Visual novels (VNs) were invented in 1982 because it was discovered that early text adventures didn’t compress Japanese kanji enough to fit on the low-memory computers in use back then. It’s been a full-blown thing in Japan since the late 1980s, but it was only about 15 years ago that English-language visual novels started becoming common in the last decade, when English-language tools became accessible enough for indie developers to make games they felt confident releasing commercially.
Visual novellists prefer, from what I’ve seen, Ren’Py, NaniNovel (a plug-in for Unity 3D), Visual Novel Maker (a plug-in for GameMaker), Visual Novel Maker, TyranoBuilder, Ethornell, Buriko General Interpreter, KAG, Ren-engine, Kirikiri, Novellate and Twine (the latter may surprise some people) - a lot of tools have appeared in the last five years to challenge the longer-standing Ren’Py. They’re not in any particular order, but I see the first three (as well as people building visual novels in plain Unity and GameMaker) more often than any other options available to the general public - especially for small indie commercial games (as distinct from free VNs, which can use any of the above and others). Larger commercial concerns often use significantly-modified versions of multiple of these options, mixing and matching according to the strengths of each platform. (Bonus for them: harder to pirate a “hybrid” creation than one created entirely in one platform).
I regard visual novels as part of the IF category, as a particular way of telling a work of fiction in a way that facilitates interaction but keeps the emphasis on the story more than other elements of gameplay. Some people classify IF more strictly and would disagree with that definition - I’ll let them state how they’d distinguish visual novels and IF.
Text adventures are different to visual novels because they rely on text to tell the story. (Some text adventures have illustrative/augmentary graphics, but if graphics are the main way the story is told, it’s not a text adventure but something else). Updates to these graphics are generally because the player did something that would logically cause that change (the castle image appears because you moved there and the compass rose shows you can’t continue north because the moat bridge is up and you’ve not solved the puzzle to bring it down yet). Traditionally, there is a parser which accepts text (either from a keyboard - physical or on-screen - or speech-to-text), although some text adventures have experimented with the form. There is also typically a focus on puzzles and/or high-concept exploration, although the last 20 years have brought a wide variety of ways to make text adventures interesting.
Visual novels, on the other hand, typically have the graphics doing a lot more work in telling the story - although rarely does a visual novel have anything in the pictures that is not at least implied in the text. The use of graphics typically resembles a cut scene from other sorts of game, but with simplified animation (a nod to 1980s computer limitations that is sometimes subverted these days). These graphics most often change simply because the player clicked to see the next part of the story, rather than because the player took a specific action (the castle appears because you were talking with another character on the road and the game assumed you were going to the castle, and the compass rose probably isn’t there because you have to read another 10 lines of dialogue and make 2 choices before you’ll next move location anyway, 1 of which determines if the drawbridge will be lowered).
Traditionally, there’s a limited number of options each time there is something for the player to do (assuming the visual novel has any - unlike text adventures, some visual novels are strictly read-only). All the ones the player can choose will be spelt out right there on the screen, and players choose the one they want using arrow keys, a mouse or their finger. One result is that conversation-as-puzzle becomes easier to code - and convincing, consistent object-based puzzles, depending on the tool used to create the visual novel, may be more difficult. Hence why, in my example, the drawbridge puzzle became a conversation choice. Unsurprisingly, this suits genres of game where conversation is expected to form the bulk of the game - dating sims being the classic example.