|
|
|
|
|
| 1 | 2 | 3 | 4 |
| (1MB) |
Figure 1: A computer rendering of our first laser scan. The subject is Michelangelo's unfinished statue of the apostle Saint Matthew. The scan was acquired on January 26, 1999, the day we moved into the Galleria dell'Accademia. It consists of several vertical sweeps of our Cyberware scanner. The depth resolution is 0.1mm, the spacing between sample points is 0.29mm, and the width of the scan is about 30 centimeters. Here is a plot of the surface profile in the area of the cheek.
This rendering shows the range data acquired from only a single viewpoint, so the data has holes in it. Look for example just below the eyebrows or to the left of the nose. These holes will be filled later by scans taken from different viewpoints. The rendering is also not in color. Although we are acquiring color data, we have not yet processed this data for this particular scan. To make the rendering shown here, we treated the surface as if it were made of a gray, slightly shiny plastic. The shininess is not natural, but it helps us see Michelangelo's chisel marks.
Figure 2: It is now February 8, and most of the holes have been filled. However, we have still not connected the scans to form a single polygon mesh, nor have we added color, so the model is again rendered in shades of gray. Here are some statistics about the model as it now stands:
Figure 3: It's February 19, and we have finally connected together the scans for St. Matthew's head using Brian Curless's volumetric range image processing (vrip) software. The volume contained 1200 voxels on a side, making each voxel 0.25mm. The resulting mesh contains 3,800,000 vertices and 7,600,000 triangles. To see what a vripped mesh looks like, look at some meshes from our model of the David.
We have also processed our color photographs to compute a reflectance value for each vertex. ("Color" is what a camera sees; it depends on how the surface is lit. "Reflectance" is a property of the surface, independent of the lighting.) Back on January 26, when we scanned the statue, we also photographed its surface with our digital color camera. For each point on the surface, photographs were taken from several angles and under both the ambient lighting of the gallery and a calibrated spotlight. The spacing between these color samples was 0.125mm. To compute reflectance values, we first compensate for ambient lighting, then we discard pixels ruined by shadows or specular reflections, and finally we eliminate the dependence of our measured colors on surface orientation. This last step requires knowing the shape of the surface with great accuracy, including the shape of each chisel mark.
Once we've computed reflectances, we can generate new images with any lighting we please. In this example, we've placed a virtual spotlight above and to the right of the head. Remember - this is not a photograph of the statue; it is a rendering made from a three-dimensional computer model. In this example we tried to make the rendering look like the statue; we could as easily have made it look like bronze, or wood, or even glass.
Figure 4: It's the beginning of June, and after months of scaling up our software to handle larger and larger meshes, we can now vrip the entire statue. The resulting 3D model contains 386,488,573 triangles, our largest to date. We have not yet mapped our color data onto this model, so it again rendered with artificial surface shading (this time rather shiny). Warning - this image is large!
If you want to fly around our model of St. Matthew, but you don't qualify for a license to download the data, try ScanView: our secure client / server rendering system.
|
|
|
| 5 | 6 | 7 |
Figure 5: Another view of our vripped triangle mesh, with per-vertex colors (actually reflectances) computed from our photographs of the statue, as explained in figure 3 above.
Figure 6: An alternative coloring of the same mesh. To compute these colors, a probe sphere is (virtually) rolled around on the mesh. If the sphere cannot reach a particular vertex because the vertex lies in a valley, then the sphere is shrunk until it can reach it. Each vertex is then colored according to the size of the sphere that finally reaches it; darker means smaller. In essence, this algorithm is computing the accessibility of each point on the surface of the statue. This algorithm was invented by Gavin Miller. The rendering shown here was made by Szymon Rusinkiewicz.
Figure 7: What insight does this visualization provide? To us, it seems to show the structure of Michelangelo's chisel marks more clearly than the color image. In this annotated closeup, the red arrows denote a ridge where Michelangelo's chisel evidently exited from the surface tangentially. The yellow arrows denote a valley where his chisel dove into the stone and stopped. What other features can you find from this visualization? Can you think of other visualizations that might be useful? If so, send us email!
Notice: The images of Michelangelo's statues that appear on this web page are the property of the Digital Michelangelo Project and the Soprintendenza ai beni artistici e storici per le province di Firenze, Pistoia, e Prato. They may not be copied, downloaded and stored, forwarded, or reproduced in any form, including electronic forms such as email or the web, by any persons, regardless of purpose, without express written permission from the project director Marc Levoy. Any commerical use also requires written permission from the Soprintendenza.
Notizia: Questi modelli elaborati al computer, immagini computerizzate, e fotografiche sono proprietà del Progetto Digitale Michelangelo e la Soprintendenza Per I Beni Artistici e Storici per le Province di Firenze, Pistoia e Prato. Non possono essere copiati, scaricati da internet su un file, inviati, o riprodotti in nessuna forma, incluso la posta elettronica o il web, da nessuna persona per nessun motivo, senza un permesso scritto da Marc Levoy, il direttore del progetto. Eventuali usi commerciali esigono anche il permesso scritto dalla Soprintendenza.
