ICTree: Automatic Perceptual Metric for Tree Models

ICTree implements the metric published under the same name at the international SIGGRAPH Asia 2021 conference. Its primary goal is to enable the user to automatically evaluate virtual tree models from the point of their visual perception. The input tree can be provided in one of several formats, including polygonal model, tree skeleton, or the native format used by the SpeedTree software. The output produced by the software is a perceptual score that reflects how realistically a human observer would perceive the tree. Compared to the standard approach, which requires interaction with the human subject, the main advantage of ICTree is its immediate feedback. This feedback is essential in several downstream applications, which include, e.g., the comparison of existing methods. Further potential use is also the building of more sophisticated botanical tree generators, which produce tree models based on their perceived level of realism.

For more information, links to supplementary materials, and codes, please visit the project website: http://cphoto.fit.vutbr.cz/ictree/