Ostwald color solid constructed by blending eight primary‐color fibres and its color prediction system

To address the challenge of lacking a conversion mechanism between color values and primary‐color fibre mixing concentrations in textile blending applications of the Ostwald color solid, this study constructs a complete data transfer chain from eight primary‐color fibres to physical color models. First, based on the double‐cone topological structure of the Ostwald color solid, a full‐color‐gamut grid point array model containing 1352 grid points is established. By anchoring grid point serial numbers, a point‐to‐point mapping between the grid point array and the mixing concentration model is built. On this basis, using eight polyester fibres (red, yellow, green, cyan, blue, magenta, white and black) as primary colors, the Ostwald color solid is decomposed into six color matching sub‐models (each containing 232 grid points), and three types of physical color models (linear mapping, measured and predicted) are constructed. Then, based on the Friele color prediction algorithm, each color matching sub‐model is taken as a unit, and an assignment iteration approach is used to solve the prediction model parameters for each training sample at each wavelength, obtaining a parameter matrix consisting of 6 × 71 × 31 parameters. A bidirectional prediction algorithm is established, including forward prediction (mixing concentration → color value) and reverse prediction (color value → mixing concentration). Finally, a Friele full‐color‐gamut color prediction system consisting of a basic data module, a prediction module, an input module and an output module, along with its evaluation system, was developed. Experimental results show that the forward prediction average color difference for training samples is 0.85, and the reverse prediction average color difference is 0.76; the average color difference for reflectance prediction of non‐grid point samples is 1.32, and that for mixing concentration prediction is 0.93. The vast majority of the predicted color differences fall within the industrial tolerance range (<1.5), indicating that the color data transfer chain from the Ostwald color solid to the physical color models established in this study has potential for application in engineering production. This provides a systematic technical solution for precise color matching in digital colored spun yarn. However, a few samples with relatively large prediction errors should not be overlooked, and further calibration is still required for color‐difference‐sensitive scenarios.