[Plugin] Solid Inspector
-
I don't know if something ain't right or what....
I really like the improved SI over what I had been using in SU 8 for years.
I'm now on SU-15.
But I just noticed that the found errors are not highlighted with big red/yellow circles. Up until yesterday, that was not a problem, as SI fixed 100% of my errors.
But then there was a new error (can't remember exactly what) that only offered info and no fix. The problem was the line was so tiny, even in red, that it took me about an hour to find it in an object with over 4,000 lines. Only by zooming in and going over the whole thing twice (2 separate but identical errors) did I find the tiny red lines, deleting them, and all was OK.
No red circle to guide me to the errors. Why?
-
@jgb said:
No red circle to guide me to the errors. Why?
Did you not see a label in the viewport pointing to the edge?
(got sample model or screenshot?) -
I do not recall seeing anything other than the warnings box.
I will try to set up a "fail" and see if I get that same error message, or see a viewport.
Stay tuned......
10 min later ====================
Well I did not get the same error msg with this test, but I got an error msg without a circle or a "viewport" pointing me to where the error was, in this case an external face.
It is at the top center of the comp.Unless I zoomed in, I would not see where it was, especially if it were much smaller.
The error is hard to find without a red circle.
-
I have an issue filed for adding back functionality to circle highlight the errors.
-
That would be great.
Thank you.
-
Another problem.....
Inspector says "all Shiny" but SU-15 says not solid.
Why??
-
I could see the problem immediately upon opening the file. There are two overlapping faces. Delete one of them and it's solid. Why SI2 doesn't detect it is a good question, though.
-
Thank you. That is in the very same small area that I had a lot of SI2 found errors (most it could not fix) that took a while for me to find (no circles!!) and manually fix.
How did you find it so fast???
-
It showed up white when I opened the SKP file and appeared to flicker as I orbited indicating z-fighting. It wasn't difficult at all to locate.
-
Use Tab to cycle through.
-
SketchUp perform a very naive manifold test - if all edges have exactly 2 faces connected.
However - there are edge cases where this fails, like with this model:
Solid Inspector use a different method to detect solids. However, in this case with the overlapping face (which shouldn't happen) it fails to detect this anomaly.
-
Now that I knew where to look, I copied the comp to 1 side and saw the white face.
Gone in 5 seconds and fixed! Thanks again Dave. In its original location those gear teeth in that area were visually covered by another gear (comp) so I could not see it. It was also the same area that SI2 found numerous non-fixable errors, and in my manual search and cleanup, missed that one. I never looked at the isolated comp I sent to this thread.
ThomThom: Is this sort of error findable or fixable in SI2?
I don't recall having this sort of error undetected in earlier SU-8 versions of SI. -
I ran it in CleanUp - which actually does look for these kind of things (but off by default because it's a very slow thing to do) but even that seemed to fail to detect it.
Not sure what's going on with that model... -
Another "shiny" but flawed non-solid. Multiple errors not detected; open faces, multiple faces (the bottom has 3 faces).
I was adding corner cubes and did an interim solid check
There were other errors it fixed, but somehow not these.
-
hm.. yea - another model where some faces have overlapping faces and invalid edge loops. Skrews up the algorithm. Have to look into catching that. Do a validity check before the manifold check.
-
I found it impossible to fix. The bottom face was 3 faces bounded by a single line rectangle. I could not delete a face, it kept coming back if I tried to fix something on an adjoining face. Then when I thought it was almost OK, I did another SI which found many problems. So I said let SI fix it. Ended up with about 8 valid lines and faces. SI "fixed" it by blowing most of it away.
I blew the rest away, started fresh, and that turned out OK.So, anybody, I don't need it fixed. I posted it for ThomThom to ferret out the why.
-
@jgb said:
I could not delete a face, it kept coming back if I tried to fix something on an adjoining face.
Ackh! One of them models - where faces start appearing all over the place. I've never found a good solution for that. I saw a good number of them in my previous job - usually ended up recreating the geometry because it was quicker. Never really figured out the cause - but I suspect it was often due to some precision issues which can often creep in from imported geometry.
-
Why this must be made by Hand ? (Close the solid) it detects 2 possibilities?
-
Because while it's easy for us humans to see the correct solution, it's much harder to code a solution. One can code a solution for this specific situation - but that wouldn't apply to other situations. Generic solution for repairing geometry is difficult.
-
The way my SolidSolver deals with that particular case is that it finds faces for the unfaced and single-faced edges on that corner, it then removes the newly added inner triangular partition-face, whilst leaving the three new triangular facets on the outer shell.
At that point it's a solid.
It finally prompts you to remove the three coplanar edges.
If you want the cube actually to have an angled triangular corner facet you need to edit it and delete the unfaced edges before attempting to make it a solid - then the one triangular corner facet is added if it's still missing...
Advertisement