Omni-Extractor: A Disentangled Texture Reconstruction Protocol

AbstractOmni-Extractor introduces a deterministic GPU-based method for disentangling texture information into structural and state components.By separating a 16-bit topology map from a 1D state atlas, the protocol achieves up to 80% VRAM reduction while preserving perceptual fidelity. Built as a modular pipeline inside ComfyUI (used here not for AI inference, but as a CUDA-based node graph interface), Omni-Extractor reconstructs textures in real-time directly on the GPU. The paper presents the concept, implementation, and evaluation of this method, demonstrating how deterministic signal separation can outperform traditional RGBA texture formats in both efficiency and clarity. This work aims to inspire new generations of visual computing systems and future semantic compression standards.