Woot Art Hard Shading Oekaki Style

This model generates the Woot art style, specifically older works such as those found here https://woot.iiichan.net/ I limited the data to just those with a "hard shading" style without dithering, so it should be more specific. If you want something like Woot's more recent work go use https://civitai.com/models/805540/w-oo-t-style-ponyxl instead (which despite the link has an Illustrious version).Trigger word is ws drawing for "woot style drawing" as this has less existing training than "woot".Long detailed captioning is recommended. Repeating ws drawing is recommended. Capping your prompt off with BREAK ws drawing is also recommended for style fidelity reasons. Otherwise just describe your image with extensive tags. 1girl and/or 1boy (with the correct number of characters) is also recommended as the model was trained on this.The model has a strong bias towards very round ladies. The model was trained on the Danbooru breast size tags, with a bias towards bigger, but it's recommended to specify this.Works best at the 1024x1024 resolution (or smaller) due to the dataset. Hi res fix to 1536x1536 works well but you won't gain much unless you're using a nice finetune.Recommended strength is 1.0, but especially with finetunes it can easily go up to at least 1.5 and give more style fidelity. See what works for you.Other helper words:oekaki - while untrained, this is an oekaki style and Illustrious knows the oekaki style so this will make it look more like a woot oekaki, but also it's a sketchy style so maybe you don't want that. Can also be good as a negative prompt, or even both positive and negative.jaggy lines - this one is trained on a few of the images with more jaggy lines and it seems to reduce the jaggy line effect when used as a negative prompt. Not normally needed, especially if you're using a finetune base model.pixel art - also untrained, while it seems like it would work I personally didn't like the results, tends to make extremely jaggy lines that outline pixels. You could probably make this work with better pixel art models/LoRAs.