Discharge phenomenon such as lightning occurs when excess charge flows inside the insulator, and the damage marks are called Lichtenberg Figures. In Computer Graphics, the expression of lightning is mainly drawn directly by the designer, and the procedural generation method is not widely used. However, considering the implementation cost in CG, it is desirable that the designer interactively manipulate the shape and generate it procedurally in a short time. Then, in order to focus on the shape of the damage mark of lightning, we first try to generate a Lichtenberg figure and seek a method that can be widely applied to dielectric breakdown. 

A related work is Diffusion-limited aggregation (DLA), which can express the growth process of clusters, but this does not take electrical properties into consideration, and it is difficult to generate shapes instantaneously. There is also the Dielectric Breakdown Model (DBM), which is a model that assumes that dielectric breakdown progresses with a probability. This reflects the electrical property, but it is difficult to extract a concrete and explicit route, and it is unsuitable when you try to develop CG expression by using new route information. 

Therefore, in this research, we try to create a procedural generation of insulation breakdown traces that the user can operate interactively, considering the electrical properties. By constructing a model that is more specialized in CG expression than existing research examples, we realize an application that is ready for users to use and that saves time and cost. 

We think much of directability, that is, expertise that enables the systems to be intentionally controlled. In this method, by combining the parameters set in advance and the information that the user operates at the time of execution, we try to generate shapes interactively.

Positioning on Reality Modeling Triangle

Since this research is specialized in CG expressions and user-friendliness, it can be said that it has a strong Artificial element. In contrast, it is not Phenomenological because it does not use actual data. In addition, it can be said that it is far from Physical because it includes many approximations in shape generation.


NameAffiliationWeb site
Ryosuke NishidaKeio University
Masanori NakayamaKeio University


Domestic presentations

  1. Ryosuke Nishida, Masanori Nakayama, Issei Fujishiro: “Generation of 3D Lichtenberg figures considering spatial characteristics,”  in Proceedings of the 82th National Convention of International Processing Society of Japan, Vol. 4, pp. 119―120 (1ZC-05), Kanazawa Institute of Techn, Ishikawa, March 5―7, 2020 (in Japanese).  Student Encouragement Award [IPSJ Digital Library]


  1. Grant-in-Aid for Scientific Research (A): 19H05576 (2019―)

Back to RM team page