Weird triangulation of faces
-
john2
Can you post your model.
When one is trying to clean a model there usually are tolerances associated with that operation which you can change and the results can be a function of the sequence of steps you follow. For example if I merge "close" vertices there is a tolerance with what close is and I can actually change my geo. with that operation. There use to be associated with the suddivide and smooth a video that shows that possibel artifact along model boundary. Like Thomthom notes under stand what changes are being made. -
I should probably add a little clarification on this:
In that first step when I say to select a single plane, what I mean is to select all coordinating faces and edges of what was originally a single plane before the artifact lines showed up. -
Here's a solution that often works to allow you to erase the artifact lines without having to rebuild the geometry:
Select all faces and edges of a single plane where the artifact lines are showing up.
(I should probably add a little clarification on this:
In this first step, what I mean is to select all coordinating faces and edges, including artifact lines, of what was originally a single plane before the artifact lines showed up.)
Go into wire frame mode and activate the scale tool, then zoom in as close as you can to the center handles. They should appear as if you have selected a multi-planar object and not merely a single plane, so in other words the scale tool should be showing a cube and not a plane, with two center handles instead of the single one we would expect to see.
Now scale the "top" center handle into the "bottom" center handle so that both handles will merge into a single handle, which now indicates that Sketchup recognizes the selection as a single plane.
You should now be able to delete the artifact lines. Just be sure you zoomed in far enough that when you scale the handles together that they really do merge and aren't just really, really close. -
In effect you are flattening the surface that went very slightly off during the intersect. And that is a very good suggestion.
But that will only work on surfaces that were (before the intersect) exactly planar to any 2 model axis, that is; the surface is aligned in 2 axis directions. If the surface is at any angle, flattening using the SU scale tool will clobber the surface, as Scale works strictly along axis directions.
For non-axis planar surfaces use Fredo's Scale tools, where you can define the scaling box's orientation.
-
@jgb said:
In effect you are flattening the surface that went very slightly off during the intersect. And that is a very good suggestion.
But that will only work on surfaces that were (before the intersect) exactly planar to any 2 model axis, that is; the surface is aligned in 2 axis directions. If the surface is at any angle, flattening using the SU scale tool will clobber the surface, as Scale works strictly along axis directions.
For non-axis planar surfaces use Fredo's Scale tools, where you can define the scaling box's orientation.
Right.
I haven't used Fredo's scale tools, and I'm sure using these would be the fastest way to do this, but I would think you could use the AXES tool and set it to a perimeter edge and face, which should align with the rest of the model (I think..) then group the surface, do the scale fix, then un-group and re-attach the surface... probably the uber slow way of going about this compared to using Fredo's scale tools. -
I played a bit with using the scale tool (not Fredo's) to correct a distorted surface.
The reason this will not work as expected is the lack of precision inherent in the scale tool. Had no time to try Fredo's scale tool, but I suspect it suffers the same as well.
If you scale anything, you will notice that the % scale is good only to +-X.01 (reported in the VCB) but you can scale to finer precision by manually entering a value. Here you want to the 2 center handles to merge. But if the distance to move is less than that .01 tolerance, then the selected handle will snap to just in front or just in back of the target handle. You can't nail it. Yes, you can interpolate the distance and enter it in the VCB, but how exact is your estimate? You would have to be within SU's tolerance limit to flatten the face, and that tolerance is way less than the scales tolerance.
Whenever I get a flat surface mangled by an intersect, and that happens a lot (to me), and seems to happen more-so in big models where I can suddenly get literally dozens of hidden lines appearing out of nowhere, I use TT's Selection tools to select only the hidden lines and delete them all. Invariably that blows away the surface. Either I reconstruct the surface or undo to just prior the intersect, and manually intersect the entities.
It seems to me that when you move a line on a surface, or intersect a surface, SU will draw temporary hidden lines connecting all the potential verticies on that surface, to form the "bend" lines. But the cleanup routine to remove the unneeded lines seems to quit before the job is done, like I said more-so in big models.
Part of the problem with intersecting is again related to SU's internal tolerances. If the line fragments generated by the intersect is near or less than the tolerance, SU may not even draw the line, resulting in an open piece of the surfaces, or it may snap the endpoints of the fragment at the tolerance point, which will throw off the "flatness" of the surface resulting in stray hidden or soft or even hard lines.
The only solution here is for Google to rationalize SU's internal tolerances and make them more consistent, and at the same time loosen up the tolerance (a bit) for a surface to remain "flat". SU also needs to be a lot more adept at handling small line fragments. People are using SU more-so as a poor mans engineering 3D CAD (especially me) therefore needing finer tolerances, than as its' original intent as an architectural sketching tool that is OK with its current larger tolerances.
-
I still have simple question.
Why do you have the need to run the clean up multiple times? I think everyone has given a rational explanation of the "auto fold" lines so if your model does not require cleanup don't run it.Cleanups can be damaging so only run as really needed and then some time you have to plan some repairs. You can usually get away with deletion of null faces, duplicate vertices, unreferenced vertices and duplicate faces but further that that ( say like merge close vertices, deletion of self intersecting faces, non manifold edges etc)) you can expect some repairs. The work flow should be:" If it ain't broke don't fix it"
Just some thoughts for you. -
Again, you are right jgb, but you can get it to work, but it must be done exactly correctly.
To make this work, it usually requires that you zoom and move the handles in more than a single step (zoom, move, zoom, move...), and it will not work correctly unless the last handle movement (before they disappear) is pushed right up to the threshold of the tolerance level. This all depends on whether you were zoomed in close enough, AND that the last handle movement was pushed as far as it would go.
-Just because the the handles disappear doesn't necessarily mean that the planes have all merged to the same coordinate.
One thing I've noticed when testing this is that the further the distance between the planes you are trying to "level" out, the more exacting this process must be in order for ALL the lines can be erased without faces being deleted. You can see this when trying to level out three planes where one plane is substantially more distanced from the other two; the line that divided the closer planes will likely be able to be erased without deleting a face, but the other line cannot be erased without deleting a face(s) in the process.
-This is why this process usually works when dealing with the artifact lines problem. The various planes are so close to being on the same planar coordinate that when the handles merge and disappear, you can almost always delete the artifact lines without problems, because all the planes merged according to the necessary tolerance threshold.
I hope this is all clear, if not, I might be able to make a video that shows how to make this process work effectively for anyone that might need it.
-
Now that I've been messing with this for a couple of days, I've noticed that sometimes it works and sometimes it doesn't. Also, at what point the center scale handles disappear is dependent on how far out you are zoomed (the closer you are zoomed in the less chance the handles will disappear prematurely).
Here's a video that shows me getting rid of some artifact lines. It's probably not all that useful, but it shows an example, and as you will see in this instance or only worked partially. -
I watched the vid. Interesting. While I've stated above why using the scale tool is not ideal, a tip on doing it that way is....
When you need to get very close to the subject (center handles) before you zoom in, turn OFF perspective. That will allow you the get very close without nearfield clipping.
After that doesn't work if you examine the model and the affected faces you will notice most of the hidden lines emanate from midway along an edge and some from corners. The Hlines from corners are what you would expect when an intersect happens, as SU draws these potential fold lines before the actual intersect is actioned. SU is supposed to clean up these lines after the intersect, except where the face becomes non-planar. SU does not do a good job of that cleanup.
But you can rebuild the face with a bit of sleuthing and some line regeneration. Using the erase tool, start deleting Hlines in small bunches. When a bunch defaces, undo and delete the Hlines 1 by 1, leaving the very few lines that deface intact. You will get most of them that way, and if there are any left you can just leave it alone, but we can't do that, now, can we?
You will notice the bulk of the remaining Hlines almost always emanate from somewhere along a line at 1 or both ends. Very rare in my experience they emanate between 2 corners. That means the line has been segmented for some reason, and 1 or more new endpoints have shifted out of "flatness" tolerance from SU's point of view. Delete that line, and its minute segments. The 2 faces it formed will disappear.
IMPORTANT; do a small left2right select at each corner, to make sure no micro fragments remain, just delete if they appear in Entity Info. That may also delete some other faces, leaving 1 or more lonesome lines, attached at 1 end.
Replace the lines endpoint to endpoint, and faces should reform.
Yes, you can end up redrawing much of the perimeter of some really bad cases, but that will cleanup a face and reduce the line count as well.
-
remember awhile back I posted about "microvertices" and was there a way to clean them up?
this is what I see when I have that issue. tiny gaps between 2 vertices, as in micro-millimeters - smaller than SU likes to work with, that cause this triangulation.all I found to do was to erase the affected area, draw a line in on the tri line (even after erasing one end, it'll remain!) then erasing that line and the redrawing the areas I had to erase.
..I hope that explained clearly.. I don't think I explained the problem well last time..
-
These micro-gaps are pernicious and extremely difficult to find. You can have multiple triangles that seem to endpoint on 1 vertex, but in fact are separate and have their own vertex extremely close to each other. As you said, the only way to fix them is to delete all but 1 line and rejoin the triangles.
However, the cause of many of these micro-gaps are micro-fragments of lines joining those separated verticies. They are so small, SU cannot make a face using them as 1/3 of the triangle, and leaves the face open. Finding micro-fragments is easy. Just do a very small Left2Right Select on the vertex, then just erase them. Then fix the triangles/faces.
They are caused by many ways, intersects, most common, but not adhering to endpoints when drawing lines is next most common. Be very careful when joining endpoints that SU does not say "Constrained by ...." which means it wants to put the end of the line you want to draw coincident with an axis line, or snapping to an endpoint just near where you want to go. That for sure will generate gaps and fragments, and faces that simply will not form.
If you do get that dreaded "Constrained by ..." going from point A to B, then Escape and simply draw the line from point B to A (reverse). For some weird reason that always works.
-
Another problem that seems to be associated with those micro gaps is that when fixed, often it creates a dual face, especially so when dealing with very small areas.
I think one aspect that the Sketchup development team needs to recognize is that Sketchup is positioned to be the most widely used and recognized program associated with 3D printing, but this means that Sketchup has to be able to deal with very small geometry, as most of 3D printing currently deals with smaller more ornamental forms.
-
@xiombarg said:
Another problem that seems to be associated with those micro gaps is that when fixed, often it creates a dual face, especially so when dealing with very small areas.
I think one aspect that the Sketchup development team needs to recognize is that Sketchup is positioned to be the most widely used and recognized program associated with 3D printing, but this means that Sketchup has to be able to deal with very small geometry, as most of 3D printing currently deals with smaller more ornamental forms.
I think I know why SU will draw a 2nd or even 3rd face between 3 planar and endpoint joined lines. The endpoints of each line are actually not connected, but within the tolerance to form a face. SU draws a face for each line. Other than using Solid Inspector, or seeing the subtle difference in a smooth face, there's no way to find them.
You are dead right about SU becoming a standard for 3D printing. A few months ago I attended a seminar for SolidWorks CAD and they had a "gee-whiz" 3D printer there that could lay down a solid model in multiple materials from soft rubber to aluminum in 1 go using inkjet technology. One simple demo part was created in SU and the drawing ported to the printer using an in-development s/w conversion. The presenter aid SU was becoming the poor-mans 3D printer app.
-
I can honestly say that thomthoms solid inspector plugin turned SU from a program that I used to use sometimes into a program that I now use almost daily.
Advertisement