Loading

Venus de Milo

by CosmoWenman, published

Venus de Milo by CosmoWenman Dec 9, 2013

Featured Thing!

Description


Update 6/23/14: I used a compilation of photos of other people's prints of this model in a presentation to LACMA on February 3, 2014. Here's an adaptation of my presentation: *[3D Printing, 3D Capture, and Opportunities for Design Custodians](https://medium.com/@CosmoWenman/3d-printing-3d-capture-and-opportunities-for-design-custodians-7985097d2ac4).* Please share it with anyone you think might be interested -- Thanks.
***
*"Supreme western works of art, like Oedipus Rex and Hamlet, preserve their indeterminacy through all interpretation. They are morally ungraspable. Even the Venus de Milo gained everything by losing her arms."* -- Camille Paglia, *Sexual Personae: Art and Decadence from Nefertiti to Emily Dickinson*
****
In the 19th century, important works of sculptural art were reproduced in plaster. Artisans carefully made molds of the ancient originals, and high-quality reproductions were then cast in plaster to be bought, sold, and traded by museums, universities, art schools, and private collectors everywhere. Plaster casts of the 2nd century BC *Venus de Milo* were very popular, and would have been found in cast collections all over the world.

But the plaster cast tradition [faded](http://goo.gl/aFX84x) in the early 1900s. Cast collections were broken up, sold off piece by piece, and in some cases actually physically destroyed. Today there are only a few sizeable collections of plaster casts left in existence.

The Skulpturhalle Basel museum in Switzerland maintains one of the world's few surviving large collections of plaster casts. They have a very high quality cast of *Venus de Milo*, which was commissioned by the University of Basel and carefully cast by the Louvre's own atelier in 1850.

In September 2013, with the museum's permission and the financial support of Autodesk's Reality Capture division, I spent a week working in the Skulpturhalle, taking 3D surveys of my choice of casts. I took hundreds of carefully staged photos of *Venus* and used Autodesk's *ReCap Photo* photogrammetry software to process them into this high-quality 3D model.

This model of *Venus de Milo* is a modern artifact and direct descendent of the plaster cast tradition, which is poised for a 3D captured, 3D printed, digital renaissance. It is to my knowledge the first high-quality 3D survey of the *Venus de Milo* to be freely published, and I am pleased to be able to offer you direct access to its ancient, enigmatic, and graceful contours, which descend to us through an unbroken chain from antiquity -- from the Greek island of Milos 2,100 years ago, to the Louvre, to the Skulpturhalle, through my camera lens, to you.

You can read more about my project, *Through A Scanner, Skulpturhalle,* at ThroughAScanner.com. I will be publishing more results, including *Winged Victory of Samothrace* and the *Medusa Rondanini,* among others, and cataloging them here: thingiverse.com/thing:83781 I post occasional updates on Twitter as well: twitter.com/CosmoWenman

If you know anyone who would be interested in sponsoring more of this kind of work by me, please send them my way.
Thanks,
-- Cosmo Wenman
cosmowenman.com
twitter.com/CosmoWenman
[email protected]
****
The 3D print shown in the photos and video was made with white PLA and finished in patinated copper using Alternate Reality Patinas (coming soon!). It's the first of its kind, and it was shown at the London and Paris 3D Printshows in November 2013. **It's for sale.**

I am also offering 3D prints of *Venus de Milo* through my Shapeways shop, shapeways.com/shops/CosmoWenman
In addition to my Shapeways offerings, [I can cast *Venus* for you in bronze or stainless steel](http://cosmowenman.wordpress.com/2013/10/24/3d-scanning-3d-printed-lost-pla-bronze-casting-and-the-art-of-the-living-dead/) in any size.
****
Information, Tools, and Specifications:
*Venus de Milo*
Skulpturhalle Basel accession number 261
1850 plaster molded and cast from the original by the Louvre atelier, with only very slight evidence of parting lines, and a false patina likely added in the 1950s or 1960s.
Skulpturhalle Basel: skulpturhalle.ch
History of the Skulpturhalle and uses of plaster casts: goo.gl/aFX84x
250 JPEGs processed with Autodesk's *ReCap Photo* cloud-based photogrammetry application: recap.autodesk.com, using every other third photo from 663 sequential positions, plus a selection of detail shots.
Camera: Canon EOS 5D Mark II
Lens: Sigma EX 50mm
Using a wireless remote shutter release and a loose tripod for shots at eye level and below, and a tripod on extended legs with wheels for shots from above.
Camera settings:
f/16
ISO-100
.8 seconds
No flash
5616 x 3744 pixels per shot
Lighting: Normal museum lights plus three additional fixed spotlights, with diffuse light from skylight on a cloudy day.
Scaling, cropping, sectioning, and print preparation of the model was done with *Blender,* on an under-powered Windows 7 PC that is dying at this very moment, as I type.
Scale references in the raw unedited .obj*:
The grey plinth *Venus* stands on is 690mm x 519mm x 337mm (345mm to the floor).
The white balance reference card is 254mm x 216mm.
The "Frauen und Sport" poster is 2000mm wide.
The "Weitere Sportarten" poster is 1700mm wide.
The "Wagenrennen und Reiten" poster is 2188mm wide and 915mm from the floor.
(*I am having trouble uploading the original .obj and textures of the unedited capture -- they are very large files. I'll keep trying and post a link if I wind up hosting it somewhere else.)
The model has been only very lightly re-sculpted with *Meshmixer* to fix one spot at the top of the head to fill a hole in the physical plaster cast, and to edit a capture defect, also at the top of the head, which the photos did not capture well. The valley where the back of the figure's right arm meets the side of the torso was made slightly crisper. A few small areas in some of the valleys in the drapery are inverted, convex where they should be concave, but these were left as-is.
****
Printing:
The complete, single-part model needs supports under the figure's right arm, and under the face. I've printed it small, with no infill, with RepG for file prep and a Replicator1.
The multi-part sections print well in PLA with no external supports. I've printed them several times with RepG for file prep and a Replicator1, with zero infill, 3 walls thick, at .2mm layer height for the body pieces, and 10% infill, 4 walls thick and .12mm layer height for the face and crown pieces. At its current scale, the assembled print stands about 19.5 inches tall, and is roughly the same scale as my [*Winged Victory*](http://www.thingiverse.com/thing:196038) model -- they make a nice pair side-by-side.
Other machine and software combos may require different settings.
****
Acknowledgments:
This project was made possible by the financial support of Autodesk's Reality Capture division. My thanks to Tatjana Dzambazova and Brian Mathews there for making it happen, and to the people in Autodesk's Reality Capture and 123D teams, who make incredible products. recap.autodesk.com + 123dapp.com
Thanks to Dr. Tomas Lochman, director of the Skulpturhalle Basel museum, for giving me access to the museum and allowing me to conduct this experiment in extending the reach of its incredible collection and the spirit that informed its creation. skulpturhalle.ch
I'd also like to thank the following:
Bernard Frischer, Professor of Informatics and Director of the Virtual World Heritage Laboratory, Indiana University, for his advice on where to try this experiment *(e dove non),* and for introducing me to Dr. Lochman. frischerconsulting.com/frischer/projects.html
FARO, for making a FARO Scanarm Edge laser scanner available to me, and to FARO's Florian Fuenfschilling and Chris Bartschat for their expertise operating it and grabbing some laser scans of additional pieces for me. faro.com
Ralph Wiedemeier, who made and lent me a custom, 10' tall tripod on wheels that was absolutely critical. He even delivered it to me from Zurich. framefactory.ch
Bre Pettis and Kio Stark for early feedback and advice on my Kickstarter campaign. brepettis.com + kiostark.com
MakerBot for featuring my Kickstarter on Thingiverse.
Kerry Hogarth of the 3D Printshow, for giving me tickets to the London, Paris, and New York shows to use as Kickstarter rewards, and for exhibition space there to show off the project's results. 3dprintshow.com
Susan Self for her help promoting the Kickstarter to the media. [email protected]
Virginia Postrel for her advice and help telling the story of the bigger picture as it's been coming into focus. vpostrel.com/power-of-glamour
Thank you to all my Kickstarter's backers. Even though it wasn't the viral hit we'd hoped for and the campaign fell through, without their support and help promoting it, the project would have remained dead in the water. kck.st/1bxMz0I
-- Cosmo
****
Related media mentions: cosmowenman.wordpress.com/media-mentions/
****

Recent Comments

view all
It's possible to select and reselect which photos to process over and over trying to find the "best" set. I chose to process as many as the app could handle and see what happened. It would probably yield similar results with 100 photos, maybe fewer, given the right selection of reference angles. (I took so many photos because after setting up the photoshoot, I made sure to get complete coverage, rather than only shoot the minimum and risk incomplete coverage. I also wanted to be able to reprocess as many photos as the next generation apps can handle.)

Only the "decimated" file is reduced resolution -- all the others have all the capture's triangles. The figure in the "Full Scale" versions of the obj and stl files are the same quality as in the (still unpublished) raw obj captures, aside from the light editing I mentioned in the description. But the unedited capture obj and phototextures include the museum gallery itself and are 800MB+ of data, and I'm having trouble uploading it.

Right now I don't have an effective way to upload the original photos or host them. For this one piece I have 3.3GB of JPGs and 15GB of the original RAW files and a very slow connection. It's going to take at least a couple days or more, but I'll post a link here when I get it sorted out.
Sorry it's not working. I wanted these to be easy to use. You might get good results with just a little infill. Maybe 5% or even less, and that might only be needed on the base part.
I hear what you are saying, but I've tried it yet again with no luck. Printing the base, at around 22 to 24mm print height, it simply attempts to air print the top of the base (the surface the foot sits on) and continues to fail through 58mm (the point at which I cancelled the print since it was obvious it wasn't going to succeed) with a tangled mess of filament. I am, as usual, totally confused.

But thanks for your reply.

And it is indeed a beautiful print, even if I can't accomplish it.

More from Sculptures

view more

Liked By

view all

Tags

License

Venus de Milo by CosmoWenman is licensed under the Creative Commons - Attribution license.

Give a Shout Out

If you print this Thing and display it in public proudly give attribution by printing and displaying this tag. Print Thing Tag
I am not
sure I liked the quality of the mesh and texture given 200 photos. I think the
same could have been achieved with 30-40 images. Could the author be so kind to
share original photos or at least obj files with the maximum resolution
possible? I am basically looking for a better texture. And I do remember a
promise to share source photos. Can you please share them as well?
It's possible to select and reselect which photos to process over and over trying to find the "best" set. I chose to process as many as the app could handle and see what happened. It would probably yield similar results with 100 photos, maybe fewer, given the right selection of reference angles. (I took so many photos because after setting up the photoshoot, I made sure to get complete coverage, rather than only shoot the minimum and risk incomplete coverage. I also wanted to be able to reprocess as many photos as the next generation apps can handle.)

Only the "decimated" file is reduced resolution -- all the others have all the capture's triangles. The figure in the "Full Scale" versions of the obj and stl files are the same quality as in the (still unpublished) raw obj captures, aside from the light editing I mentioned in the description. But the unedited capture obj and phototextures include the museum gallery itself and are 800MB+ of data, and I'm having trouble uploading it.

Right now I don't have an effective way to upload the original photos or host them. For this one piece I have 3.3GB of JPGs and 15GB of the original RAW files and a very slow connection. It's going to take at least a couple days or more, but I'll post a link here when I get it sorted out.
I tried slicing the "Head" pieces last night but found problems when I came to review the GCode produced. The issue was that printing would start in mid air on the second layer. That is, the STL model does not lie flat on the bed - same problem as encountered by gzumwalt but lower down the model.
Please see my comment above in reply to gzumwalt.
I'm not seeing that the "Head" piece is not level to the bed. and I'm not sure this is the same issue gzumwalt is having but I'll double check everything.
Does "hollow" mean "0% infill"?
Yes. I usually print with no infill, at 3 or 4 walls thick.
I cannot get it to print as per your instructions on a Replicator 2 using Makerware. It cannot be "hollow" since the top of the base requires some form of support to print. Also, the top of all of the sections cannot be printed "hollow" again without some sort of support. Any suggestions? Thanks.
I've downloaded all the sectioned stl files directly from this
"Thing" page. I re-sliced them all with ReplicatorG0400. I set it to no
rafts, no supports, 0% infill, 3 walls thick (total), .2mm layer height
for the body pieces and .12mm for the head/face/crown pieces. I got no
slicing errors, and saved them out to .s3g.

Using white PLA at 220C on blue tape, I've re-printed all the
head/face/crown pieces, and got level, flat, full cross-section
footprints on the build plate, and no skipped layers or air printing.
Re-printing the re-sliced "Venus1" stl (the base/feet), I got the same
results and the top horizontal surface of the statue's base (on which the
feet rest) printed fine (not perfect, but well supported and OK by me),
and the topmost surface at the part's crop line printed too. Again,
this is with 0% infill, only the cross-hatched threads RepG spans
between the perimeter to go under unsupported features that are
suspended over empty interior volume (which I assumed Makerware would
make too). These are all my normal settings that I've been using for all
my similar multi-part prints for a year and half with no problems.

As for slicing errors and level-to-buildplate/layer-skipping/air-print errors mentioned in the following comment by Alzibiff, if the cause is a problem with my stl files, I'm stumped as to what it is.

This all makes me wonder if you are either experiencing mechanical
issues, or if Makerware+Rep2 is slicing the files very differently than
my own ReplicatorG0400+Rep1 setup. It's possible that Makerware simply
requires infill, and maybe the pieces need to be reset level to the Makerware onscreen floor prior to slicing.

I
only recommend zero infill to save time and material, but you might
have better luck with different settings for these pieces.

However, I did double-check my notes and realize I got *better
looking* results just on the head/crown/face parts when I used 10%
infill, .12mm layer height and 4 walls thick. There's some caving-in on
the top of the head without infill and thicker walls on those, sorry.
I'll update the print recommendations above.
I hear what you are saying, but I've tried it yet again with no luck. Printing the base, at around 22 to 24mm print height, it simply attempts to air print the top of the base (the surface the foot sits on) and continues to fail through 58mm (the point at which I cancelled the print since it was obvious it wasn't going to succeed) with a tangled mess of filament. I am, as usual, totally confused.

But thanks for your reply.

And it is indeed a beautiful print, even if I can't accomplish it.
Sorry it's not working. I wanted these to be easy to use. You might get good results with just a little infill. Maybe 5% or even less, and that might only be needed on the base part.
I used ReplicatorG0040 to create gcode and s3g files, and printed on a Replicator1. Even with infill turned off prints internal crosshatching spandrels to support the flat surfaces. These are the same files I printed, twice each, and my normal settings which I've used to print hundreds of big prints, so I'm not sure what's going on. It's possible Makerware behaves differently with horizontal surfaces and needs infill.
As I feared, at the 22.2 mm level, the Replicator 2 is extruding plastic into thin air. I guess I'm doing something else wrong.
Thanks, I am beginning a print of the full scale version, and am unsure how the top of the base and the top of each section can print without infill for support. Keeping my fingers crossed.
As Greek people with ambition and knowledge about art and
mythology we want to mention that Venus is a Latin translation from the Greek Goddess
Afroditi. Afroditi is born our wonderful Greek Island Cyprus.
Top