Pietà Project - Results



Synthetic images obtained from our model using both geometric and photometric data (Click on images for larger size.)

Our primary goals for the Pietà project were defined by Dr. Wasserman's research questions, and our presentation of the results was shaped to fit his needs. The results we delivered to him were various combinations of:

Precisely-defined views
Impossible views
Imbedding the statue in virtual environments
Precise measurements
Modifications
Interactive viewer

In order to answer certain questions about Michelangelo's composition, Dr. Wasserman needed to see the statue from points-of-view that are impossible in reality. These included views from directly above the statue, to reveal details of the composition not normally visible; and from various angles at a height several meters below the base of the statue, to illustrate the statue as it would have appeared used as Michelangelo originally intended.

The Pietà currently stands on a small pedestal in a rather small room in a museum. Michelangelo intended it to stand in a niche over his tomb. Over its history, it was displayed under the huge vault of the Florence Duomo and in a garden at the Bandini palazzo. We re-created some of these settings with 3D models and produced images and animations illustrating the visual impact of the statue in these various environments.

Measuring the distance between points of a real statue can be difficult: The statue itself can interfere with a precise measurement. In order to analyze Michelangelo's use of proportion, Dr. Wasserman needed to know the lengths of limbs and the dimensions of torsos. The ability to modify the statue, impossible in reality, is particularly compelling for the art historian, especially for a statue with such a complex history.

Using the 3D model, we re-constructed the statue with Christ's missing left leg, approximating its appearance before Michelangelo broke it. We removed the pieces that Calcagni reattached, illustrating the statue as it would have appeared without his efforts.

We separated the four figures that make up the statue so that they can be examined in isolation.

To enable Dr. Wasserman to study the statue at his leisure, we developed an interactive viewer tuned to the specific requirements of the project. The final model is far too large for interactive display even on today's best graphics hardware. On Dr. Wasserman's laptop, it was inconceivable. Therefore, we are developing a hybrid viewer that provides interactive examination of a geometrically-simplified version of the model; and a detail viewer that provides more limited interaction with a selected high-resolution image.

The simplified model acts as a kind of map of the more detailed image. By enhancing the geometry with simplified textures, we discovered that a model of about 10% of the original triangles was sufficient to provide the necessary high-level positioning of the viewpoint. The user can quickly select an area of interest, using simple navigation controls to reach the desired view. Dr. Wasserman came to us as a very inexperienced computer user. Using a mouse and keyboard to navigate in three dimensions is non-trivial even for an expert; for this project, we implemented an extremely simple interface: Rotation and zooming are defined relative to a center point that the user can reposition by a simple point-and-click operation. Translation is available by click-and-drag, but rotation and zooming are controlled through buttons that can be clicked or held down for continuous motion. A home button guarantees that the user can always recover a familiar viewpoint.

When he wants to examine a region in more detail, the user frames a section of the scene and invokes an offscreen renderer (perhaps on a remote server) that creates the desired image from a database that represents the full-detail model. The 2D image is enhanced by the addition of a surface normal at each pixel, allowing the lighting to be re-computed in real time. The user can drag the mouse across the image to move the virtual light from one angle to another. This technique was designed to model a method that we observed Dr. Wasserman using on the statue: He moved a small flashlight slowly back and forth across the statue to highlight small surface irregularities. The virtual light editor produces similar effects.