This might sound odd... but how is a function of utilizing and combining relational texts from photo "semantic thought representation" being offset for the need of uniqueness? The gibberish produced has to pass the copyscape type test at minimum. This does not yet seem useful for "human readability", so its use as a "spinner" still has the drawbacks of other spinning technologies that need to be addressed. As new ideas are not sought out, only remaining semantic connections, might not an ontology based structured triple graphing technology be more suitable?