Coordinate Systems

[IAN~JAMES]

We have started using FORMA to get our projects off to a start, however the coordinate systems seem to be a bit unpredictable and you can only see this by exporting the OBJ file which comes with a Readme.txt file that contains the reference to the CRS (Coordinate Reference System).

Working in Australia there are a  number of grid systems across the continent called MGA (Map Grid Australia) and for some time FORMA has exported and loaded into Revit using, for example, MGA56 (EPSG:7856) for East Coast (Sydney and Brisbane). However a couple of times recently on a couple of projects it has just switched to WGS84/UTM Zone 56s which isn't the same thing.

The result is that rather than the coordinates being alignable, the same numbers appear to be circa 8.219km East and 0.387km North of where we are expecting it to be which then only leaves manual alignment as an option.

What would be ideal if you have plans to resolve this would be the ability to select from valid CRS's that apply to the area in question allowing the user to select. Not wanting to over complicate this but in Australia a recent update was issued to teh GDA Standard too so each MGA Zone has the old GDA94 standard and the new GDA2020 standard which each have their own EPSG codes the delta between these ranges across the coutry from circa 1.5m to 1.8m due to teh fact australia is drifting and we often have to align with data from the old GDA standard too.

I believe that the maths to translate between these CRS's is well established and available in code libraries, however it is hard for us as end users to implement this at our end, so it makes sense that this is built in at the point where the data is exported.

It would be amazing if this could be a feature, because when it works, it is amazing, but when it doesn't it presents us with a lot of really useful information that we can't use without risk!

I'd be happy to hear your thought...

Thanks in advance and Best

Ian

[IAN~JAMES]

I did some further testing today, because I was suddenly concerned that I have been working on a number of projects this year assuming that FORMA was somehow just assuming MGA56 (EPSG:7856) because we setup the MGA Grids, specify the coordinates as the first step and bring everything else in to that location and until yesterday everything seems to drop in perfectly.

I've exported our MGA Grids to DXF (Shared Coordinates and in meters) which allows me to bring them in to QGIS and specify the CRS. I've brought them in to QGIS at both EPSG:7856 and also EPSG:32756 and they seem to be millimeter perfect, so my assumption is that either EPSG:7856 is a subset of EPSG:32756, either that or they just conform to the same coordinate system...? By that I mean that a real world point in the CRS has the same numeric Easting and Northing value irrespective of which CRS is used...? ¯\_(ツ)_/¯

The grids aren't super visible in the attached view, but they are in there and they do align perfectly...

The residual problem I have is that the FORMA import seems to be coming in over 8km to the east of where the site is setup, or at least that is where the coordinates of the Project Base Point (PBP) put it.

I've relocated the Survey point to the location where FORMA should be aligning. The coordinate specified at the PBP during the FORMA Load Proposal is as follows, which mathematically represents the above delta

FORMA Origin

 Currently I have had to just manually relocate the FORMA model (link) within our site model to align with some fairly dubious old CAD information just so the team can hopefully hit their deadline, but I need to figure out exactly what is going wrong here and it is a bit of a head scratcher...

just wondering if anyone has had any similar experiences or can spot from my description what may be going wrong...?

If you've read this far then HUGE THANKS ❤️

Ian

[kristoffer.dyrkorn · Accepted solution]

Hi,

I am a software engineer on the Forma team.

Forma uses UTM zones as the basis for the internal coordinate system for projects. Which UTM zone this will be, will be chosen based on which zone that will enclose the project's location on the globe.

Then, the coordinate values Forma uses internally are relative to some reference point within - or very close to - the site. The reference point is specified in coordinates using that same UTM zone.

All exports produce data in the internal coordinate system. So, the coordinates you will get in an export are in the project's UTM zone, but relative to (offset from) the reference point. The Readme.txt file describes which UTM zone this is, and the location of the reference point. So - by taking the exported coordinate values, and adding the offset values in the Readme.txt, you will get the true coordinates for the geometries, in the specified UTM zone.

I think what has happened here, like you already mention, is that the national grid and one of the UTM zones are 1:1 compatible in some cases (inside a limited region). So things work inside that region, but outside, the differences will suddenly become large as the coordinate systems are no longer compatible. And this is what causes the unexpected offsets.

The export mechanism for Forma has not changed for a long time. I would recommend assuming the coordinate values are in the UTM zone, and with the coordinate value offsets, as specified in the Readme file.

Best regards,

Kristoffer

Software Engineer,

Autodesk Forma

[IAN~JAMES]

Thanks for the reply @kristoffer.dyrkorn that all makes sense now...
We had found over many projects that the output (or Load Proposal) from Forma was compatible with models that we had previously setup to MGA coordinates, not realizing that this was a happy accident and that MGA56 is actually the same (ish) as WGS84/UTM Zone 56s. I did notice recently when trying to setup a couple of projects in South East Asia that we couldn't get the same alignment in Singapore and KL, and ended up aligning those as best we could manually, but your explanation explains why now and I understand that.

This problem last week seems to have been more of a glitch somewhere I think... the coordinates came out just 8219m out of location, which while being a lot, is close enough that it could have been mistaken as correct... I re-did the export again on Friday and weirdly it came out in the location I expected it to, so maybe it was something that just went awry in the proces...? ¯\_(ツ)_/¯

I had already given it a second go earlier in the week, but on the same day as the orignal one, just to be sure, but got the same erroneous output, but the Friday version seems to have addressed our concerns, and I've learnt a little more about teh Aussie coordinate systems.

I'll mark your explanation as answering the issue, and bear that new knowledge in mind for the future. It would be good if the Load Proposal yielded some confirmation of the CRS, maybe into the REVIT_LOCAL folder along side the topo texture...? We do occasionally import into Rhino as well, but not always. Or maybe make it a separate option on the drop down to just download the README file...??? just a thought...

Thanks again 

[valentinafantini]

 Hi I am opening the obj in Rhino however the model is rotated 90 degrees and I don't understand where the point mention in the "Readme" file is in the site. I have also checked in Google map and that coordinate goes elsewhere. I am a bit confused

[kristoffer.dyrkorn]

Hi,

The OBJ format does not specify which way (which axis) is up. In the export from Forma, the Y axis is up - and most likely Rhino follows a different convention. Regarding the point coordinates in the README file, the values are in UTM coordinates (which UTM zone this will be will depend on the location of the site). Those coordinates values are not compatible with the ones Google Map is using - you would need to convert the UTM coordinates to lat-lon (WGS 84) values before Google will be able to show where that point is.

The reason for having the point's coordinates in the README is that all coordinate values in the OBJ file will be relative to this location. So without knowing about the location it will not be possible to place the OBJ correctly at some location on a map or in a 3D model.

Best regards,

Kristoffer

Software Engineer,

Autodesk Forma