It's high fidelity because it's closely matching the visual and interaction design the final product is likely to have. How closely depends on how much work you put into it and your workplace's standards. I've heard a rumor that Apple always makes 10 "final" products and discards all but the "best", but more generally high-fidelity prototypes are meant to incorporate everything that compromises the final product at that stage. The visual design is a big part of this, especially to differentiate it from low fidelity prototypes.
However, visual design isn't necessarily concerned with practicality, especially prototype-level practicality. Your prototype might not be 100% there because some programming/animations/art assets don't exist yet or are too hard to put together for what's not actually a final product. It's still a prototype, so the visual design isn't 100% represented in all cases.
I don't think you can really "bring changes" to a prototype to make it a visual design. You can make a prototype into a final product, which then has the visual design, or use the prototype as an example of the visual design, but you shouldn't be thinking them as the same entity.
High fidelity prototypes are dangerous exactly because people think they are the final product and they start to assume things can't be changed; test users will be reluctant to mention high-level problems with the design. Your team will likely be reluctant to change them as well; you put a lot of work into that! Isn't it ready to ship yet?
It's important to keep a layer of abstraction here; you can kill your prototype at any time to refine it, and that doesn't mean you're throwing away your visual design, you're just improving it. Ditto for interaction design. Your prototype might go on to look 100% exactly like your final product, but never assume it is.