SU 2015 Automatic Check & Repair

#1 (267.2 KB)

We are exporting from our professionnal kitchen software some cabinets with C++ SDK.
Since SU 2015 version when opening our models, SU propose to Fix geometry inconsistencies.
The Dialog box is weird for end-user with “Don’t Panic” title and non-obvious options to choose and understand.

The fix report says
CEdge (2693) and CEdge (2709) connect the same 2 vertices - fixed

We have edges using same vertices ?
It’s not the case from our point of view. It is just 3 simple faces (files enclose here) . We tried to find out why and we didn’t succeed. In our sample, “j.skp” has a message where “k.skp” doesn’t.
The difference between them ?
A simple translation of a face. In j.skp case the 3 faces have common edges to faces. In the other case (k.skp) the 3rd face is oustide from the two other faces

It seems there is an “analysis” on the geometry and a merge seems to be done for us on edges.
We would like to output our models with no error like these. Is there a way to achive this ?

Thanks for any suggestion.



Your models contain defects that are right at the approximately 1mm threshold of SketchUp’s internal tolerance for vertices to be distinct. In model j, the defect happens to trigger the distinct vertex trap, whereas in model k it does not, but this is a matter of fine details - both models contain the same kind of flaw. SketchUp’s merging of too-close vertices is only triggered under certain conditions, one of these being the integrity check as a model is opened. Many other operations do not trigger the merging of vertices. For example, if you scale up, draw something, and then scale down, vertices that are ok when enlarged but too close when scaled back down are not merged. There are also numerous methods in the programming API that do not trigger merging.

I have attached screenshots that show the flaws. In both models you have slight mismatches between the widths of the surfaces. In model j, the back is 2mm wider than the bottom and there is an extra 1mm wide rectangle adjacent to the bottom. In model k, the back is 1mm wider than the bottom but the extra rectangle isn’t present.

The best way to address these problems is to revisit how the models were created and figure out why those surfaces were drawn at such close, but not quite equal sizes.