Abstract: Batik is one of cultural heritage in Indonesia and it consists of various patterns and objects. Most of these objects are plant. By conventional way, batik pattern is generated manually. So, the batik design generating rate is slow. By computational technology, batik pattern can be generated faster and more various. In this research, we propose the new and simpler Dendrobium modeling, so that, it can be generated as batik pattern. Dendrobium is one of orchid plant genus with huge number of species. In this research, we use random walk method as a stochastic approach, so that, there is uncertainty pattern and path in the result. We model all parts of the orchid plant includes: root, stem, leaf, pedicel, peduncle and flower. By using this proposed model, the Dendrobium based batik pattern is successfully generated. In this research, traditional Kawung pattern is used as background pattern. In this research, there are three tests to evaluate the relation between the adjusted variables and the observed variables. In the first test, we observe the relation between the number of stem segments and the number of leaves. In the second test, we observe the relation between the flowering ratio and the number of peduncles. In the third test, we observe the relation between the number of peduncle segments and the number of flowers. Based on the tests in all tests, the relation between the adjusted variable and the observed variable is positive linear.