The final cleavage leaves , which remains covalently bound to the target surface.

The most famous implementation of C3D was proposed by Du Tran et al. (2015) in their paper "Learning Spatiotemporal Features with 3D Convolutional Networks."