Metrics of A Hug
RGB/Thermal Photogrammetry, Python, OpenCV, Cinema4D
Carnegie Mellon University
2020
Metrics of a Hug is an investigation on how to measure our affective footprint on other beings and spaces using RGB and thermal imagery, computer vision, and photogrammetry.
The capture system employs a DSLR-camera with a fixed lens and an Axis Q19-E surveillance thermal camera (connected as IP cameras to TouchDesigner) that simultaneously captured an RGB and a thermal map of a person after receiving a hug. After, the images were aligned using a Python and OpenCV algorithm that computed the homography matrix needed to adapt the perspective of one image into the other. The system was previously calibrated using a perforated dotted grid pattern visible by both the RGB and the thermal camera which allowed to determine each image's deformation.
Using high-contrast markers and a spinning platform I was able to capture enough frames to perform a 3d photogrammetry of the 'hugged' individual. After a hug, the thermal load stored in the clothing's fabric was easily visible with the thermal camera which images were used as the final texture of the reconstructed hug model in Cinema4D.
• Project supported by the Frank-Ratchye STUDIO for Creative Inquiry.