First thing first, yes, you will have to pay for it, it's what it is, second, a quick tutorial: 1 . Go to https://novelai.net/ 2 . On the top right, click the on "Login" 3 . Now you can click "Sign Up" 4 . Create an account (use of a reliable burner like protonmail is advised)

>>11780 you already have a thread >>11282

5 . You can now go back to step 3 and connect yourself 6. You are now connected, you will see a menu and plenty of stuff that you won't care about (for now) so just click on the "Image Generator"option 7 . You will now see where the magic happen, first you need to select the model that shan't be advertised using the drop down menu, second you will need to prompt things you want to see in the "Prompt" tab while also filling the "Undesired content" tab with things that are too degenerate for our chaste eyes to witness like hand holding (I know you're a bunch of reprobates)

>>11782 This thread isn't in competition with the other AI thread, This one is more of a tutorial to explain to people how things works and what resources to use, also and much more importantly to share prompts and tricks between users rather than talking about the "grease monkey" stuff and sharing the end results. I know AI can be a daunting task and there's many facets to it. Speaking of prompt, where to find them and in what order are you supposed to use them? Danbooru is a solid way to start, it contain most of the vanilla tags that most if any LORAS/Stable Diffusion models will use https://danbooru.donmai.us/wiki_pages/tag_group%3Anudity https://danbooru.donmai.us/wiki_pages/tag_group%3Asex_objects https://danbooru.donmai.us/wiki_pages/tag_group%3Asex_acts https://danbooru.donmai.us/wiki_pages/tag_group%3Aattire https://danbooru.donmai.us/wiki_pages/tag_groups

Now, using those "Prompt": monster girl encyclopedia, species: lamia, inverted nipples, bulging breast, detailled background, scary, creepy, detailed, and those "Undesired Content" "futa, snout, gay, homo, bisexual" I was able to generate this, it's not perfect (this isn't the point, not yet). In the next part of the tutorial I'm going to explain the importance of {} and []

To generate an image, an AI have to create a base drawing, a primordial soup if you prefer and following the list of prompt you did input it's going to add layers of things you asked for, so the most important prompt is always the first one and the least important one is always the last. It's important to note than when you are starting, it's important to separate each tag with a "," to not confuse the model. Also, every time you put a prompt in {} like {monster girl encyclopedia} you are ordering the model a 5% priority focus on that prompt, on the other hand if you put something in [] like [inverted nipples] the model will give that prompt 5 less importance. Note, there's no limit on how many bracket you can input, it's possible to put {monster girl encyclopedia} as a first prompt and with 10x curly brackets like so: {{{{{{{{{{monster girl encyclopedia}}}}}}}}}} this would ensue that the model will do everything in it's power to create something from the MGE universe before anything else is created and only then add the other prompt you asked for. Here's a concrete example, because I modified the brackets, the end results is different, here's the list of "Prompt" that I used this time: {{{{{{{{{{monster girl encyclopedia}}}}}}}}}}, species:lamia, inverted nipples, bulging breast, detailled background, [[[scary]]], [[[creepy]]], {{{detailed}}}, The lovely Lamia still isn't perfect, but she's less aggressive looking than the previous one and the forest a bit more welcoming.

Now, what if you generated something to your liking, but it's not "quite right" or if you like a certain pose or design but don't want to have to beg RNGesus to bless you with the deviant stuff that keep you getting out of bed? Simple, 8 . you will use the "Add base image (optional) and upload the image you just generated 9 . You will select the "Image2Image" option, this will guarantee that instead of working with a blank canvas, what you just uploaded is going to be used as the baseline 10 . You now have access to two sliders, one of "Strength" which dicate how much the model will try to follow the uploaded image and "Noise" that will dicate how smooth the end results will be.

11 . As you can see, a strenght of 0.1 barely changed the "Image2Image" 12 . Meanwhile a strenght of 0.9 might have kept the general idea of the drawing but did some significant changes That is all for now, at a later date I will explain how to create pictures involving sexual predicament and why using "," between prompts isn't necessarily a good idea, at the end of the day it is still an IA and it can be quite lazy so some finesse is required.

>>11788 >11 . As you can see, a strenght of 0.1 barely changed the "Image2Image" I see zero changes, are you sure you uploaded the new image?