WARNING: Negatives often breaks the style and diminish the quality, even a single word like watermark, read to know how works the model, but as is based in previous version works similar.If you still don't wanna read, here it is a good starting point:<img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/76361458-b91f-4d80-9b80-cea3ca69b630/width=525/76361458-b91f-4d80-9b80-cea3ca69b630.jpeg" />FOALIIPOSITIVE QUALITY BOOSTS (add at the beggining of the promt)Nothing(It means don't add anything, just prompt, it does more cartoon and flat colors, don't understimate it, it can do amazing images specially if your prompt is long)score_9, score_8_up, score_7_up, score_6_up, score_5_up, score_4_up,Standard quality9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, score_9, score_8_up, score_7_up, score_6_up, score_5_up, score_4_up,Idk why but adding nines boosts the quality, seriously, try it you will be suprised, you can also add just nines, for each 9 it does different results, also you can pt the score at beggining or at end and it does different results.score_9, score_8_up, score_7_up, score_6_up, score_5_up, score_4_up, score_9, score_8_up, score_7_up, score_6_up, score_5_up, score_4_up,score_9, score_8_up, score_7_up, score_6_up, score_5_up, score_4_up,score_9, score_8_up, score_7_up, score_6_up, score_5_up, score_4_up,Repeting the scores it also boots the quality, you can repeat x2,x3 or whatever you want9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, score_9,It does similar to nothing but boosts the quality a bit, also looses cartoon and probably tends to do more loliNEGATIVESI recomend to no put any negative, often breaks the style and diminish the quality, though is good to do still more cartoon, also 3D, painting, works, but i warn you, it will do different styles.Scores and ratings works like pony.CFG/STEPS (depends a lot of the lenght o the prompt)5/25 is what works best i thinkBut the model can go as low as 1/8 and as high as 15/100Generally it benefits of low cfg/steps as is anime based.As a advice i would move between 15-35 Steps and 3-8 CGIAnother thing that works is lock at 25 steps and put higher cfg, is the cfg that adds the details! (if you prompted them)RESOLUTIONIs better and faster put higher resolutions without high res, than lower resolutions and high resBut these are the best resolutions for me:512*768+highres x2, 800*1200, 1024*1024, 1024*1280, 1024*1364, 1024*1424, 1280*1280The model as based in pony, works better with tags, it likes long lenght prompts , still you can use descriptions cos is merged with SDXL, but i recomend tags.Stye loras may not work as intended, some words may have more weight than others, adjust it.Finally, to create these models is lot of hours drinking coffe that i give for free, if you wanna support me, or want me to continue, consider buy me one!: <a target="_blank" rel="ugc" href="https://ko-fi.com/xdyexd">Kofi</a>FOAL II tried to tame the pony but the pony kicked meThis was a personal use model but i decided to share. Its a finetuned merge, the goal of this model was create an anime model without negatives and color corrected but i think i failed, pony is untamable, so this is what it is, a model to do anime right off the bat. Most of the weight of the model is from <a target="_blank" rel="ugc" href="https://civarchive.com/models/288584?modelVersionId=324692">AutismPDO</a> a bit of <a target="_blank" rel="ugc" href="https://civarchive.com/models/314840?modelVersionId=353166">mfgcpd</a>, a bit of <a target="_blank" rel="ugc" href="https://civarchive.com/models/371840?modelVersionId=425909">white unicorn</a> and of course <a target="_blank" rel="ugc" href="https://civarchive.com/models/257749?modelVersionId=290640">pony</a>, then is mixed with some loras for style like <a target="_blank" rel="ugc" href="https://civarchive.com/models/264290?modelVersionId=418782">styles for pony </a>, some of mine and some couple more but sorry if i dont remember cos the model has been mixed, remixed, reremixed and mixed again. The model dont have a lot of secret despite the hours of headache tryng to correct it, so consider this a beta althought idk if i would continue to work on it.Don't waist time like me! (put this exactly settings for best quality?)Positve prompte: score_9, score_8_up, score_7_up, score_6_up, score_5_up, score_4_up,(only this, and exact like this, no half, no separate, etc, is what gives more quality, althought you then can put sources or others things, but start putting only this seriously)Negative: None (Seriously don't put anything, you can get schizophrenic with negatives in pony)TAGS (SUPER IMPORANT)put tags, lot of tags, as more tags you put more quality, this model works with longs prompts better, to get tags copy direct them from sites like danbooru or use WD14 tagger or any other extension, but put tags, or long prompts, the difference is abysmal.Steps: 25CFG: 5(Steps only repeats the prompt and draws, but if the promt is not long enough redraws, and eyes gets messed, CFG only gives more quality till it breaks the lines or ddraws limbs)Sampler: DPM++ 2M Karras or Euler AResolution: 1024x1024, 1280x1280 or any SDXL compatible (as more resolution more quality, its better put more resolution than hires fix, but it may create double things)Hires fix: if u are using it 4x_NMKD-Yandere2_255000_G, 4x_Sourcetex-v2-Smooth_120000_G, R-ESRGAN 4x+ Anime6B, 4x-AnimeSharp or any anime upscaler<h3 id="how-to-make-high-quality-images-super-easy-(step-by-step-no-handpicked-example)-ncn78oz6z">How to make high quality images super easy (step by step, no handpicked example)</h3>First of all i would suggest install forgeUi, is faster with SDXL models: <a target="_blank" rel="ugc" href="https://github.com/lllyasviel/stable-diffusion-webui-forge">https://github.com/lllyasviel/stable-diffusion-webui-forge</a>Then you need to install WD14 tagger extension: <a target="_blank" rel="ugc" href="https://github.com/picobyte/stable-diffusion-webui-wd14-tagger">https://github.com/picobyte/stable-diffusion-webui-wd14-tagger</a>(you can dowload automatic in the extension tab if you search wd14)ok we are set up, now lets put essentials things, put DPM++ 2M Karras, 25 steps, desactivate hires fix, put with 1024 and height 1280, and cfg 5<img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/0c811036-0801-4862-8600-20f2798a5468/width=525/0c811036-0801-4862-8600-20f2798a5468.jpeg" />We are set, now lets get a photo, for example this: <a target="_blank" rel="ugc" href="https://danbooru.donmai.us/posts/7440983">https://danbooru.donmai.us/posts/7440983</a>(can be any photo from any site or your pc)Now lets go to the extension tag tagger that we dowload and paste the image, select a model, I like the moat tagger v2 model is what gave me best results, then we put for example 0.5 threshold (optinally you can exclude tabs in the last row like censor, monochrome, loli, etc) (lower threshold more tags, higher less tags)<img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/c55737b5-8898-4270-8428-ccf0f5c40add/width=525/c55737b5-8898-4270-8428-ccf0f5c40add.jpeg" />and the tagger gave us this: 1girl, musical note hair ornament, akamatsu kaede, solo, pantyhose, animal ears, musical note, blonde hair, breasts, playboy bunny, long hair, one eye closed, hair ornament, bangs, smile, rabbit ears, ahoge, leotard, necktie, fake animal ears, open mouth, simple background, strapless leotard, pink eyes, high heels, bare shoulders, looking at viewer, pink leotard, tail, black pantyhose, strapless, white background, sittingWe are almost done, but before copy the tags, rember to put at the begining score_9, score_8_up, score_7_up, score_6_up, score_5_up, score_4_up (i would make a style to forget it forever: score_9, score_8_up, score_7_up, score_6_up, score_5_up, score_4_up {prompt},Now we put the the tags and nothing in negatives!, it should look like this<img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/5f796e48-6826-469b-9245-9b299592030e/width=525/5f796e48-6826-469b-9245-9b299592030e.jpeg" />So easy! Now we press generate and give us this:<img src="http://127.0.0.1:7860/file=D:/ForgeUI/webui/output/txt2img-images/2024-04-10/00108-1123983826.png?1712776084.2128406" />we are alredy done, but we can tune to our taste, for example i will change blonde hair for pink hair and white background for pink background and gives me this:<img src="http://127.0.0.1:7860/file=D:/ForgeUI/webui/output/txt2img-images/2024-04-10/00109-1123983826.png?1712776223.175306" />generate images till you are satisfied, and when you have a one that likes its time to hires fix!Lock the seed, and activate the hire fix, put an anime sampler, i like anime 6b, denoising strengh i like about 0.35 or 0.4 or even 0.5 to add extra details, and upscale it till your maxium computer can afford, i usually scale to 1.5 or 2. All in all it should look like this:<img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/7a31f826-eb89-4f7c-9659-19b4e9d7496d/width=525/7a31f826-eb89-4f7c-9659-19b4e9d7496d.jpeg" />and here is the final result: <img src="http://127.0.0.1:7860/file=D:/ForgeUI/webui/output/txt2img-images/2024-04-10/00117-1123983826.png?1712776862.737024" />And thats it! its really simple!

AIOS V101Foal

aiosV101foal_v1100foalii.safetensors

V1100FOALII

WARNING: Negatives often breaks the style and diminish the quality, even a single word like watermark, read to know how works the model, but as is based in previous version works similar.If you still don't wanna read, here it is a good starting point:FOALIIPOSITIVE QUALITY BOOSTS (add at the beggining of the promt)Nothing(It means don't add anything, just prompt, it does more cartoon and flat colors, don't understimate it, it can do amazing images specially if your prompt is long)score_9, score_8_up, score_7_up, score_6_up, score_5_up, score_4_up,Standard quality9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, score_9, score_8_up, score_7_up, score_6_up, score_5_up, score_4_up,Idk why but adding nines boosts the quality, seriously, try it you will be suprised, you can also add just nines, for each 9 it does different results, also you can pt the score at beggining or at end and it does different results.score_9, score_8_up, score_7_up, score_6_up, score_5_up, score_4_up, score_9, score_8_up, score_7_up, score_6_up, score_5_up, score_4_up,score_9, score_8_up, score_7_up, score_6_up, score_5_up, score_4_up,score_9, score_8_up, score_7_up, score_6_up, score_5_up, score_4_up,Repeting the scores it also boots the quality, you can repeat x2,x3 or whatever you want9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, score_9,It does similar to nothing but boosts the quality a bit, also looses cartoon and probably tends to do more loliNEGATIVESI recomend to no put any negative, often breaks the style and diminish the quality, though is good to do still more cartoon, also 3D, painting, works, but i warn you, it will do different styles.Scores and ratings works like pony.CFG/STEPS (depends a lot of the lenght o the prompt)5/25 is what works best i thinkBut the model can go as low as 1/8 and as high as 15/100Generally it benefits of low cfg/steps as is anime based.As a advice i would move between 15-35 Steps and 3-8 CGIAnother thing that works is lock at 25 steps and put higher cfg, is the cfg that adds the details! (if you prompted them)RESOLUTIONIs better and faster put higher resolutions without high res, than lower resolutions and high resBut these are the best resolutions for me:512*768+highres x2, 800*1200, 1024*1024, 1024*1280, 1024*1364, 1024*1424, 1280*1280The model as based in pony, works better with tags, it likes long lenght prompts , still you can use descriptions cos is merged with SDXL, but i recomend tags.Stye loras may not work as intended, some words may have more weight than others, adjust it.Finally, to create these models is lot of hours drinking coffe that i give for free, if you wanna support me, or want me to continue, consider buy me one!: <a target="_blank" rel="ugc" href="https://ko-fi.com/xdyexd">Kofi</a>FOAL II tried to tame the pony but the pony kicked meThis was a personal use model but i decided to share. Its a finetuned merge, the goal of this model was create an anime model without negatives and color corrected but i think i failed, pony is untamable, so this is what it is, a model to do anime right off the bat. Most of the weight of the model is from <a target="_blank" rel="ugc" href="https://civitai.com/models/288584?modelVersionId=324692">AutismPDO</a> a bit of <a target="_blank" rel="ugc" href="https://civitai.com/models/314840?modelVersionId=353166">mfgcpd</a>, a bit of <a target="_blank" rel="ugc" href="https://civitai.com/models/371840?modelVersionId=425909">white unicorn</a> and of course <a target="_blank" rel="ugc" href="https://civitai.com/models/257749?modelVersionId=290640">pony</a>, then is mixed with some loras for style like <a target="_blank" rel="ugc" href="https://civitai.com/models/264290?modelVersionId=418782">styles for pony </a>, some of mine and some couple more but sorry if i dont remember cos the model has been mixed, remixed, reremixed and mixed again. The model dont have a lot of secret despite the hours of headache tryng to correct it, so consider this a beta althought idk if i would continue to work on it.Don't waist time like me! (put this exactly settings for best quality?)Positve prompte: score_9, score_8_up, score_7_up, score_6_up, score_5_up, score_4_up,(only this, and exact like this, no half, no separate, etc, is what gives more quality, althought you then can put sources or others things, but start putting only this seriously)Negative: None (Seriously don't put anything, you can get schizophrenic with negatives in pony)TAGS (SUPER IMPORANT)put tags, lot of tags, as more tags you put more quality, this model works with longs prompts better, to get tags copy direct them from sites like danbooru or use WD14 tagger or any other extension, but put tags, or long prompts, the difference is abysmal.Steps: 25CFG: 5(Steps only repeats the prompt and draws, but if the promt is not long enough redraws, and eyes gets messed, CFG only gives more quality till it breaks the lines or ddraws limbs)Sampler: DPM++ 2M Karras or Euler AResolution: 1024x1024, 1280x1280 or any SDXL compatible (as more resolution more quality, its better put more resolution than hires fix, but it may create double things)Hires fix: if u are using it 4x_NMKD-Yandere2_255000_G, 4x_Sourcetex-v2-Smooth_120000_G, R-ESRGAN 4x+ Anime6B, 4x-AnimeSharp or any anime upscalerHow to make high quality images super easy (step by step, no handpicked example)First of all i would suggest install forgeUi, is faster with SDXL models: <a target="_blank" rel="noopener noreferrer nofollow" href="https://github.com/lllyasviel/stable-diffusion-webui-forge">https://github.com/lllyasviel/stable-diffusion-webui-forge</a>Then you need to install WD14 tagger extension: <a target="_blank" rel="noopener noreferrer nofollow" href="https://github.com/picobyte/stable-diffusion-webui-wd14-tagger">https://github.com/picobyte/stable-diffusion-webui-wd14-tagger</a>(you can dowload automatic in the extension tab if you search wd14)ok we are set up, now lets put essentials things, put DPM++ 2M Karras, 25 steps, desactivate hires fix, put with 1024 and height 1280, and cfg 5We are set, now lets get a photo, for example this: <a target="_blank" rel="noopener noreferrer nofollow" href="https://danbooru.donmai.us/posts/7440983">https://danbooru.donmai.us/posts/7440983</a>(can be any photo from any site or your pc)Now lets go to the extension tag tagger that we dowload and paste the image, select a model, I like the moat tagger v2 model is what gave me best results, then we put for example 0.5 threshold (optinally you can exclude tabs in the last row like censor, monochrome, loli, etc) (lower threshold more tags, higher less tags)and the tagger gave us this: 1girl, musical note hair ornament, akamatsu kaede, solo, pantyhose, animal ears, musical note, blonde hair, breasts, playboy bunny, long hair, one eye closed, hair ornament, bangs, smile, rabbit ears, ahoge, leotard, necktie, fake animal ears, open mouth, simple background, strapless leotard, pink eyes, high heels, bare shoulders, looking at viewer, pink leotard, tail, black pantyhose, strapless, white background, sittingWe are almost done, but before copy the tags, rember to put at the begining score_9, score_8_up, score_7_up, score_6_up, score_5_up, score_4_up (i would make a style to forget it forever: score_9, score_8_up, score_7_up, score_6_up, score_5_up, score_4_up {prompt},Now we put the the tags and nothing in negatives!, it should look like thisSo easy! Now we press generate and give us this:we are alredy done, but we can tune to our taste, for example i will change blonde hair for pink hair and white background for pink background and gives me this:generate images till you are satisfied, and when you have a one that likes its time to hires fix!Lock the seed, and activate the hire fix, put an anime sampler, i like anime 6b, denoising strengh i like about 0.35 or 0.4 or even 0.5 to add extra details, and upscale it till your maxium computer can afford, i usually scale to 1.5 or 2. All in all it should look like this:and here is the final result:And thats it! its really simple!

Files with SHA256: fa6c1517d9853ab764bcd273565faac3c9de74816032b0b0473f06f5766638d4