Help needed to flatten cracked face one thousandth of a foot out of plane

Flatten was the first plugin to flatten geometry to Z=0.

Fixing a bunch of wonky geometry certainly is a waste of time.
Albeit, the exercise will help one develop their geometry tweaking skills.
Generally, it’s faster/easier to erase and do it again and again if necessary, until one finds the right way.
It’s called practice.


One might also blame the hammer when they hit their thumb instead of the nail.

• The user has infinite control of the Drawing Axes
• Drawing Axes steer the Inference Engine
• Inferencing steers the tools
• Inference Locking puts the tools on rails

Experienced modelers tend to have very shiny Shift and keys.

3 Likes

It’s back again, this time on floor 93, which I copied up from your Floor 91 and then modified, apparently to the breakage point.
Some things I have learned NOT to do:

  1. Don’t use the arrow keys when pulling holes to a different location as this will almost always cause cracks in the floor. Just open for editing and surround the holes and move them carefully along red (in this case) or green.
  2. When moving beveled edges, move along an existing edge.
  3. Don’t attempt to move a hole over inside geometry. It’ll get stuck and/or crack the floor. Clean up the insides of the floor first.

Well, I did all this and still the underside cracked. I was able to remove all but 4 of the cracks without losing the underside face. It passes Solid Inspector with just 1 nested instance and passes SU’s Fix problems, but I’m afraid it’ll again fail at some point, possibly when I have to copy and modify this floor for the final time on 95.
F93.skp (605.4 KB)

The explanations and fixes suggested are not simple, at least to me, and I am wondering why I keep having these issues when I even started with a previously fixed floor (91)? Can this be fixed for good?

Why not start with a flat horizontal rectangle instead of using a face that may start as coplanar but not being horizontal. I doubt that SketchUp can’t keep vertices on exactly the same level.
On a slightly slanted plane this could very well be difficult for other programs as well.
image

1 Like

I could do that, but then I’d have to re-bevel the elevator holes in the middle (the elevator shafts go at an angle). But, I suppose that might be simpler than a lot of the fixes available here.

These aren’t the cause of your problems.

1 Like

Neither the top nor the bottom of @Box’s ‘fixed’ floor (91) are perfectly coplanar.
Here are the coordinates of the vertices.
F91 floor only - top face crackedBox_verts.txt (4.5 KB)

For ease of reading, I moved the geometry, placing a corner at the Origin.
Positioning the top face parallel with the Ground Plane proved impossible, because it’s not flat.

Any program needs to keep track of the Z value depending on where it is on the face.
If the vertex A is tied to a plane through the four corner vertices at any time, then it will be coplanar with these corners.
But SketchUp “only” sees A as an endpoint of two new edges, starting with coplanarity. You can (or maybe not) imagine that when going on adding edges (and vertices) somewhere on the rectangular face, at some point there may be slight differences introduces.
This will not be the case with a horizontal face.
Why do you want to work with the slanted face?

We don’t. But they somehow get that way, even if started dead horizontal. And there’s enough similarity between floors that it saves a lot of re-drawing if one can use a copy of an earlier one.

As long as the floor is flat but slanted you could use ‘Align Axes’ on top or bottom face and make the floor a new component.
Then placing a new instance of this component in modeling space with the reset axes it will be horizontal.

I’m not at all sure it IS flat - unless showing as a singe face would indicate that? Must try it, though.

I was wondering if an easier way to fix it would be to draw a large rectangle, bigger than the floor, and a little above it would work to get the flat, horizontal plane.

Use Top view parallel projection.

Make the new face moderately transparent or turn on Xray mode, then using On Face locked inference, draw over the floor outline but on the new face.

I’ll try both of these in a few minutes.

This could result in other dimension issues depending on the floor angle. I’m not sure it is an ideal method.

What causes the floor to become not horizontal in the first place?

1 Like

Maybe, but the off-level is so small, any differences are completely negligible.

Here’s my redrawing, using the method I described. A few lines missed the plane initially and even when I got them all drawn and on-plane, the top face didn’t form until I drew a line across the middle (redrawing an edge didn’t work).

After that a pushpull down 2ft, then moving edges in the bottom face ‘on red’ locked axis inference, to line up with the bevelled edges in the original. And finally, a move down to the original floor level.
F93 redrawn flat.skp (638.1 KB)

PS. I tried using the Text Tool to label the coordinates of the corners. But often, even though I had a clear inference on an endpoint at the corner, the tool only placed Enter text in the label.

I checked with the Query Tool, and all the corners seem to be dead on a whole-foot height.

PPS. I wasn’t timing it, but looking at the difference in time between my last two posts, it took about half an hour to redraw.

I see I made two mistakes in the version posted here. I missed an outdent and a glass railing on one edge. But the principle worked.

@john_mcclenahan

That which you dismiss as ‘negligible’ is precisely the stuff modeling errors are made of.
The result; tiny lonely edges just waiting to cause more issues.


Screenshots from F93 redrawn flat.skp




Here’s a screenshot of F93 redrawn flat.skp as the model first opened.
Little wonder why accuracy and performance are suffering.
• Length Units settings like these invite errors and make errors far more difficult to detect.
• Profiles slow performance and make it difficult to see geometry.
• Textures slow performance and make it difficult to see geometry.

1 Like

Two months ago we discussed performance issues at length here in the forum and privately.

You were shown how light and responsive the model became when all the dross from 3DWH was deleted.
You were shown how performance improved even further after exploding the cacophony of over-nested Groups and Components that comprised the building structure itself.

Nonetheless, here we find this overwrought 3DWH component in the most recent model F93 redrawn flat.skp.

The greatest impediment to completion of this model is not the hardware you’ve thrown at it or the software.
It’s ignoring sound advice from the many experienced modelers here trying to help you.

2 Likes

John and I are discussing having another go at simplifying some of the worst 3DW offenders. I’ve held off while getting to the point where I won’t need many more of these, near the end of the interior design of the building. The glass panels can certainly be one of them. In my defense, sort of, there are simply too many complex needs in the model for me to create with my limited time and skillset, and I don’t always want to just use simple walls or glass panels, for example, and complex things like kitchens and bathrooms are all the work of others. John simplified the bathrooms some time ago, but now I have a new one for the luxury apts. at the top of the building, so that’s another area to attack. Lots to do and there’s only two of us for a 430mb model …and counting.

In Scott’s defence, I would just like to say that this is his very first SketchUp model. I think it’s amazing that he has got so far, so fast.

Please don’t feel we are ignoring the experts’ advice, but are working slowly on cleaning up the modelling, and simplifying or redrawing the over-complex 3d Warehouse models, and overly deep nesting of components.

Scott is still, however, creating faster than I can keep up with cleaning up. And he’s getting better at drawing more cleanly, though still not perfect - sometimes far from it. Hence the kind of problems we are still encountering or creating.

It’s like the jokey school reports in successive terms.

John is trying.

John is still trying.

John is still very trying.

2 Likes

I do agree that small errors are cumulative and problematic.

What I meant was something different. When drawing on a truly horizontal plane, and inferencing from the out-of-plane and/or out of horizontal floor below, you (should?) only get one point on the the new plane. The tilt in the plane below is a very small portion of one degree, and any horizontal offset from such a small tilt will be at most the difference between cos(small angle) and 1.

Which IS negligible when the vertical error is of the order of .001 inch over a horizontal distance of hundreds of feet.

Agreed about the miniscule horizontal differences between the actual result and what it should have been.
But somehow (and I think @Geo is refering to that) you managed to get some very short edges in your new floor model. I found at least one.

Then that I don’t understand.

Scot said the Solid Inspector 2 said it was clean with no stray edges.

How did I create them?