[MStn CE U17.1] WMTS dysfunctional

Hi,

this is my contribution to (sporadic) discussions (and complains) about WMTS support quality in MicroStation and other products.

Recently, I was asked by my customer for help with WMS and WMTS in MicroStation CONNECT Edition and how such data can be printed. Because we found a lot of issues during WMTS testing, and the final decision is that there are so many disfunctions and bugs that WMTS in MicroStation cannot be used in production (especially when compared with other products), I put together a few examples, using public data sources, so anybody can repeat the testing (and it’s used as test case for Bentley support as well ;-).

A basic requirement is simple: To use WMTS in MicroStation instead of WMS, because of speed and also, because some data providers migrate their servers from WMS to WMTS.

We use typically one coordinate systems only and the systems is “just XY coordinates”, so the task should be simple: To place raster to the right position in model (it is how WMS works with no big problems). The system is S-JTSK and usually EPSG 5514 is recommended to use (in the past it was 102067, now treated as obsolete).

A summary of tests:

No EPSG set in model (no-EPSG.dgn):

EPSG xwmts file result
5514 CUZK_ZM_epsg_5514_(jtsk) map displayed, but wrong position (bug)
5514 CUZK_ZM_epsg_5514_(jtsk) moved manually to the right position nothing is displayed (bug)
4258 CUZK_ZM_epsg_4258_(etrs89_geographic2d) cannot be used without transformation
32633 CUZK_ZM_epsg_32633_(wgs84_utm33n) cannot be used without transformation

EPSG 5514 set in model (EPSG-5514.dgn):

EPSG xwmts file result
5514 CUZK_ZM_epsg_5514_(jtsk) MicroStation crashes (bug)
4258 CUZK_ZM_epsg_4258_(etrs89_geographic2d) probably correct transformation, but nothing is displayed (bug)
32633 CUZK_ZM_epsg_32633_(wgs84_utm33n) correct transformation, displayed in the right position in JTSK coordinates (lower precision than native 5514), tiles loading is not stable (result is not good, but the only usable working combination)

I also tried how the files mentioned above (no-EPSG.dgn and EPSG-5514.dgn) work when used as references. Because only map in EPSG 32633 with applied transformation to EPSG 5514 works (displayed in the right position), I tested only this combination.

  • Master file with no EPSG assigned (master-no-EPSG.dgn): Nothing displayed (but borders looks in the right position).
  • Master file with EPSG 5514 assigned (master-EPSG-5514.dgn): Map displayed (but loading of tiles again works in a strange way, not refreshed correctly, when attached to some other DGN files, not displayed at all).

The results are not good:

  • When everything is done by a user right (EPSG assigned to model, the right map selected), MicroStation crashes.
  • When EPSG is not assigned to model, map is displayed in wrong place.
  • The only way how to display the map in the right position is to select another coordinate system and apply transformation. But even then, with some files, it does not work (no WMTS data is displayed).
  • Such file cannot be used as reference without EPSG assigned in master model (in reality, there is no system assigned, because it’s XY for users and they have no reason to take care about EPSG).

Other products, like QGIS (tested myself) or Autodesk Civil 3D (told me by my customer ;-) work fine without any problem (I am not sure about AutoCAD or ArcGIS, but I do not hear any complains).

The test data is attached.

To log the found issues officially, I plan to create service ticket with link to this post also.

With regards,

  Jan

test_data.zip

  • everything related to WMS/WMTS is not production ready and only somewhat barely works. 

    I agree for some unknown reason, WMS and WMTS implementations in PowerPlatform are like school beginners project :-( Which is sad in situation, when not supporting INSPIRE data (WMS, WMTS, WFS) perfectly means you are quickly out from business. Many users have pointed to this fact for last many years, and the result is that any software I know (commercial competitors, open source, local tools...) is much much much better, working in the way "enter server URL, click, and it works".

    Regards,

      Jan

  • (Microstation CE update 17)

    I agree that the implementation of WMTS is not production ready.

    In my case, SDFI.dk, the National Danish provider of WMS, WFS and WMTS, has opened several WMTS services, but they all fails in Microstation. I can connect to them and the service does appear in the Raster Manager, but nothing shows in the map.

    But in the Message Center it says that "WMTS map tile matrixes are not organized as definition of well-known scale set GOOGLEMAPSCompatible". And this is true, they're not organized that way, because it is not recommended to do so anymore.

    And here's the twist, in the 'OGC Two Dimensional Tile Matrix Set and Tile Set Metadata' document (https://docs.ogc.org/is/17-083r4/17-083r4.html#toc46) (publication date 2022-09-09) it is stated in Annex C, that

    "[....] The WKSS concept was introduced in WMTS to improve interoperability, but experience has demonstrated that the use of common TileMatrixSets such as those registered on the OGC Naming Authority’s register, and defined in the common tile matrix sets and variable width tile matrix sets definitions annexes, is even better. The use of WKSS is no longer encouraged by this standard." 

    Summa summarum: The implementation is outdated/error prone since it relies on WKSS and should be fixed as soon as possible in order to facilitate the proper use of OCG services.

    Note to Bentley: A year ago, I had much trouble with the WMS and WFS implementations at customers  (https://communities.bentley.com/products/microstation/f/microstation-forum/157595/use-of-wfs-web-feature-service) - and I don't think they're solved yet. No further inputs to that tread, which I think is quite sad, Bentley, as the OGC services looks like their here to stay (all my customers are using them every day). And when QGIS seems to be able to handle these services effortlessly, so why can't you? 

    Happy New Year, best regards Claus

  • CUZK_ZM_epsg.7z

    Hi Jan,

    These position and display issues were caused by a wrong GCS reprojection calculation when some map tile location are out of the target GCS valid range. In this condition, the tile location correction outputted from our base 3rd party lib does not be guaranteed. we have not gotten enough stuff to do a fully test that time for 17.1, and falsely thought this situation is not so common. In 17.2, a fix has been pushed, now it works(see the attached video). In order to make it work properly, you need to reattach the xwmts file and references in test_data.zip in 17.2 . Very sorry for the Bad experience.

    BTW, By now , we only offer a minimum wmts map support as KVP visit with well-known scale set GOOGLEMAPSCompatible.

    Best Regards
    Chris 

  • If iTwins are about openess and interoperability, referencing GIS type info is key.

    I hope Bentley will fix these long standing WMS, WMTS, WFS problems on the dgn side before focusing resources on Phase 2, where its not quite clear how referencing imodels are going to be more productive compared to referencing dgns and other formats in Mstn.

    Or is the intention to provide an iTwin Connector for these formats? The user would then ref attach an iModel from the Ref dialog.