The vertices of the triangulation are always supposed to follow the closest neighbors and shortest path. It doesn’t look like it’s doing that in this case. This might be a bug in the system. Most stockpiles that are taken off are on fairly flat ground and even when there is a brake line it’s not very steep. @Jamespipe do you agree? There’s something wrong with the triangulation here.
Bug or no bug… Smoke is coming out of my ears… Fairly close to a melt down…
Enough for tonight.
Do you have some kind of hint on where to find more info regarding this?
R
I think that slope is an exaggerated condition, but points out a flaw in the process. You could try putting some points on the midpoint of the slope to see if it will create new triangles to capture it, but normally with that kind of slope you are doing something different than creating a flat plane on all of the existing surfaces.
I think the info we are giving you from experience is probably as good as it’s going to get as this is a fairly new function of drone use with different base planes.
@MichaelL. But when do I know when i’ve reached the point where it’s enough with vertices? Impossible to know since the volume changs with every new point I add.
I’m after to reduce the “guessing” - part of the procedure.
I have been trying out various confugurations of vertices, and… well, the only thing that changes is the volume (cut/fill) AND the grey “curtain” i asked you about earlier. Is this grey curtain supposed to disappear (or disappear into the roof) when i’ve reached the correct amount, and placement of, verticies? ’
Here (Triangulated base plane) managed to almost get rid of this grey curtain usin just a few verticies, but then the volume went down to just 13 cu meters. Why?
When I switch to Linear I’m getting this:
Let me think: Am I seeing fill followed by cut followed by fill again? The base plane is not levelled because the base plane is a reflection of the area starting on one height and ending at a higher level? Correct?
The Triangulated view givers me 13 cu meters and linear gives 208 cu meters…
Once again apologies for harassing you with my stupid questions, but man, this is difficult…
R
The problem here is we don’t know the intent of the takeoff. Are you trying to cut the berm out, smooth the slopes, fill to a point? The volume is very small because you have a cut at the top of the berm that almost equals the fill at the toe of the berm. Here are the points I see needed if you are just trying to capture the existing slopes and end up with a 0cy balance. Keep in mind from what I am seeing I don’t think it is 100% working correctly. Let’s see how this goes.
Correct. All the points and the slope across have been averaged. This is a perfect example of when triangulated is better unless you are going for the exact condition. There’s not much more fill here which you should be able to tell from the report, but allot more cut.
@Jamespipe, @MichaelL
I made a new annotation this morning only covering a part (not the whole since it’s hidden) of the ridge:
DD calulates the value to 594 cu meters (Cut) and 0,3 meters fill.
Manually I calculate this portion to 650 cu meters. Seems OK. I released a huge “Sigh” when I saw this…
But… Then I started to add vertices (equally distributed along the long sides)
the volume drops to 12 cu meters. And this is my main concern regarding this whole discussion. The method dependency. I don’t like a methods in which the whole result depends on how many verticies I add to my annotation. I have to more or less guess how to place my verticies and how many, and if I add the wrong amount of verticies the volume goes bananas. It’s like I’m solving an mathematical equation, and can pick my own way to solve rather than use the scientific (validated) method proven to be correct. “Yes, I have a result here. I’m not sure it’s OK according to science, but still a result that looks nice”, so to speak.
Validation. The result cannot be validated since it’s so dependent on the number of verticies (=the user) and nowhere is there a written rule of thumb about this. I have to guess. Or harass you guys with endless discussions an a ton of questions…
There is something fundamentally wrong with the whole method. I do really hope this is a bug. Otherwise I don’t know what to do.
Is there an explanation to why the volume acts like this?
R
I think there is still a misunderstanding of how these base planes are designed to function. When digitizing we are designing what we want the ground to look like once the material is removed. Even though you used triangulated on both you emulated a linear plane with only 4 points on the first example.
The linear plane averages the values of the vertices that are input in an X line and a Y line establishing slopes accordingly. Changing vertices cause a proration of the elevation on the side of the plane that you changed.
Triangulation takes the specific value of ever vertice into account and basically creates several hinged smaller planes. What you have done in the second example is all but matched the existing surface which is why there is very little volume. Are you seeing the gray plane? This is showing you exactly what you are designing.
I can see tiny bits of the grey plane, which is obvoius and why it only returns some 12 cu Meters.
Reading your replies I can tell that it’s long time since you guys were in the same situation as I am at the moment. Or is it the fact that I’m lacking language skills and my explanations are not understandable. I’m not reaching you guys.
I’m feeling that you are replying to something else than what I’m asking for…
It is like to have a dozen of soaps in a bucket of water and try to catch one of them. Every time I think I understand something it slips out between my fingers. I’m missing documentation om how to get good quality (quality markers), and the lack of docs makes every volume measurement a gamble. When do I stop? How do I know when the numer of vertices is enough. How should I place them. What to think here, and what to think there.
There is ONE page with instructions I’ve read at least a thousand times.
Imho I think the whole software should be redone and a kind of a “guide” be introduced.
There are other flaws as well: You have to click on the polygon to add verticies to your volume mesurement.
Then if you want to take a look on the baseplane you’ll have to back a couple of steps and go into the “Model” selection. While in there you cannot add verticies, just delete and move the existing ones around, and to add new ones you have to reverse two steps and start all over agan.
A “normal” pile on a normal, flat ground works perfectly. The trickier the task gets, the more you have to gamble and guess. And somewhere along the line there is a customer that ordered this volume measurement. Which of all the results (depending on the number of verticies) do I stamp “Correct”? That’s the question.
I kind of resigned. I give up. It doesn’t feel ok to harass you more. Enough.
I cannot rely on DD:s calculations since it’s so extremly method (=vertex) depending and therefore i cannot deliver a result i can defend to the customer. And there is nothing to read. No explanations. No rules of thumb. Nothing.
R
I think there is quite allot of confusion on your end as to what the end result expectation is. You have to be able to visualize what the intent for the final ground is and how material is going to be removed. This is not a skill that can just be had at a whim. Many of us come from backgrounds where we have been dealing with this kind of information for many years and it is obvious to us that people unknowingly expect to just pick up a drone, some software and start making and quantifying maps. I can assure you the deeper you go the less this is the case which is why many of us are on here. IMHO it’s not DroneDeploy’s responsibility to teach us theory, but to provide the tool to do the job and provide simple instructions on how to use that tool. They don’t do what we do.
It really is pretty simple the design result which each base plane produces. Having a better understanding of vertices and grade-breaks is also required to get it right the first or even second time. The best workflow is to get the general shape in 2D with or without the aid of the elevation layer. The rest of the fine tuning should be done in 3D and you CAN add vertices simply by clicking and dragging the midpoint to the desired location. There should be no switching back and forth once you get to the 3D view. I draft all the areas in 2D first and spend the rest of my time in 3D.
Sorry we couldn’t help more, but we will be here if you have any more questions.
And I am totally confused now. I thought I had a grip, but this last project really killed everything i knew. Which wasn’t much, to be honest…
However: I slighly disagree here. I think that DD should provide us with some kind of knowledge to increase the over-all quality. Of course we’re the ones doing the job, but DD has so much, and profound, knowledge.
If I’m buying a new car the car dealer provides me with knowledge especially for this car (service). The car normally has a thick technical manual and a IFU. The manufacturer has the profound knowledge of their product and are trying to spred the knowledge to me as a buyer which will spread out on the market: “Volvo has an excellent service”… or “They have tons of material”. In this case the dealer is sending me out to other car owners to ask for help.
The last was a bit far fetched, but what I descibe is what I’m missing…
I’m returning back to the quality issues, or validation if you like, of the process. In my eyes the process is not validated and not consistent in quality when a slightly wrong placement, or too few, vertices can spoil a whole measurement. And no way to control what’s happening. And nothing written anywhere.
I think I have to look elsewhere for a provider that can deliver what I’m looking for: a quality assured service. Although I think there are none out there that are even remotly close to DD, so I guess I will return. With new questions…
Meanwhile I’ll stick to simple piles. They are at least working…
And finally: The end result and expectations. Yes. I hoped too much. My simple land fill project didn’t work here… 
Although I haven’t been able to get you to understand my questions due to language problems I thank you for at least trying to help.
R
Who do you think they learned these processes from in order to engineer the product? Us. They are not surveyors, agronomists and construction workers… I think they know less about this process than you give them credit for when they know more about the design intent.
Does the dealer teach you how to drive the car?
I have to disagree right here. The process cannot be validate if it is not input correctly in the first place. If you have no idea of how to input then what are you claiming a lack of validation against? The points are totally in your control.
Maybe you should ask for a formal training class. Good luck getting any more elsewhere. I’ve gone through pretty much every other processing solution out there and none have even close to the practical support that is available here. Like I said before this is not something that some master genius came up with ages ago and developed a textbook for and performs classes on. We have only been at this for about 5 years and I still learn something new every week, sometimes every jobsites. You strike me as a really intelligent fellow that is use to picking things up instantly and have just gotten aggravated. If you don’t want to grind on it don’t blame DroneDeploy. I don’t work for them, but I can tell you that they do a fantastic job at what they do and I think you are making what you need to do harder than it actually is.
That said, I would be happy to collaborate on the map with you and provide a little consulting time if you’re interested. Better than you are going to get anywhere else.
Of course the car dealer won’t teach me to drive, but he/she is teaching me what’s unique with the car in question. 
Haha… I really do not understand the sentence “harder than it actually is”. This is hard to understand, Maybe not for you, but…
Regarding to help me out. I really, I mean REALLY appreciate it, but I’ve already told the customer that I can’t deliver a volume measurement on this land-fill since the tools are not collaborating with me. I think I have some 10 000 cu meters, but I can’t quality asssure this value due to factors I tried to explain earlier
And besides that: You, and the others, have already helped me so much. I can’t ask for more. If i’d understood how the base planes works this would be easier to understand. It would also help me A LOT if the colored cuts/fills would be available for me. I think this would clear things up…
Maybe next time. This gig was the last for this season. Sad that it ended up in a failure… :(, My trial will expire in one week and the winter is closing in, and with it low light and sub-freezing temperatures. .
But i would have given much…
Of course I’ll be back, as Terminator said.
R
This aggravates me that DroneDeploy is withholding what I consider a mandatory piece of information in a 3D cut/fill scenario. It might actually make understanding the cut/daylight/fill scenario a little easier.
The offer is still in on the map if you change your mind. If prudent maybe you could share the images and I could process and takeoff on my end as your validation… We are on Enterprise so I could get you full access once it is completed. I just want people to be successful selfishly so my company has a better product to work with. 
I have to say OK to your offer… How do I share the 3D, project or map?
You won’t be able to on a trial, but you could PM me a link to download the images or just put your email and I will send you an upload request.
roger@humanteknologi.se
Some 270 images in total.
Link on the way. That many images shouldn’t take long to turn around.
On its way now. This is the area I’ve been fighting with trying to measute volume…
As I mentioned I THINK the volume is some 10 000 cu meters, but… well, I don’t know.
So grateful for you taking time to take a look.
I have to get some sleep. Lets talk more tomorrow.
1 Like
Roger, when you finally get this, you will then understand that you are really making it much harder than it is. We promise!
Take your first polygon with no extra vertices between the low point on either side of the berm. What you are telling the software is that you want a straight line drawn under the berm between those vertices and include all the volume from there to the surface. And that is exactly what it did.
In your second image, you added more vertices across the berm. What you are telling the software with every vertex is that you want to sample the elevation of the SURFACE of the berm at that point. So, in effect, you are telling the software that you want to include much less volume because with each vertex, it is going back to the SURFACE elevation to plot it’s angle through the base of the berm. And that is exactly what the software did.
Any time you add a vertex you are telling the software that you want to draw a line at the base that connects those SURFACE elevations. When that sinks in, you’ll understand that you’re making it way too hard.
This site consists of fill inside a berm as shown in this perspective picture:
If I click around the perimeter similar to what you did:
and select the Lowest Point method to define the base surface, then this gives a volume of 51,425 cu.m. as shown in this perspective view:
The base surface is the flat plane beneath the pile in this perspective picture and the pile is the 3D shape above it. The volume estimate for the pile is the sum of the volume beneath all the triangles in the pile and this base surface. The volume includes the volume of the berm and all of the contents inside.
If you use this same outline later on to check this site with the 3D model created from a later mission, then the change in the volume will be reflective of the amount of material added or removed from inside the berm. For good accuracy, the outline needs to be carefully positioned over the pile in the same way.
Is this what you want?
Roger, like for you, the DD pictures do little to help me understand what is going on. They do not seem to show the base surface used to compute the volume. The color in DD looks nice and is something I would like to add to my Rhino App when I display the pile over the base surface.
There is a conundrum in entering the points of the outline:
(1) On the one hand, these points need to accurately define the X-Y region of the mesh that should be included in the volume calculation
(2) But on the other hand, the Z of these points need to be used to define the elevation of the base surface beneath the pile.
For this case it is only easy to do (2) if you use the Lowest Point Method for base surface definition. With the other methods, the base surface starts moving upwards towards the pile and rapidly decreasing the volume estimate.
Regards,
Terry.