[Requests] Review of 3d-tiles Community Standard Justification and related spec files

Carl Reed carl.n.reed at gmail.com
Thu Aug 11 17:16:00 EDT 2016


Dear Scott and 3D-Tiles submission team -

Thank you for submitting the 3D-Tiles specification as a candidate for
using the Community standard Process.

As you may know, when I was the OGC Executive Director Standard, I reviewed
and commented on numerous candidate standard submissions. Just ask the Open
GeoSMS or SensorThings SWGs how extensive my comments and edits can be!

So, I decided to take the same approach with the 3D-Tiles submission.
Attached are two documents:

1. The Justification document with edits and comments.
2. The 3D-Tiles specification as a Word file with edits and comments. This
file does not include the TileFormats clauses, although I read those also.

Please note that I had to removes some graphics to make this email small
enough for the OGC list server.

I also browsed the schemas, the glTF spec, and other supporting documents.
I appreciated the ability to review all of the supporting documentation,
presentations, and so forth.

However, I have to object to the submission as a potential candidate for
the Community Standards track for the following reasons:

1. Maturity. As clearly stated in the 3D-Tiles document "We expect the
initial 3D Tiles spec to evolve until fall 2016. If you are OK with things
changing, then yes, jump in. The Cesium implementation is in the 3d-tiles
<https://github.com/AnalyticalGraphicsInc/cesium/tree/3d-tiles> branch.
>From other documentation, the idea for 3D-Tiles first appeared in 2014 and
the spec announced in august 2015. This does not indicate a mature
specification. from my perspective, mature specs/standards are GeoTIFF,
GeoRSS, and ShapeFiles. Further, changes are being made on a regular basis.
Maturity suggests a stable baseline. Finally, a number of components are
shown as still in progress.

2. Evidence of Implementation: The 11 provided examples is not sufficient
as evidence of a strong and diverse implementation community. Bentley is
the only major international company on the list and there product that
includes 3D-Tiles has been announced but not released. Bentley also
supports i3s, another tiling system. To me, GeoJSON has strong evidence of
implementation with hundreds of open source and commercial implementations
(Google, Twitter, Esri, AutoDesk, and many more). This is the level of
evidence I believe is required  for a spec to be submitted as a community
standard. Check the comments in the attached files.

3. The spec document: The specification document needs work. There are
TODOs scattered throughout the various components as well as the word
"draft" in the schemas. A variety of informative (and marketing) content is
included in the specification. These should be moved to a separate
informative document. I comments notes in the attached spec file.

4. Dependencies: The core spec needs to clearly state what the 3D-Tiles
spec dependencies are on other specs/standards, such as glTF and WebGL. The
current version of 3D-Tiles seems to be quite tied to a Khronos type
workflow. If this is the case, what is the impact on a.) learning curve and
b.) the ability use use other visualization/rendering workflows? I have no
problem with glTF or WebGL but I do have concerns about overly tight
couplings that discourage other approaches.

5. Use cases: The OGC has not (yet) clearly identified and documented use
case for 3D-Tiles. There is the current work in Testbed 12 on vector tiles.
Trying to process 3D-Tiles in the absense of the larger requirements and
use case context may be pre-mature.

6. Requirements: While a community Standard submission does not need to
follow the OGC modular specification, requirements should be clearly
stated. This is not the case with the 3D-Tiles spec. Some work is required
so that mandatory elements (requirements) and optional elements are clearly
identified.

7. Coordinate Reference Systems (CRS): A section needs to be added to
clearly state the "whats" and "hows" of CRSs. Simply sating WGS 84 is
inadequate, especially since "long/lat" coordinate order appears.
Basically, the CRS aspects of the spec are lacking and as a result
confusing.

8. There are other 3D Tiling approaches. The OGC should have time to
consider other approaches to ensure that the broader requirements of the
entire community are best addressed.

9. All the examples and implementation references seem to require Cesium.
While the 3D-Tiles spec states that 3D-Tiles is an agnostic approach to
streaming geospatial content, this statement is followed by: We expect to
see other visualization engines and conversion tools use 3D Tiles. This
suggests that there are no other implementations other than using the
Cesium framework. I would like to see implementations independent of Cesium.

10. Vector data. This is in progress. Currently, the spec points to the
Cesium tutorial on geometry. Cesium geometry is graphics oriented although
there are some references to ellipsoid. I think an interesting exercise
would be to map Cesium geometry to ISO 19107/GML/Simple features so that
developers have a roadmap for streaming 19107 specified geometry into
3D-Tiles.

My recommendations would be to:

1. Do a new interoperability experiment with the 3D Portrayal Service SWG.
The focus of this IE would be to write an OGC Best Practice for the use of
3D-Tiles with the OGC 3DPS Interface standard.

2. Work on and mature the physical core specification document.

3. Allow more time for the community to implement and use 3D-Tiles -
increase the evidence of implementation.

4. Look at the provision of an implementation that does not require Cesium.

I am not trying to be overly harsh. However, the Community Standard track
and related PnP was designed for the submission and processing of mature,
robust, widely implemented and used specifications. 3D-Tiles is simply not
at that level of community adoption yet.

Regards

-- 
Carl Reed, PhD
Carl Reed and Associates

Mobile: 970-402-0284

"When the power of love overcomes the love of power the world will know
peace." Jimi Hendrix
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.opengeospatial.org/pipermail/requests/attachments/20160811/cbd7d77b/attachment-0001.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: 3d-tiles spec CNR Comments and edits.docx
Type: application/vnd.openxmlformats-officedocument.wordprocessingml.document
Size: 1811382 bytes
Desc: not available
URL: <http://lists.opengeospatial.org/pipermail/requests/attachments/20160811/cbd7d77b/attachment-0002.docx>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: 16-123_3D_Tiles_community_standard_justification CNR Comments.docx
Type: application/vnd.openxmlformats-officedocument.wordprocessingml.document
Size: 2758894 bytes
Desc: not available
URL: <http://lists.opengeospatial.org/pipermail/requests/attachments/20160811/cbd7d77b/attachment-0003.docx>


More information about the Requests mailing list