Stable Diffusion

Hardcore Hentai V12

model_id: hardcore-hentai-v12

<p>https://civitai.com/models/136054?modelVersionId=232266<br><br>[This model is using BAKED VAE]</p> <h3 id="heading-361">Prompt</h3> <p><strong>best quality,</strong></p> <p>I like to keep the filler tags to a minimum, but you can get good result with more tags too.</p> <p>Use danbooru tags or sentence-based prompts</p> <p>Extension for tag autocompletion&nbsp;<a href="https://github.com/DominikDoom/a1111-sd-webui-tagcomplete" target="_blank" rel="noopener ugc">tagcomplete</a></p> <h3 id="heading-594">Prompt Order for good composition</h3> <p>Tags at the start have more influence than the ones at the bottom.</p> <p><strong>The 7 blocks template</strong>&nbsp;(Work in progress):</p> <ol> <li> <p><a href="https://danbooru.donmai.us/wiki_pages/tag_group%3Aimage_composition" target="_blank" rel="noopener ugc">view angle</a>,</p> </li> <li> <p>participants -&gt;<a href="https://danbooru.donmai.us/wiki_pages/1girl" target="_blank" rel="noopener ugc">1girl</a>,&nbsp;<a href="https://danbooru.donmai.us/wiki_pages/1boy" target="_blank" rel="noopener ugc">1boy</a>,&nbsp;<a href="https://danbooru.donmai.us/wiki_pages/tag_group%3Asex_acts" target="_blank" rel="noopener ugc">sex acts</a>,&nbsp;<a href="https://danbooru.donmai.us/wiki_pages/tag_group%3Asexual_positions" target="_blank" rel="noopener ugc">sexual positions</a>,</p> </li> <li> <p><a href="https://danbooru.donmai.us/wiki_pages/tag_group%3Aposture" target="_blank" rel="noopener ugc">posture</a>, secondary actions,</p> </li> <li> <p>body -&gt;&nbsp;<a href="https://danbooru.donmai.us/wiki_pages/breasts" target="_blank" rel="noopener ugc">breasts size</a>, body fat, age, stature,&nbsp;<a href="https://danbooru.donmai.us/wiki_pages/tag_group%3Abody_parts" target="_blank" rel="noopener ugc">body part focus</a>,</p> </li> <li> <p><a href="https://danbooru.donmai.us/wiki_pages/tag_group%3Ahair_color" target="_blank" rel="noopener ugc">hair color</a>,&nbsp;<a href="https://danbooru.donmai.us/wiki_pages/tag_group%3Ahair_styles" target="_blank" rel="noopener ugc">hair style</a>,&nbsp;<a href="https://danbooru.donmai.us/wiki_pages/tag_group%3Aeyes_tags" target="_blank" rel="noopener ugc">eyes</a>,&nbsp;<a href="https://danbooru.donmai.us/wiki_pages/tag_group%3Aface_tags" target="_blank" rel="noopener ugc">face expressions</a>,&nbsp;<a href="https://danbooru.donmai.us/wiki_pages/cum" target="_blank" rel="noopener ugc">cum location</a>,</p> </li> <li> <p><a href="https://danbooru.donmai.us/wiki_pages/tag_group%3Aattire" target="_blank" rel="noopener ugc">clothing</a>,</p> </li> <li> <p><a href="https://danbooru.donmai.us/wiki_pages/tag_group%3Alocations" target="_blank" rel="noopener ugc">location</a>, furniture, objects,</p> </li> </ol> <p>You don't need to fill in every category, it's just for reference.</p> <h3 id="heading-362">Negative prompt</h3> <p><strong>worst quality, low quality, normal quality,</strong></p> <p>And whatever you don&rsquo;t want to be on the image</p> <p>Some people use the standard (worst quality, low quality: 1.4) + negative embeddings with good results too.</p> <h3 id="heading-363">Samplers</h3> <p>I like&nbsp;<strong>DPM++ SDE Karras&nbsp;</strong>but you can use any sampler. Some samplers converge at a defined image, others keep changing with each step.</p> <p>You can find better explanations&nbsp;<a href="https://www.youtube.com/watch?v=JAMkYVV-n18" target="_blank" rel="noopener ugc">here</a>&nbsp;and&nbsp;<a href="https://stable-diffusion-art.com/samplers/" target="_blank" rel="noopener ugc">here</a>.</p> <h3 id="heading-364">Steps</h3> <p><strong>25+</strong></p> <h3 id="heading-365">CFG</h3> <ul> <li> <p><strong>7</strong>: standard</p> </li> <li> <p><strong>3&ndash;15:</strong>&nbsp;working range</p> </li> </ul> <p>Lower cfgs don&rsquo;t follow the prompt as much. Higher cfgs will make &ldquo;burned&rdquo; images</p> <p>The extension&nbsp;<a href="https://github.com/mcmonkeyprojects/sd-dynamic-thresholding" target="_blank" rel="noopener ugc">sd-dynamic-thresholding&nbsp;</a>can play around with different CFGs at different steps</p> <h3 id="heading-366">Sizes</h3> <ul> <li> <p>1:1&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; [512, 512], [768, 768]</p> </li> <li> <p>2:3&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; [512, 768]</p> </li> <li> <p>3:2&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; [768, 512]</p> </li> </ul> <p>You can use Hires fix to get a bigger picture with more details</p>