This paper applies a texture model regression method to characterize seismic facies with particular reference to frontier, deep marine depositional settings. The algorithm introduces and designs a seismic texture model in an adaptive manner in terms of dimensionality, size, amplitude, frequency, and phase using mathematical, synthetic, or actual seismic data. The model serves as a reference for seismic feature discrimination by comparing actual seismic texture at each location with the model through a linear, least-squares regression analysis.