Raw LLM Responses

Lavender found the new cash flow for the moment, get that bag. Even if I disagree, with a lot takes or the stupid uninformed comments both AI and non-AI users are saying here lol. Including thinking Nightshade or filters like that actually work. Unfortunately, they don't. I wish they did. Generation LLM is a tool. always will be. Anyone trying to make it out to be anything but, is selling you snake oil or hyping it up to be something it isn't. The best you can do is control the LLM - which in the right hands is a boon, useful for artists or designers who use this new tool to their advantage. Which for the 0.1% using this stuff for faster and better results already know this, and probably aren't even watching this video lol. Not everyone is super passionate about repeating the same process in every little detail and would rather have an AI repeat it instead after proving to themselves that they can do parts of mundane work. Whether people like it or not, mundane can exist in the very craft many people enjoy. Whether that's traditional or digital, 2D or 3D. It's pretentious to pretend that all artists enjoy the entire craft at the same level. The craft is about self expression, taking shortcuts to build your own work up is valid if you know what you're doing and for what purpose. All that said, never use AI by itself and try to pass it off as your own, or barely change it and proclaim the same. Be smart and be derivative with it. Don't use it as a crutch, don't use it as a shortcut without deep knowledge of the subject and end result. Otherwise you'll end up like the people who use 'ART shortcuts' and lost in the sauce when you truly have to figure something out yourself. Or at worst people will see you're using AI as assistance. EDIT: Jesus Christ I write a lot of word vomit. If anyone ends up reading this and wants to know why Glaze or Nightshade doesn't function? It's because they were designed to work against VERY specific versions of LLM's (SD-XL) and how they functioned. Well, as nice as that is - we're many versions past where these sensationalist anti-AI buzzword filters 'worked'. It didn't even modify working models before it became widespread. This means the people behind these tools would have to not only adapt these filters or changes to new models (Ironically using LLMs against LLMs) but also expect the training process and algorithms used to stay relatively the same. They probably knew this but wanted to see if it would theoretically work anyways. It does, but only on those specific backends. And if those backends never update. Or use poisoned images as a no-no and avoid that alteration.