<h1 id="generation-guide">Generation Guide</h1><h2 id="model-information">Model Information</h2><ul><li>Model Name: <code>{model_name}</code> (replace with the actual filename you downloaded, e.g., <code>gngsfimZIB.safetensors</code>)</li><li>Trigger Word: <code>{trigger_word}</code></li></ul><h2 id="recommended-settings">Recommended Settings</h2><h3 id="resolution">Resolution</h3><ul><li>2:3 ratio: 821×1232 (portrait)</li><li>3:2 ratio: 1232×821 (landscape)</li><li>Square: 1:1</li><li>Note: You can vary these resolutions with limited success</li><li>FT15 models: Lower max resolution at 512×768</li></ul><h3 id="generation-parameters">Generation Parameters</h3><ul><li>Sampler: Euler (typically)</li><li>CFG Scale:<ul><li>Standard models: 3-7</li><li>Turbo models: 1</li></ul></li><li>Steps:<ul><li>Standard models: 20-50</li><li>Turbo models: 9</li></ul></li><li>LoRA Strength: 0.6-1.0<ul><li>If images look "cooked" or overprocessed, lower the strength</li></ul></li></ul><h2 id="model-series-identifiers">Model Series Identifiers</h2><ul><li>FT15 - Stable Diffusion 1.5 (max resolution: 512×768)</li><li>XLrd - SDXL Run Diffusion X based</li><li>CHHD - Chroma models</li><li>ZIMG - Z-Image Turbo</li><li>ZIB - Z-Image Base</li><li>FKFB - Flux Klein 4B</li><li>QWN - Qwen</li></ul>Note: LoRA files are large and can be resized if neededCurrent Recommendation (January 2026): Use ZIB/ZIT or Chroma models for best results.<h2 id="dataset-type-indicators">Dataset Type Indicators</h2><ul><li>mx - Vastly larger datasets with less consistency, typically trained at lower learning rates for longer durations</li><li>lncc - Smaller, more specific aesthetic-focused datasets</li></ul>Training Data Scale: Datasets vary from 20-30 images to over 1,000,000 images. The median dataset size is closer to 10,000 images.Training Techniques: Models starting at SDXL use mixed resolution training, multi-subject crop, and flips for improved generalization.<h2 id="using-the-wildcard-prompt-template">Using the Wildcard Prompt Template</h2>The piped string format below is designed for ImpactPack Wildcard Processor or Automatic1111 Dynamic Prompts. Copy and paste it into either extension to generate a new randomized prompt each time, built on the distribution of the training dataset.<h3 id="prompt-format">Prompt Format</h3><pre><code>&lt;lora:{model_name}:{0.6|0.7|0.8|0.9|1}&gt; {trigger_word}, {wildcard_tags}</code></pre>Example:<pre><code>&lt;lora:gngsfimZIB:{0.6|0.7|0.8|0.9|1}&gt; example_triggerword, {additional|tags|here}</code></pre><h3 id="understanding-the-wildcard-tags">Understanding the Wildcard Tags</h3><ul><li>More pipes (|) in a tag group = rarer tags in the training data</li><li>Fewer pipes or repeated options = more common tags with better model performance</li><li>More examples in the training data mean the model is better at that particular task or concept</li></ul><h3 id="manual-usage-(without-wildcards)">Manual Usage (without wildcards)</h3>If you're not using dynamic prompts:<ol><li>Load the LoRA manually in your interface</li><li>Start with the trigger word <code>{trigger_word}</code> at the beginning of your prompt</li><li>Add additional tags after the trigger word to vary the composition</li><li>Tags that appear more frequently in the wildcard examples will produce more consistent results</li></ol><h2 id="tips">Tips</h2><ul><li>Always start with the trigger word (the first tag) for best results</li><li>Check sample images for embedded generation parameters</li><li>Add additional tags to vary composition and style</li><li>Experiment with LoRA strength if results don't match expectations</li><li>Tags with more training examples will be more reliable and consistent</li><li>Reference the sample images on this page for working parameter combinations</li></ul><hr /><h1 id="faq:-dataset-filename-and-trigger-word-conventions">FAQ: Dataset Filename &amp; Trigger Word Conventions</h1><h3 id="what-problem-does-this-filename-format-solve">What problem does this filename format solve?</h3>The filename is designed to avoid collisions with generic or common names while also serving as a programmatic signal. It encodes both the trigger word and the dataset type, making it easy for scripts and training pipelines to identify and handle the dataset correctly.<h3 id="why-not-use-a-generic-filename">Why not use a generic filename?</h3>Generic filenames tend to overlap across projects and environments. This format ensures:<ul><li>Uniqueness across datasets</li><li>Clear intent when parsed programmatically</li><li>No ambiguity about dataset content or usage</li></ul><h3 id="what-do-the-suffix-codes-mean">What do the suffix codes mean?</h3>The suffix in the filename specifies:<ul><li>The resolution of the dataset</li><li>The model architecture tier it is intended for</li></ul>This makes it immediately clear what kind of model configuration the dataset targets and helps avoid compatibility issues.<h3 id="what-does-&quot;mx&quot;-stand-for">What does "mx" stand for?</h3>mx means mix. It indicates that the dataset is diverse and vastly larger (potentially hundreds of thousands to over a million images), though less consistent than focused datasets. These models are typically trained at lower learning rates for longer durations to accommodate the dataset diversity.<h3 id="what-does-&quot;lncc&quot;-stand-for">What does "lncc" stand for?</h3>lncc indicates smaller, more specific datasets focused on a particular aesthetic. These are more consistent but cover a narrower range of content.<h3 id="how-are-trigger-words-determined">How are trigger words determined?</h3>Trigger words are embedded in the dataset and filename structure. They function as activation tokens that help the model recognize and generate content consistent with the training data. Always use the specified trigger word at the start of your prompt for best results.<h3 id="how-large-are-the-training-datasets">How large are the training datasets?</h3>Training datasets vary significantly:<ul><li>Minimum: 20-30 images</li><li>Maximum: Over 1,000,000 images</li><li>Median: Approximately 10,000 images</li></ul>Larger datasets (mx) enable broader capabilities but may be less consistent. Smaller datasets (lncc) are more focused and aesthetically coherent.<hr />For best results, always check the sample images on this model page—generation parameters are embedded in the metadata. v3.0 -early releaseimproved, more flexible, larger,512, 768 sd 1.5Vertical 2:3Horiztonal 3:2768:512fp32&lt;lora:onlfnmxFT15:1&gt;, {onlyfans, }{1girl, }{solo, }{realistic, }{looking at viewer, ||||}{navel, |||||||}{sitting, |||||||}{standing, |||||||}{holding, |||||||}{ass, |||||||}{parted lips, |||||||}{selfie, ||||||||}{upper body, ||||||||}{, ||||||||}{full body, ||||||||}{phone, ||||||||}{midriff, ||||||||}{cellphone, ||||||||}{smartphone, ||||||||}{bare shoulders, ||||||||}{holding phone, ||||||||} skinny,v2.0improved, more flexible, larger,512, 768 sd 1.5Vertical 2:3Horiztonal 3:2768:512&lt;lora:onlfnmxFT15_v2:1&gt;,{onlyfans, }{1girl, }{solo, }{realistic, }v1512, 768 sd 1.5Vertical 2:3Horiztonal 3:2768:512*This is an extremely powerful model.**pending overbaked version, known issues, can produce some monsters, pending further training.No actual persons likeness were used in the training of this model, all likeness is purely coincidental. Methods involved interpolation of existing weights.Only Fans Concept / onlyfans style&lt;lora:onlfnmxFT15_p1-step00008000:1&gt;, __onlfn_names__, {onlyfans, }{1girl, }{solo, }{realistic, }{looking at viewer, ||||}{navel, |||||||}{sitting, |||||||}{standing, |||||||}{holding, |||||||}{ass, |||||||}{parted lips, |||||||}{selfie, ||||||||}{upper body, ||||||||}{, ||||||||}{full body, ||||||||}{phone, ||||||||}{midriff, ||||||||}{cellphone, ||||||||}{smartphone, ||||||||}{bare shoulders, ||||||||}{holding phone, ||||||||} <hr />tokens wildcard for help steer the model through the latent space: __onlfn_names__:Abby ExposingAbigail Rat Ch FordAdorable Petite Babe PutriAlex XxAlexaAlexa and Rubs Her TwAlexis RenAli Cruz and Ali CruzAli DisplayingAli ExposingAli FlauntingAli FlauntsAli PosingAli Princess AliAli ShowsAli TeasesAli Yah J Az MineAllie Aut OnAn Ella MillerAndrea ShowsAsian Mary JaneAsian SchoolgirlAudrey Shan IceBabeBabe Al Is SBabe AlexaBabe AlinaBabe AmateurBabe Anastasia Vi FlauntingBabe AndreaBabe BegBabe Beg He B FlauntsBabe Beg He B RemovesBabe El IzzBabe FlashesBabe FlauntsBabe Gymnast AlinaBabe HoornBabe IoanBabe Kat IiiBabe Kat Iii ABabe Kitty ShyBabe LollaBabe LuminaBabe LyaBabe M AshyBabe MadisonBabe Men and Cia Francis PosesBabe Mia MilkerBabe MonoBabe Mya Mira FlauntsBabe Mya Mira ShowsBabe PatriciaBabe PutriBabe Sam Ly PuffBabe Sam Ly Puff PosesBabe Sam Ly Puff PosingBabe Sam Ly Puff ShowsBabe Sam Ly Puff and ShortsBabe ScarlettBabe ShowsBabe SunnyBabe TeensyBabe Velvet Vixen FlauntsBabe Yul Chi KBabe Yuri LiBabe and Yuri LiBald CuntBam BeiBarbie Eva PosingBaskin ChampionBeauty Ali CruzBeauty MadisonBeauty PutriBeauty TeensyBeauty Vi Angel ShowsBeauty Yul Chi KBeg He B Displays Her Beautiful TBeg He B Squeezes Her GreatBlake BlossomBlonde Model AlexaBonnie ChloeBri A MylesBrunette AmateurBrunette Amateur Babe FlauntsBrunette BabeBrunette Babe CelineBrunette Babe E Rel ShowsBrunette Babe E Rel TouchesBrunette Babe FlauntsBrunette Babe Patricia BloomBrunette Babe PutriBrunette Babe Sam Ly Puff ShowsBrunette Babe Sam Ly Puff Shows OffBrunette Babe Shows OffBrunette Babe WithBrunette Beauty Ness A Moore PosesBrunette Beauty Patricia Bloom Posing ProvocativelyBrunette Beauty TeasesBrunette Cutie PutriBrunette ExposesBrunette HottieBrunette Hottie and Sandra Belmont CoversBrunette Kat Iii ABrunette Model and Que SanchezBrunette Nerdy OnlyBrunette Scarlett Oh Arra FlauntsBrunette ShowsBrunette SweetieBrunette TeenBrunette Teen PosesBubble BumBushBuxomBuxom BabeBuxom Babe Mia Ryan ShowsBuxom Babe TeasesCam BabeCam Babe Adriana OliverCam Babe Ali CruzCam Babe AliceCam Babe Alter Eva X ShowsCam Babe Aly Bunny FlauntsCam Babe AnastasiaCam Babe Andrea PardoCam Babe Beg He B FlashesCam Babe Brooke Mills FlauntsCam Babe CCam Babe Dais SCam Babe E Rel ShowsCam Babe Elsa ThoraCam Babe Elsa Thora FlauntsCam Babe HoornCam Babe Larissa S IiCam Babe Lex I Gogo GoCam Babe LolaCam Babe LollaCam Babe MadisonCam Babe Mono MurCam Babe Mya Mira FlauntsCam Babe Nasty ACam Babe PatriciaCam Babe Patricia BloomCam Babe PutriCam Babe RedheadCam Babe SandraCam Babe ScarlettCam Babe Suki X OxoCam Babe TeasesCam Babe Vi Xy Di FlauntsCam Babe Your and Lick AbleCam Babe Yumi BumsCam ModelCam Model Gymnast AlinaCam Model MonoCam Model Sam Ly Puff FlauntsCam Model Victoria Bros HkCary S SchulzeCasey JamesCassie VCatherine LynnCecil Ie Nord AhlChai Lee SonCharly JordanChristina E OmChristine Cardo ExposesChristine Cardo FlauntsChubbyChubby Babe Kristi Kkk FlauntsChubby Babe Kristi Kkk TeasesChubby Babe ShowsChubby Brunette and Kristi Kkk ShowsChubby Naked BodyChubby and Cam Babe Kay La BumsChubby and Cam Babe Kristi Kkk ShowsChubby and Cam Model KristiChubby and Kristi Kkk Wipes SudsCia Francis PosesCia Francis PosingCia Francis RemovesCindy KimberlyCj MilesClaudia ValentinaCurly Haired Babe Sky Alex XxCurly Haired and Sky Alex XxCute Asian Babe PutriCute BabeCute Babe AmateurCute Babe EmilyCute Babe TeasesCute Blonde BabeCute Brunette Babe E Rel PosingCute Brunette Babe E Rel SpreadsD Annie RielDark Haired and Babe E Rel GrabsDebbie S A ThDina De NoireDray A MicheleEating AElsa ThoraEun BiEun J IiiEva LopezFiona BarronFirm BootyGabi MoonGatoGlamorous Babe Anastasia Vi CoversGlamorous Babe Sam Ly Puff Shows Off Her Big As S Wearing Sexy ShortsGlamorous Babe Stefani IsabelGlamorous Brunette A Very A Lena FlauntsGlamorous Brunette Babe ShowsGlamorous ModelGlamour BabeGlamour Babe FlauntsGlamour Babe PutriGlamour Babe ShowsGlamour Babe Shows OffGlamour BabesGlamour Model NastyGolfer SkirtGra C IiiGraceHorny Babe Alina FlauntsHos In LeeHot BootyIsabelle MathersIulia VoJ Ace KimJana AyyJana E RobertsJasmin El AgoJasmine BaresJasmine FondlesJenn QJordan Hall FlauntsJordan Hall X Jordan HallJug Gs Brooke Mills PosesJug Gs and Eating AJug Gs and Touches Her CuntJun HasegawaKate K Kate RbKay La BumsKay La Puff FKay La Puff F Kay La Puff FKay La Puff F Shows OffKei LahKei Lah BenKei Lah KangKendall KayKimberley GarnerKinky AmateurKinky Only Fans Model Lumina Tria ExposesKinky Only Fans Teen Baby Yon A Poses In Her LatexKo Keto KaKris TopLatina Andrea VelascoLee Ho JungLei LeiLemy BeautyLi El OhanaLinda MartinLingerie CladLong Haired Centerfold Anastasia Vi FlauntingLorena RaeLorna Spain ELou X DisplaysLou X ExposesLt NdaLu Ra NormandLuis Iana Da CruzLumina Tria TeasesLy FiMaddie GeMaddie Ge ChMaddie Ge Ch PosingMaddie Gec and Maddie Ge ChMaddy CrumMads RaffertyMaiMaid Babe Sasha and Jug GsMarilyn MeloMarissa Dana EMary FosterMary Foster and Mary FosterMath Il DMeredith MickelsonMia Milker S and Mia Milker SMil F PrincessMila Pan FlauntsMini JelloModel Ko Keto KaNe Rize Pey PerNet A Al Chi MisterNeva GOh T To MomPatricia BloomQue SanchezRaven HairedRaven Haired and Babe TeasesRossRubs Her Big BoobsS A Har LunaSalacious Babe Ness A MooreSam Ly PuffSam Ly Puff PosesSam Ly Puff Sam Ly PuffSam Ly Puff Strips Her ShortsSela Maw It Shi Fe RawShay BrownShows Her BoobsShows OffSierra SkyeSkinnySkinny AsianSkinny Asian BabeSkinny Babe Angel AdeleSkinny Babe Ariel A Moon PosesSkinny Babe Kay La Puff F Teases WithSkinny Babe TakesSkinny Teen FlashesSkinny and Ko Keto KaSky Alex XxSnow BlackSofia Jam Or AStacked Babe Jenn Ny Santorini FlauntsStacked Babe RedheadStephanie SherryStocking Ed FeetStripsStunner Megan Lou X TeasesSuarez ShowsSue MscSweet Blonde Babe Carol FlySweet Nerdy Babe WithSweet Petite Babe PutriTarynTattoosTattoos AngelicaJenny SantoKay La Puff F ShowsKo Keto KaYumi BumsTouches HerselfTri St In May STria Lumina TriaVery Sexy Teen Anastasia Vi ExposesVictoria Bros HkWai Kei TongYo D It Ye ManeYummy As SYummy BootyYummy PussyYuri Lionlyfans

Sarah Petersons OnlyFans Whores modifier FT15

onlfnmxFT15_p1-step00008000.safetensors

v1.0

v1.0 ZIMG

v1.0 CHHD

v1.0 SDXL XLrd

v3.3

v3.2

v3.1

v3.0

v2.0

Files with SHA256: 8962c398cd429f627842f8e7fb85f1e3ab72881a5ef5f7772819140cef302d63