment will be turned off in the VRML97 output, meaning that all faces will be rendered opaque in both directions.