Solid working area

Duncan,

That's not how the solids working area is handled, it's a scale, not a volume. No single solid can have extents larger than the solid extent, but it can be positioned anywhere you want in the dgn*.

* Keeping in mind that with double precision floating point you have the most accuracy near 0.

-B

Thank Brian. I'm not sure I was clear. The mdl didn't create any geometry. it just indicated the area which I understand to be the swa within which the accuracy was high. I want to encourage our users to draw near 0 and remap coordinates if required so that when we export geometry to other apps ie Rhino and Revit we can bring it in without issue.

Duncan Macdonald said:

it just indicated the area which I understand to be the swa within which the accuracy was high.

Modeling close to 0 is always a good idea, and wanting to do so in order to facilitate export to another format is a reasonable requirement, so while I understand why you liked this app, it was erroneous that this app would lead someone to believe a solid that is 2 km from the origin when the swa is 1 km is invalid for the dgn format.

The solids extent is implemented as a scale from uors to solid kernel coordinates, operations on the solid are performed in the solid's local coordinate system, not design file coordinates. I don't know the history of this app, but the reason such a feature doesn't exist in MicroStation is that it would re-enforce the incorrect notion that the swa is a volume centered at 0.

-B

In our case if we are using Ordnance Survey coordinates and modelling a building in London, then we are more than 500 kilometres from the origin. Clearly .dgn can do this but I know from past experience that doing so causes problems with rendering and accuracy as well as translation to other apps. How does it work with 2d geometry? My experience is that lines drawn that far from 0 have problems with snapping and accurate location of snaps. 

Duncan Macdonald said:

My experience is that lines drawn that far from 0 have problems with snapping and accurate location of snaps. 

Yes, there can be double precision floating point issues when you are millions of uors away from 0,0,0 that are unrelated to the swa, the swa only applies to breps (smart solids, etc.).

The recommended approach is to model near 0,0,0 (uors) and specify a global origin to define the geometry's location in the world. But again, this is just good practice for how numbers are represented, it's not because of the solid extent setting.

What you don't want is people changing the swa to 1000 km and thinking, now everything is inside the swa and I'm good to go...because again, the swa is just a scale. It's basically just determining how far apart in uors 2 points need to be to be considered unique, etc.

-B