Topic: Nuisance with surface creation.

[therock003]

I have a file that contains some polyline contours (lots of them) and when i define a surface by adding the contours there are some parts that the created c3d contours do not match the initial polyline contours.I had some help at cadtutor as well,till i reached the point that i realised by exploding the polyline contours,that they are a mixture of lines and arcs,and c3d seems to have problem continuing the contour when an arc part is located.

Is it generally a problem for contours to create arc segments?

What can be done in my case,so that the contours i have to match completely the c3d created contours?

[mjfarrell]

You will need experiment with your WEEDING factors, and the other surface creation factors, (smoothing, face flipping, etc.)  And you may never get them to match exactly, (with any product).  As you are asking an algorithm to return the same resultant from different data. Those contours are the interpolation of the original data, (points, break lines), and the data extracted from them is rarely ever going to match exactly.  You can get close; just keep the above in mind.
You might post your data here and we can poke at it to see just how close we can get.

[therock003]

Well it's just that it matches almost every segment except the arc segments!That's y i'm concerned with that!Ivtried the lowest weeding and supplementing factors so as to try and retain the original parameters but still the contours break where an arc exists!

Here is my file you can chek it for yourself if you like to do so.

http://rapidshare.com/files/80952143/contour-file.dwg.html

[mjfarrell]

The 'wrong' areas are NOT only at curves, there are some ares where the data given produces the result model.
The image attached is from my first pass. IT matches what is given, probably into the upper 90 percentile. With a few edits, adding some data, and inferring some breaklines from the data I see in the file, one could probably push this model closer to 100% pretty quickly.

[therock003]

there are some ares where the data given produces the result model.

You mean that contours are also created at parts where the drawing has none?Cause thats also something i wanted to adress.Is there a way to automatically discard this instead of having to manually delete triangles at the corners of the file where false contours are created?

How can i manually edit and add contour data to continue along arc segments.I've already set the lowest of weeding and supplementing factors but that does not seem to slve this problem.

[mjfarrell]

You may NEVER be able to recreate the EXACT contours that you were given, unless you can get the point and break line data they were created from. Before you drive yourself crazier than me, remember Contours are only accurate to within 1/2 the base interval, AND they are an interpolation of the underlying data. OK?

If you think my data is closer than yours I will post it for you to continue adjusting. You might find that adding a few break lines, and or manually add some points to recreate some of the missing contour islands. If you resort to this be sure too add a CALCULATED to your point descriptions, and Name the Breaklines so that you know they did not come from survey data.

[therock003]

I dont want to recreate a 100% of the initial contour data but i want to approximate as far as i can go.

And since this is my learning stage,and i'm deeply concerned with surface creation i think this is a good point to adress such problems.

I'm just searching for ways to improve the models as far as possible.

This is just a template file given on my university for roadway design purposes.When i passed this course i used the plain vanilla autocad and i manually interpolated my alignment and drafted the whole profile and cross section and i'm now looking for ways to automate the whole procedure for future references.

I do realise that hte data to be collected on the field would be points and breaklines so in majority i wont have to be concerned with this,but then there's the whatif scenario of being given a file for analysis that containspolyline contous already being breated one way or the other and i want to be prepared.

PS.Where can i find general knowledge about dem/dtm information?

[mjfarrell]

A great thought exercise then.  Only be forwarned that rarely will a TIN from contours match those same contours, regardless of software applied.

A question I did not answer earlier, one can later apply a Boundary polyline to the TIN to restrict the limits to be only where one has actual data.


Perhaps this could illuminate: http://www.isprs.org/commission2/proceedings06/pdf/karel.pdf

[therock003]

Thanx a lot for all your help,will return though if something else comes up on this matter!

[scout]

Well it's just that it matches almost every segment except the arc segments!That's y i'm concerned with that!Ivtried the lowest weeding and supplementing factors so as to try and retain the original parameters but still the contours break where an arc exists!

Here is my file you can chek it for yourself if you like to do so.

http://rapidshare.com/files/80952143/contour-file.dwg.html

I just pulled it down. The contour adding engine doesn't like arcs, as you have discovered, but you might try adding them as breaklines and cranking the mid-ord distance way down, to like 0.01. I just did that with your file and it seemed to work well.

Another thing you can do is eliminate the arcs first by using some of the pline fit curve tools and just adding a lot of vertices.

[mjfarrell]

As posted the gaps and omissions of the contours happens at other locations including ARCS, the errors have more to do with trying to create EXACT duplicates of contours from contours. I know of no product that will return with 100% fidelity contours on a surface with only contour data as the input.

You will see that using all some or none of the Minimize Flat option in the modeler will also alter the fidelity of the produced surface as well. I found the resultant contours actually closest when I disabled ALL of those options, used small weeding factors AND adding the polylines as breaklines as well as contours to the surface. This does produce a number of duplicate point errors, however the resultant contours are still closer than without them.