glTF : capacity, limitations, performance?

[image_95]

Dear all,

As part of a state of art about Textured Digital Surface Model (3D geospatial modeling) formats, we have identified some candidates formats whose glTF. We would like general informations focused on :

• The capacities
• Advantages/Limitations
• Performance

I’m taking the liberty to ask following 10 concise questions on glTF.

1. Type of geometry supported (Mesh, TIN, triangles, other) ?
2. Type of “Spatial Reference System” supported (geographic : Reference system+latitude/longitude/ellipsoid heigh) and/or projected : Reference system+ellipsoid heigh+ projection) and Spatial Reference System description : Spatial Reference System description Identifier(SRID) ?
3. Capacity to support the semantic ?
4. Ability to associate TIN/Mesh (3D model) with Points clouds ? Ability to associate several N-property with a Mesh /TIN ? (Point clouds + Texture,etc.)
5. Capacity to support metadata (for instance qualifying the product on planimetric/altimetric accuracies,etc.) and glTF 's specifications ability to provide clarification about metadata mecanism ?
6. Capacity/optimisation of data compression ?
7. Tiling management capacity ?
8. Multi-resolution management (2 levels or more) ? One more general and another one with a better level of detail ?
9. What about Performance and Limitations ? (Data access, storage, Data Volume, display, adaptibility to client/server diffusion, bandwith optimization)
10. Standard maturity (level of interoperability, level of GIS tools implementation : especially qgis, esri, global mapper)

In advance, thank you very much for your usefull lighting.

Kind regards.

[donmccurdy]

I’ll chip away at a few of these. I can’t say much about the GIS-related questions.

Type of geometry supported (Mesh, TIN, triangles, other) ?

Points, lines, and polygons / triangle meshes. Various triangle draw modes (strip, fan) are also supported.

Type of “ Spatial Reference System ” supported
Tiling management capacity ?
Multi-resolution management (2 levels or more) ?

This is not a GIS-specific format, and does not include these features. The OGC 3D Tiles format can reference glTF files, so that would be a good way to handle this.

Capacity to support the semantic ?
Ability to associate TIN/Mesh (3D model) with Points clouds ?

I’m not sure what these questions mean, sorry.

Capacity to support metadata

glTF supports metadata, which are simply key/value pairs defined in an extras field. Meaning of those properties is entirely up to the application. For something with a formal structure we encourage you to define a vendor extension.

Capacity/optimisation of data compression ?

glTF allows the use of Draco compression for triangle mesh geometry, which is state-of-the-art for that purpose. Support for compression of other things like animation and point clouds may be considered in the future.

What about Performance and Limitations ?

I’m not sure how to answer this without more specific questions. glTF is designed to be transmitted over a network and rendered in realtime GPU-based engines efficiently, that is its purpose. It does not generally include features that aren’t common in current realtime rendering technology, like certain types of animation. For GIS-specific constraints, like streaming data access, I would recommend OGC 3D Tiles for the best of both worlds here.

Standard maturity (level of interoperability , level of GIS tools implementation

I think you’ll need to do this research sorry. glTF is widely supported in the realtime 3D graphics world now, but support in GIS tools specifically is not something I’m familiar with. I do know that there’s a qgis plugin and the Cesium team supports glTF and has been heavily involved in its development.