Reproduce an existing sculpture in 3D

What would be some methods to reproduce an existing sculpture in 3D so it can be
3D printed? A very kind Rhino tech suggested posting this on the forum to find the
most straight-forward approaches to achieving this result. I am attaching a pic showing
various views of the piece to be reproduced. Any help would be very greatly appreciated.
Thank You!

composite2S.pdf (3.4 MB)

Hi Peter - digitizing some curves or points off of the real thing is where I think I’d start - do you have access to a digitizer or service that can do this?

-Pascal

Hi Pascal!

I’m not sure what kind of service that would be- I’ve called places that do 3D scanning,
but they quote for a complete scan… and they’re out of my price range.

Thank You for your reply!

peter

Peter,

You might try AGI Photoscan. It uses photos to build the 3D model by photogrammetry. Although, you might need 50 or more pics or more to get a good model. The Standard version is very affordable, you probably would not need the Pro version which has some extra functionality for geographic information systems.

It is possible to draw 3d curves from two views.
You PDF appears to have most of the curves showing and since you created and have them, getting the missing curve bits might be possible.

The Rhino User’s Guide has a series of progressive tutorials.
Towards the end, there is one called Trace Images where given top and side images of a dragonfly, the tutorials shows the techniques to make a 3D model.
It’s worth looking at I think.
http://docs.mcneel.com/rhino/5/usersguide/en-us/index.htm

Hi Peter.
If it’s an existing sculpture and you have freedom to take pictures from every angle (in particular on the other with an angle of 45 degrees) without a doubt I would definitely with Agisoft PhotoScan (available for Mac, but possibly also of virtual machine like Parallels ) is a program that works well in tandem with Rhino. The decisive thing is that the sculpture is NOT reflective material. Other minimum requirements for drives for the Mac (in my opinion) 16gb of ram and CPU 4-core.
I’d say about 70 pictures to take. It generates a point cloud (.ply) and if you want a mesh (OBJ) both can be imported into Rhino. By using the basic tools to generate curves (for point cloud use “PointCloudSection” for mesh the “section” command that generates a polyline, then use the command “CurveThroughPolyline”.
Good luck.

Simon

It is possible to use a spray on dulling product, if the material or environment allows. Ideally, the surface should have some slight texture.

I run PhotoScan on Windows PCs. 16 GB memory and 4 core CPU is good to have but I’ve previously used a laptop with 8 GB memory and a 2 core CPU. It takes longer but PhotoScan will produce results… A 4 core CPU is definitely preferred with a desktop PC but my experience with a 4 core CPU in a laptop is the CPU slows because of heat and may not be much quicker than a 2 core CPU with the same speed rating per core.

@Peter_Stetler Does your reproduction sculpture need to be an “exact” duplicate or does it just need to “look the same”.
If it needs to be an exact duplicate then you will need to have it scanned; either with a laser system or by using photogrammetry such as PhotoScan.

Photogrammetry usually needs a surface with some visual texture, and doesn’t work with a uniform color. I have had good luck though using PhotoScan with boats with dirty painted surfaces.

emmh … I know, but I doubt that this can be done on a sculpture … As for the characteristics of the computer, I assumed that you are on a Mac as we talk about Rhino Mac (which is why I recommended 16 gb of ram and Parallels). Then of course if you use a pc or 10 pc in series with 32 cores and 1TB of RAM (you read that right) you can generate a cloud of points of whole Venice …:stuck_out_tongue_winking_eye:

Of course it depends, but it is certainly worth investigating as an option. The spray I use is typically non reactive on metals and can be brushed of easily after the photographing or scanning is done.

Noise projection is another consideration, and I don’t have any experience with that. I would like to try noise projection with an auto type finish to see if there would be any benefit. I imagine noise projection would be completely useless with a mirror finish, but perhaps there are some shiny surfaces where it would be effective.

Have a look at this:

This is KScan3d, which, if you have a kinect kicking about the house, is pretty good. It works with V1 and V2 of the kinect (although you need to buy a psu for either model). When I downloaded it, the software was free.

I used it for a friend who wanted a sculpture capturing her last month of pregnancy:

The cleanup in Rhino was a fair bit of work, but I think the results are pretty decent.

Have a play around with 123D Catch. I’ve used it a few times to generate rough meshes to build upon.

http://www.123dapp.com/catch

It worked well on my iPhone

Andy

Make sure you have a couple of high contrast things laying about the place when you take the photos to help the software stitch it back together. It won’t be 100% but it will easily be enough for it to work it out.

I’ve been working with a mate who is a sculptor. He carves knots out of stone from little maquettes he creates so I had a go at creating one in Rhino.

Still from 3D model from scan:

Here’s a quick render of the capture from 123D Catch:

Final Model(ish)

Then I discovered T-Splines and the pipe command… But I haven’t had another go yet.