SDXL
SDXL
Meet the revolution! v0.3 is here750+ artist styles, few general stylesHundreds of characters, just prompt your waifuOutperforms NAI3 in (fluffy) tail conceptsAnything from 1girl_standing_masterpiece to hardcore stuffLora for character, 2x lora for style, 3x tweaker lora... - leave this madness for sd1.5, all you need is here, just prompt. Nevertheless, the model is fully compatible with PD loras.This is finetune of Pony Diffusion V6 XL, just pure and full-scale training without any merges of loras/other checkpoints. With new large dataset (~0.4M pictures with special tagging) it gives great base aesthetic, lots of new (or originaly hidden) knowledge, better prompt control, backgrounds, etc. maintaining anatomy and nsfw capabilities of the original model.v0.3 comes with more styles, characters, concepts, and lots of general improvements.Various kemonomimi, gacha, touhous, vtubers that have been originally removed/hashed in pony. Also in this version styles work more smoothly without "burning" and character outfits can be changed much easier.In addition to artists styles, you can control or tweak it using following general styles:2.5d, bold line, cgi, digital painting, flat colors, minimalistic, ink style, oil style, pastel stylecan by used in combinations and both in positive and negative prompt.Quality classification has been applied withmasterpiece, best quality, low quality, worst quality tags. Score is based on specially designed classifier system and user ratings (where accessible). It just gives better result for new training then adjusting original score_x_up (they still work).Lots of tail concepts:tail censor, holding own tail, hugging own tail, holding another's tail, tail grab, tail around own leg, tail around penis, tail through clothes, tail under clothes ...and many others with natural text. Some works perfectly, some requires rolling, some can produce unwanted/enjoyable mindfucks.Prompting:by ARTISTNAME, [by ARTISTNAME2, by ARTISTNAME3,...]start with on artist style(s) if needed, use weights to adjust strength, like (by gweda:0.5).Then add quality tagsbest quality, masterpiece, score_9, score_8_uptags like source_anime, uncensored, spamming of score_x_up and others can give unwanted bias. General style tags from above can be also applyed herethen comes main prompt with booru tags and/or natural text.Basic negative:bad anatomy, bad hands, error, missing fingers, extra digit, fewer digits, (worst quality, low quality:1.1), jpeg artifacts, signature, watermark, usernameAvoid unnecessary tags like grayscale, monochrome, 3d, source_pony etc. unless you know what are you doing. Base saturation and contrast are already fine as is, for more flat anime style better use "flat colors and/or minimalistic" in positive (with desired weight) or some artist style.Backgrounds can be improved simply by adding simple background, blurry background, abstract background to negative prompt, but beware of overusing it.Known issues (to be fixed):Random watermarks, less then in v0.2 but still, most comes from base modeSome rare artists and characters may be barely recognizable, please report if you find anyStill WIP, some issued may be discovered.Requests for artists/characters in future models are open. Follow for a new versions.Leave link to original if using in merges, behavior of the model is easily recognizable.Special thanks to my fellow brothers who share same cultural beliefs and helped to make it real.AI is my hobby, so I'm wasting money on it and not begging for donations. If you want to support - share my models, add feedback, make a cute picture with fox/cat-girl. And of course, support original artists.