Page 7 of 9

Re: 3D surface modeling of a giant redwood trunk

Posted: Tue Jan 31, 2012 6:45 pm
by fooman
HI Michael,

The company I work for has recently purchased Rapidform XOR, a reverse engineering software package. I just installed it and had a play with a point cloud you uploaded. See below:
Melkor in Rapidform XOR
Melkor in Rapidform XOR
I've only just started to play with this software, so was just using your data to familiarise myself with it. The computer I'm on chokes on it a bit - probably needs a better graphics card. It can do textures, but not texture tiling (or at least I haven't found that).

Out of interest, I've been at a plant shutdown where another company was doing an internal laser scan of a large vessel (~ 6 m in diameter, 30 m tall). They were getting mm-scale accuracy of the internal lining with a Leica ScanStation (http://hds.leica-geosystems.com/en/Leic ... _79411.htm) and Rhino3D (http://www.rhino3d.com/). They did about 2 days scanning, 3 days on the software (but a lot of that was generating models from the engineering drawings to compare the scan against). Looked very impressive. Maybe $200k worth of kit, and they would probably charge ~25k to 50k for the job.

Cheers,
Matt

Re: 3D spacial modeling of a giant redwood trunk

Posted: Thu Feb 02, 2012 3:36 pm
by mdavie
M.W.Taylor wrote:Don and Frank,

Yes, this is exactly what I had in mind. A complete tree scan is currently in the works. Climber deployed linear scanning up the rope is going to be an important part of the project.

Michael
Um, if you need any more climbers to help with that scanning, let me know and maybe I can swing a trip out there. I'd guess you've got it covered, though.This is amazing stuff you're doing!

Re: 3D surface modeling of a giant redwood trunk

Posted: Fri Feb 03, 2012 1:49 pm
by M.W.Taylor
Mark,

I will keep your request in mind. National Geographic's Wild channel producers have taken interest in sponsoring this project. Can you be in NW California with 48 hours notice ? We may need addtional climbers if they don't hire their own.

Michael

Re: 3D surface modeling of a giant redwood trunk

Posted: Fri Feb 03, 2012 2:04 pm
by M.W.Taylor
Matt,

Your mesh and color choice look fantastic on Melkor. I can see I missed a little spot near the basal burl. I will need to go back there and those surface points. When you get a chance, check out the attached. It's a lower trunk scan ply file of Stratosphere Giant, now 371.5' tall. The trunk is highly out of round.. This cloud set is not as dense as the Melkor one. I am curious how it would look meshed and colorized on RapidForm XOR. I notice your copy will soon expire.

I will need to go back to Stratosphere and re-scan the trunk from closer to get the fine details. If I do, I'll repost the denser cloud set.

Michael
fooman wrote:HI Michael,

The company I work for has recently purchased Rapidform XOR, a reverse engineering software package. I just installed it and had a play with a point cloud you uploaded. See below:
The attachment Melkor.jpg is no longer available
I've only just started to play with this software, so was just using your data to familiarise myself with it. The computer I'm on chokes on it a bit - probably needs a better graphics card. It can do textures, but not texture tiling (or at least I haven't found that).

Out of interest, I've been at a plant shutdown where another company was doing an internal laser scan of a large vessel (~ 6 m in diameter, 30 m tall). They were getting mm-scale accuracy of the internal lining with a Leica ScanStation (http://hds.leica-geosystems.com/en/Leic ... _79411.htm) and Rhino3D (http://www.rhino3d.com/). They did about 2 days scanning, 3 days on the software (but a lot of that was generating models from the engineering drawings to compare the scan against). Looked very impressive. Maybe $200k worth of kit, and they would probably charge ~25k to 50k for the job.

Cheers,
Matt

Re: 3D surface modeling of a giant redwood trunk

Posted: Fri Feb 03, 2012 6:48 pm
by Don
Mike, WNTS LiDAR users-
I recently read an article "LiDAR as a Tool to Characterize Wildlife Habitat: California Spotted Owl Nesting Habitat as an Example" (Journal of Forestry, December 2011, p.436).
Easy to overlook for most forum members by virtue of it's source (not me, it's peer reviewed and tries to bring forward the most recent papers on technology advances), by title (what's Spotted Owls got to do with it), and California...
I recommend it for the following reasons...the authors used multi-return LiDAR to identify 1)first return (mid-tree, Distance 1), 2)multiple return (Distance 2), 3)Multiple Return (Distance 3), 4)last return (Bare Earth). THis provides a very good reference data for the topography, and while expensive, it does provide excellent crown assessment data.
Where of? California...Sierra Nevadas...Mixed Conifer Zone...in Tahoe NF, El Dorado NF, Sierra NF...size (107 square kilometers)...up to about 6500 feet.
Thought you might want to know!
-Don
PS:As an aside, I noticed that the authors used a densitometer to obtain presence/absence crown canopy data...one similar in function to the GRS Densitometer (which was designed in Humboldt County!).

Re: 3D surface modeling of a giant redwood trunk

Posted: Sat Feb 04, 2012 2:11 am
by fooman
Matt,

Your mesh and color choice look fantastic on Melkor. I can see I missed a little spot near the basal burl. I will need to go back there and those surface points. When you get a chance, check out the attached. It's a lower trunk scan ply file of Stratosphere Giant, now 371.5' tall. The trunk is highly out of round.. This cloud set is not as dense as the Melkor one. I am curious how it would look meshed and colorized on RapidForm XOR. I notice your copy will soon expire.

I will need to go back to Stratosphere and re-scan the trunk from closer to get the fine details. If I do, I'll repost the denser cloud set.

Michael
Hi Michael,

Don't be too worried about the gap at the base - that is probably an artifact of the semi-automated meshing that I was playing around with. Same for the flat slices at the top and bottom - that was to remove some of the gaps at the edges of the cloud. We've got a how-to course on the software coming up in a couple of weeks - I'll see what I can do after that. There is a lot of manual healing that can be done on meshs.

Cheers,
Matt

Re: 3D surface modeling of a giant redwood trunk

Posted: Tue Feb 07, 2012 4:27 pm
by fooman
A quick play with the Stratosphere Giant data. Still have not figured out textures that well.

The software has dedicated measurement tools (it is actually designed to measure the difference between a mesh and generated solid models, as a way of quantifying the accuracy of reverse engineering). The mesh as shown has a volume of ~1.5 mm3. It obviously reads the .ply data a certain way. Next stop, figure out the scaling required to get an accurate volume.

Cheers,
Matt

Stratosphere Giant mesh, with partial texture
Stratosphere Giant mesh, with partial texture

Re: 3D surface modeling of a giant redwood trunk

Posted: Wed Feb 08, 2012 12:29 am
by M.W.Taylor
Matt,

That looks great. I am going back to Strat soon for a better scan with 5x the cloud points this time + I will put an 8ft red rod along side the trunk for scale. These cloud sets are in correct proportion but not scale as you can see... if I scale the entire cloud set to the rod only, everything else should be scaled correctly. So far that has worked with the other scans. I am curious what volumes your RapidForm XOR comes up with compared to my custom VBA code ? Let's compare volume solutions as one way to test my "untested" code. Also, I am coding a solution for total surface area of the tree form. Does RapidForm XOR do that ? Thanks for your interest in this Matt and also for posting your tree meshes.

Michael

Re: 3D surface modeling of a giant redwood trunk

Posted: Thu Feb 09, 2012 5:41 pm
by fooman
Hi Michael,

Another quick play, to get volumetric data. Method as below:

Slice of redwood creek giant, 0<z<20ft.

Imported spreadsheet data containing vertex data with 0<z<20ft as .ply into Radidform. Automatically mesh. Filtering of extranous data points, 80% sampling ratio. Automatically filling holes resulted in smoothed, interpolated surfaces at z<0ft and z>20ft. Trimmed resultant mesh back to 0<z<20ft. Global remeshing resulted in slight smoothing of edges. Volume calculated at ~3076 ft3, with face normals pointing inwards (calculation instantaneous). Corrected normals to obtain "correct" model display, but unable to obtain a volume calculation. Mesh shown below:
Redwoods creek giant, mesh of slice from 0 to 20 ft.
Redwoods creek giant, mesh of slice from 0 to 20 ft.
Ran free form volume solver-RGC.xls with the following settings:

Range Filters # of Points = 61627 127 = # of Slice Points
Xmin -10.000 Xmax 11.000
Ymin -10.000 Ymax 11.000
Zmin -1.000 Zmax 30.000
Pole XYZ 0.000 0.00 0.000
Zcookie 0.000 Thickness 0.200
Ht Start Pt 0.000 Ht End Pt 20.000 Int 20.000
Slice Thickness 1.000 Ray IncÂș 1.00

Took about 10 minutes to get the result:

Total Volume 3163.027

The mesh volume is ~ 97% of the numerical intergration volume. The difference is likely to be in the filtering/smoothing/filling undertaken on the mesh plus the residuals inherent in the numerical integration method (e.g. as a function of ray angle/slice sizes).

This seems to be a reasonably good result.

Cheers,
Matt

Re: 3D surface modeling of a giant redwood trunk

Posted: Thu Feb 09, 2012 8:41 pm
by Don
Matt-

Eventually, I suspect some of the details of volume will need to rise up to consideration, such as inside bark volume (while timber people might want to know this, but I'm thinking more towards mass, specific gravity, weight per unit volume, etc.). Any ideas how to, remotely?
-Don