This is a sneak peek of our development of a new full body scanning system.
The three-dimensional shape of the body is reconstructed departing only from a few images taken with a custom photographic system. The video shows the iterative output of the reconstruction algorithm. The exact specifications and details of the capture system are yet to be published, but the design priorities are the following:
- Portable system. No need for a full booth nor rig structure. All equipment fits in a briefcase, can be installed in any ordinary room and adjusted in a few minutes.
- Reduced equipment costs. Using just a few inexpensive cameras and a controlled illumination system.
- Capture time is irrelevant since all poses are typically shot in a fraction of a second, depending on lighting conditions and capture settings. Moderate involuntary movements are taken into account.
- Incremental level of detail. Basic shapes can be obtained faster and using fewer images. Detailed shapes require additional shots and take further processing time until reaching the desired level of detail.
- Quick reconstruction time. Each refining iteration takes now ~3 seconds in an ordinary personal computer.
- Textured model. Variations in skin tone/color can be captured for the naked body. No need to wear a custom suit, whereas tight underwear can be still used to reduce intimacy concerns.
The reconstruction is achieved just from the images by a mixture of photogrammetric triangulation/calibration, photometric stereo, and visual hull reconstruction methods. This hybrid technique is inherited from recent developments in cultural heritage digitization, but specialized here for full body capture, with two key additions:
Firstly the reconstruction algorithm is aware of the specific human body articulation model by means of an specifically designed deformation mesh. This optimizes 3D reconstruction processes and allows to track moderate body movements in multiple images. Secondly a BSDF (Bidirectional scattering distribution function) model is used for the skin and/or some fabrics. This provides fine details containing valuable information of bone, muscle and fat distribution right under the skin and improving reconstruction robustness reducing capture requirements.
This project is now in late development stages, entering soon into evaluation and validation tests to compare with other existing technologies and to be optimized for specific applications in different fields such as medicine, sports, clothing or arts.
Advison, February 2013