CAMILA: Context-Aware Masking for Image Editing with Language Alignment | Read Paper on Bytez