As you have rightly observed, the translation of multimodal texts is difficult because they combine two or more semiotic systems. Notably, in modern multimodal environment, the translators have to go beyond the linguistic level by considering macro-level characteristics such as socio cultural, ideological and pragmatic aspects of text exploiting various semiotic resources contributing to the message created by ST writer. For more information, I refer you to the following texts, which can hopefully provide you with what you are looking for.