AI-Stories

Hi Everyone, I am really really excited after trying out ControlNet for 2 days now the results are amazing !!!! ... i can set poses and nearly get them Nearly Same pose different settings also it works perfectly with other non fighting , this this very neat images also i can change the mood while keeping nearly the same pose ;) what and amazing software to use !!! #AI #News #ConrolNet #stable-diffusion

Howdy, this is my second installment of this lesson plan, I will continue to check the poll and keep making these so long as a top secret ratio threshold is reached/maintained given a certain number of votes. So if you like these, vote, if you don't, then vote no. This will be a more broad set of tips, teaching the things that go overlooked. So to start I want to explain some things. CFG is effectively how heavily you want the prompt to effect the outcome, this is basic so I don't feel the need to explain further. The next to piggy back off CFG is the prompt quality. Naturally the AI model will be able to associate a lot of words with other words and get a better understanding of what you want. BUT because it uses word association, on some prompts it may be better to be exact and descriptive. Most prompts will work fine with a prompt generated like you are describing a scene to a friend, ie "A photo of a fat potato on a couch". Then you can add descriptors, like HD, highres, 4k, UHD,

Hello everyone, I have for you today, a brief guide to help you with your AI prompts. In order to get the best result, its important to understand what data your AI model was trained on. That becomes easy if you did the training yourself. However, not everyone has the luxury or skillset or time to do such a thing. So the first thing I want to go over is this, most AI art will be a derivative of SD 1.5, that is broadly true for most models. Thusly, they will generally have similar weights(the general sway a word has towards the final end image result). Because these AI art works this way, generating anything with very specific exceptions or contradictions is nearly impossible. An example being, using "nude" and "brown pants" in the same prompt, This doesn't mean it won't get you an image with that, but it means you will likely get more bad image results before you get what you were looking for. Instead, using a prompt with wording like "topless" can be more straightforward. For