Training-free Color-Style Disentanglement for Constrained Text-to-Image Synthesis | Read Paper on Bytez