Surface ‘Blending’ in AutoCAD Civil 3D
Out of the box, AutoCAD Civil 3D supports a technique called ‘pasting’ for combining data from multiple surfaces to form a composited ‘finish grade’ surface. The paste technique is basically a cut-and-replace process – the boundary of the surface being added is used to clip data out of the receiving surface, and then the resulting void is filled with the added data. Retriangulation then occurs using the remaining points and breaklines from the original surface combined with the pasted points and breaklines.
If there is any disjoint between the surfaces which are being combined, the resulting triangulation can sometimes be quite ugly where the new and the old data meet. To illustrate this, let’s look at an example where we are pasting a corridor surface into an existing ground surface to create a finished grade surface. At corridor stations, we expect the surfaces to meet up nicely, but what happens in between? Typically something similar to this:
Because the corridor surface comprises a series of straight lines between points at specific, usually regular stations, and because the existing ground surface is typically randomly populated with points, it is almost certain that the two surfaces will not meet precisely along the boundary except at corridor stations. In between, there will be a gap or overlap as shown above. If you consider how this might triangulate in a 3d surface model, it becomes quite apparent that the results can easily be less than satisfactory.
Where a data point exists on the existing ground surface near the boundary of the pasted surface, but in between the data points of the pasted surface, a poorly formed triangle with a steep slope will be created. If the data points on the existing ground surface are farther away from the pasted data, this effect becomes mitigated and the triangles become closer to being considered ‘well-formed’.
When this situation occurs, common practice is to ‘offset’ the clipping boundary to make it larger, thus preventing any data points from being ‘close’ to the actual edge of the pasted surface data. This often produces acceptable results as long as there is not too much variation in the data. Even then, it is still possible to see some undesirable triangulation in the transition area between the existing and new surface data.
Another situation where a similar issue might occur is when you have two separately imported/surveyed surfaces that must be combined to for a single existing ground surface. If there is a gap between the surfaces, or the surfaces don’t agree perfectly where they meet, you will run into the same sort of problem as described for the corridor surfaces above.
In either example, the worse the ‘agreement’ between the two surfaces, the more exaggerated this problem will become, and the larger the transition zone must be in order to achieve anything resembling acceptable triangulation. This comes with its own pitfalls. By enlarging the transition zone, you are removing data from a surface and replacing it with a straight-line transition:
Intuitively, you can see in this diagram that a straight line is not a realistic approximation of what an actual surface would look like. This got us thinking…what if it were possible to ‘blend’ the two surfaces together in such a way as to fill the gap in the transition zone with a curve that is proportionately influenced by the adjacent surfaces (the closer to a given surface, the more influence that surface will have on the shape of the curve).
Thus was born the new application we call Surface Stitcher.
(Please note that Surface Stitcher is still in a prototype phase, but is currently available for preview/beta testing…please contact us for details).
To illustrate what this new application does, we have a pair of surfaces as shown below (to see a paste-and-blend operation similar in nature to the corridor surface scenario described above, please refer to our previous post, Surface Smart Paste Operation ):
Of course we would expect a closer agreement between the boundaries of these surfaces in a ‘real’ project; this particular dataset is intentionally a ‘worst-case scenario’ model we use for testing. It contains a large degree of variation along the edges of the transition zone between the surfaces, voids which penetrate into the surfaces along these edges, etc.
To combine these surfaces, we simply run the Surface Stitcher command to ‘Join Surfaces’ (inside of AutoCAD Civil 3D), select our surfaces, and press the ‘OK’ button:
The result is a new surface containing all of the two source surfaces, plus a ‘blended’ transition zone (the new surface is shown here as ‘background’ contours):
Note that it is also possible with this command to join the surfaces without blending. We’ve created such a surface as a comparison and have imported both variants of the ‘joined’ surface into Terra Power Tools to make visualization and comparison easier (the ‘blended’ variant is shown in blue and the ‘unblended’ variant in brown):
From the 3d display of the surfaces as contours, it is easy to see that the ‘unblended’ surface is far more jagged and angular in the transition zone. If we zoom in a little tighter, it becomes even easier to see the difference:
…and, if we play around with the visualization settings a bit, the blended surface model looks quite good, especially considering how poorly matched the two source surface models were:
For more information about Surface Stitcher, surface blending, Smart Paste, Terra Power Tools, or to request to join our beta testing program for Surface Stitcher, please contact us .