<![CDATA[Traditional Batik production relies on skilled artisans to plan multistage dyeing sequences that yield culturally meaningful and visually harmonious color combinations. This manual practice is time-consuming, difficult to document, and hard to scale to novice designers, and prior work on Batik color design has rarely treated the dyeing process explicitly as a learnable temporal sequence. This study addresses that gap by modeling Batik dyeing as a sequence learning problem and applying a Recurrent Neural Network with Long Short-Term Memory (RNN–LSTM) to support automated color design. We construct a dataset of 72 fabric samples obtained from single- and two-color dyeing procedures that follow traditional Batik wax–dye–dewax workflows. For each sample, RGB values are extracted at each dyeing stage and encoded as time-ordered inputs, while the final fabric colors are used as target outputs. The proposed RNN–LSTM learns to predict harmonious color sequences that are consistent with examples in the dataset. It achieves a prediction accuracy of 0.869 on held-out data, outperforming several feedforward and recurrent neural network baselines under the same training protocol. An interactive simulation interface then integrates the trained model, allowing users to explore and visualize predicted color outcomes step-by-step. The results show how AI-based sequence modeling can help preserve Batik color traditions while making expert color design strategies more accessible.]]>
