It seems obvious that removing those vertices and placing a new one in the northwest corner at a much lower elevation would add volume above the plane. The whole plane would dip or tilt in that direction depending on what base plane was used.
1 Like
Rodger, youāre changing the base plane for a huge amount of area. Leave the vertex (magenta in my markup) and then extend the polygon. Leaving that vertex will allow you to change the shape and include that little corner without changing the base for that large section and your result will be less drastic.
Is there a reason you donāt include more vertices on that top line in order to sample the elevation at more spots? Remember you only want to skip sections if you know that pulling an elevation at that spot would result in a less accurate base plane. Typically, you would not omit a vertex in such a long run unless it was against a severe elevation change like a wall or cliff.
Well the reason why I am not using more vertices on the top line is that (I think) Iām following advices from you a couple of days ago.
and your words: āJust donāt place a vertex in the ājoinedā area in order to interpolate between the vertices on either side.ā.
This line runs along a āslope moving into another slopeā as i described in my initial posting.
This is exactly the ājoined areaā we are speaking of. Not a peak, rather like two huge, 200 meter loafs of bread lying besides each other joined. One of them considerably higher that the other, and Iām trying to measure the volume on the lower loaf, so to speakā¦
An I totally agree that my ātop-viewā drawing wasnāt the best one aroundā¦ 
It is obvious that I need an Enterprise subscription to get this understandable. In fact, and to tell you guys the truth Iām so disappointed at the moment. I like DroneDeploy, but this is just too much, and there is no way that a noob as I am can undestand how this is working (or rather: NOT working) without your help and an Enterprise subscription.
I think iāll have to leave this.
That is true. So what happens is that the last vertex where you decide to stop is very important. In you last example, you removed a few vertices and added another a little farther away. But then you wondered why a seemingly small edit made such a large difference in volume. Iām just trying to help you understand why. Your edit didnāt just add a few feet horizontally. It probably added depth all the way across to the other side.
As I also said earlier, when less than ideal pile conditions exist, you will have less than ideal calculation. Understanding how the processor is coming up with a figure, you just do the best you can under the circumstances.
This IS confusing. I think Iāve gone through the pages covering volume measurements and different base planes at least thousand times, but there is something in the core that confuses me. And there is not much more to read about the algorithms behind. Might be company confidetialā¦ I donāt know.
I realize that the move of the vertices added more volume, but I cannot understand how it can add
7 000 Cu meters by moving a vertex 30 meters (which adds, letās say 900 sq meters to the area). When Iām standing on the site I cannot āseeā this volume, let alone defend the calculation if the customer questions it.
Of course we are doing our best all times, but now I have to rethink if the result is deliverable to the customer. I mean: When a small flaw from me can give this result. āWhat if i placed a vertex on the wrong spot and got + 7 000 cu meters added to the sumā. That is my concern. And I canāt defend the result if somebody questions it.
The vertex, and how itās placed, and what it does holds an important key to all my questions.
Thakās for at least trying to help me out.
How much does the volume calculation change if you leave in the vertex in my markup above?
This:
Gives 20654 cu meters. Compared to 27 000 cu meters with only one vertex in the middle
If I add two more:
The volume drops to 12500.
Another two gives 12250 and yet another two gives 11982 cu meters after which more or less nothing happens to the volume no matter how many verticies i add.
BTW: Where can i read more on what the vertices are doing and which effect they have on the calculations and on the base plane? You guys seem to know it quite well, but to me it is utterly confusingā¦
How in earth do I know where, and when, to stop adding vertices if the calculations are that method (or amount) dependent? I can more or less get any number I want by adding more vertices, if you understand my confusion. Iāve read somewhere that āthe more vertices you add, the better and more exact the calculations will getā.
The main issue for me as beeing an engineer i donāt like klicking around guessing, if you see what I meanā¦ Now Iām guessingā¦
The vertices you enter are used to create the base surface that goes underneath the 3D mesh representing the site. The volume of a pile is the sum of the volume between each triangle in the mesh and the base surface. The volume computed for each triangle can be positive, when a triangle is above the base surface, or negative, when a triangle is below the base surface. The sum of the negative volumes is the fill which would be required to bring all the low spots up to the base surface. The sum of the positive volumes is the volume of the pile above the base surface.
The base surface can be defined from the vertices in many ways. Three which are popular are:
- Triangulated. In this case, the base consists of triangles defined by 3 vertices. If the vertices are uniformly distributed around the perimeter of the pile and none of them are on the pile, then the triangles go from side-to-side of the pile and can follow the gradual terrain variations beneath the pile. We want the triangles to model the earth surface beneath the pile over its whole area so that the volume calculated is an accurate estimate of the actual material in the pile. For your pile and vertices selected this does not work well because the vertices are very non-uniformly distributed and there is one vertex on the pile near the middle of the long edge near the top of the picture.
- Linear fit. For this case, a plane is fitted thru all the vertices you enter. Then the volume of each triangle in the mesh over this plane is computed and summed. For your pile, this works better because almost all the points are not on the pile. But the 1 point on the pile does still tilt the plane some which will reduce the volume calculated. If you removed this 1 point, the volume would increase and may be more accurate.
- Lowest point. In this case, a horizontal plane is defined with a Z equal to the Z of the lowest vertex you entered. This is pretty brutal and can result in a much larger compute volume if the elevation of the earth is changing around your pile.
In terms of the number of vertices to enter, you need to enter enough to enable the creation of a base surface that closely follows the earth beneath your pile. Thus the optimum number varies with the method used to define the base plane. With lowest point, all the vertices are thrown away except the one at the lowest point; just look on an elevation map and be sure to enter 1 vertex at the lowest point around the perimeter of the pile and then only add enough more to accurately define the boundary of the pile. With linear fit, all the vertices are averaged together to create the base-surface plane. For this case you want uniformly distributed vertices so you give equal weight to the earth surface around the entire pile. For triangulated, you need only enough vertices to capture variations in the earth level around the pile; enter more vertices where the level is changing rapidly. If the level is only changing gradually, then uniformly distribute the vertices. In no case should you place a vertex on the pile (although it should make no difference when using the lowest-point method).
With this as background, perhaps it will help you to better understand why the volume changed so much when you moved the upper left corner. If you were using the lowest point method and this created a new lower point, then the volume estimated would increase significantly. Certainly being able to view the base surface in each case would help your understanding. I am working on enabling this in Rhino but it has been about a year since I last worked on this. Once I get going again and can reproduce your results, I will look at adding a 4th base-definition method that works better for your case. And of course lets you see the base.
Regards,
Terry.
How is the base surface calculated in the first place?
The long edge without vertices is directly connected to another āpileā (or a loaf of bread - formed pile) and the one Iām trying to measure is a similar looking pile, but much lower. It sort of climbs up on the larger pile. I got an advice not to use vertices along the border between the two loafs.
Exactly how are tre triangles bult up from 3 verticies? Cannot find anything written on this. In this case I tried to follow an advice. I might have misunderstood the adviceā¦ It has happened beforeā¦ 
I normally try to use LOTS of verticies beacuse somebody somewhere told me to do so to increase the accuracy. But I really donāt uderstand why, to be honest.
Maybe I have to move to knitting insteadā¦
For the Linear Fit method, the base surface is a flat plane, tilted in x,y,z so that the distance between this plane and the vertices is minimized (thus it is a best fit plane).
For Lowest Point method, the base surface is again a flat plane, this time it is horizontal with its Z at the Z of the lowest vertex.
For the Triangulated Method, the base surface is comprised of many triangles. The vertices in the triangle fall on the vertices you entered such that none of the triangles overlap. How this looks in detail depends upon the program used. You do not want any of these vertices to fall on the pile or some of the many surface triangles created will not lie on the earth beneath the pile. This is why you got the advice not to put vertices where they fall on top of a pile.
Is this any help?
Regards,
Terry.
Sorryā¦
Iāve read this many times, but still gasping for airā¦ :-/
Or you want a straight grade slope.
Sure, or that. So long as you understand that dropping a corner 6 or 12" on a 1/4 mile leg is going to have an effect on your calcs !
1 Like
1/4 mile? I thought the corner drop was for some 30 meters in width and some 3 meters in height? Does this corner-drop affect the whole area by changing the whole base plane?
I didnāt know thatā¦
However, @Dave, @MichaelL and @SolarBarn. Do you know where to find something more written about the subject for this entie discussion? I see what you are trying to tell me, but the basics are missing, which resuilts in a lack of understanding, and also that I canāt provide the customer with the msmts for the piles since I donāt know what is right and what is not. Neither can i explain how three clicks on the long side can reduce the mesured volume from some 27 000 cu meters down to 12 000 or so nor do I understand it myselfā¦
It becomes so embarrasing if the customer asks questions.
I canāt ask for more help. Youāve helped me so much already. Links or hints on where I can read and learn more would be appreciated? For time beeing i have to stick to piles that are ānormalāā¦ 
R
Donāt work about taking up time. This is how we all learn so keep the discussion going. I think the most important piece that is missing that would provide understanding are visuals of the differences of the different bases and how they look in comparison the each other. This is best noticed in the 3D model view, switching between the planes and looking for the gray line.
Linear Plane - It is a flat plane that causes fill on the bottom and cut on the top. Averaging the difference between the two.
Triangulated - Matches the condition according to where the points actually are by creating triangles on each set of 3 closest neighboring points.
Lowest Point - All points are defaulted to the lowest found point.
Custom Elevation - I originally requested āhighest pointā which this accomplishes and betterā¦
yes - this is exactly whatās happening when you add a new vertex.
Wonderful illustrations @MichaelL - and glad you found custom elevation - weāre just testing that out ahead of announcement at DDC. Great for cut/fill to a known base/slab elevation, or for filling a retaining wall / pond / retention berm to a certain depth. It defaults to āhighest pointā before you edit the number.
There is a known bug that the 3D visualisation for custom elevation does not update when the elevation is changed (unless you refresh your browser or change the geometry).
1 Like
Haha, I guess I let the cat out of the bag?
Just an idea on the base plane visualization, would it be possible to change the color? The gray can be quite hard to see against many of the materials we deal with. I find DroneDeploy blue or yellow work wellā¦ Maybe just the outline?
Geee. I wish I had the possibility to fiddle with coloursā¦ Would be easier to understandā¦
I did some tinkering with one of my maps.
This is a section of a wall rising up to about 5 meters. It is roughly 15 meters wide at the base, and about 5 meters wide at the top (the gray area) after which it starts descending on the other side. This structure surrounds a land-fill area.
To begin with I added a āsquareā on a slope with just 4 vertices in each corner. Just to see what happens:
834 cu meters
Then I doubled the amount of vertices on each of the longest sides:
This is exactly what gives me a huge anxiety: The method is so extremly sensitive and almost impossible to validate, let alone explain to the customer. When/how do i know how many vertices is correct to give a correct reading? This example is spot on what bothers meā¦ Iām getting hilarious differences while adding and/or removing vertices and the only way to learn this seems to be by harassing guys llike you.
DrodeDeploy is handing out an excellent piece of software, but the rest is up to me. Which normally is fine if there is stuff to read and dig in to.
Meanwhile there are customers out there asking for helpā¦
R
I can see how that would be nerve racking, but you have to ask yourself what is the purpose of the takeoff. If I could see the elevation layer or the 3D view I could tell you why there is such a delta, but this is exactly what triangulated does. It feeds from every point instead of doing some type of interpolation of all points.
It looks like maybe the middle points you added are on the top shelf? Those bushes are on the ridge? If so my guess is basically you created a fairly flat plane on top and a slope on the north end that looks like to me whatever material is plane off from the south included with the square edge of the ridge just happens to balance to fill the bottom slope. Quite a lucky (unlucky) pick.
The short line is the top plateau. Some 5 meters in length. The Whole wall is some 5-5 meters high. Some 15 meter down at the base.
I can see something āfloatingā above the āroofā. Some gray shadow. Can you explain this:
A thought: The vertices are so few and far from each other thet the triangulation gets inexact? If I add more verticies it will more and more fit into the actual surface, and the measurement gets more and more exact? True?
Butā¦ The ore verticies I added the worse it got. Quite the opposite what i thought:
Still these gray shadows. Looks worse the more verticies I add. So the quality and reliability does not correspond to the number of verticiesā¦ Somwhere along the line you must know how many verticies to add and when to stop and where to place them. In advance, so to speak.