Why not semiotics?

The process of reading an image is wholistic, not linear - the component parts only makes sense in relation to the whole (hermeneutic?) So how can you conduct a semiotic analysis (breaking it into a 'grammar' of compositional parts) to an image?

How do you analyse a found image, something that hasn't been consciously composed for the context you put it in, using a semiotic model?

Further to why I'm not focusing on semiotics, a quote from Kate Sweetapple's PhD thesis:

"Although the application of semiotic theory to the field of design has enabled a greater understanding of how meaning is produced in visual communication it does not account for how the designer affects the type of engagement the viewer has with the material, which is a significant aspect of the communication process. The absence of such an understanding results in designers having limited control over the viewer response to their messages, which in turn compromises the intended viewer experience."

Semiotic analysis interprets existing images by coding/decoding the elements (signs) within those images. It does not (to my admittedly limited understanding) account for the process of creation; the intention of the creator to affect the way the viewer/reader responds to the designed outcome (image or document).

Kate's thesis develops a model based on literary theory and the notion of 'distance' to identify four types of visual narrators (idiosyncratic, implicit, imperative and esoteric) and provide a new understanding of the relationship between the designer and their visual outcome. She provides "a theorised model of practice, and a method of visual analysis".