<h1 id="un-phony-34ni0r0rm">un-phony</h1>go use <a target="_blank" rel="ugc" href="https://github.com/mcmonkeyprojects/SwarmUI/">SwarmUI</a>! [it has comfy!]an xl-illust-noob-vp-fork like model with enhanced background and "uguu"-ness capabilities finetuned on some recent pixiv data; in other words, a good vpred ztsnr-esque model; loras work, need to use vpred though [see my other two recent loras for examples]recommended:<ul><li>euler cfg ++</li><li>kl optimal</li><li><s>cfg</s> scale = 1.7</li><li>mahiro</li><li>1.5 mp+ works</li></ul>train method:<ul><li>i did not finetune it</li><li>i just did a big (256/128) lokr</li><li>my 3060 can only do so much</li></ul>notes: dont use defaults, other than artists i just use <code>4 fingers, very awa, very uguu</code> as positive and <code>3 fingers</code> as negative | all images other than teto x miku kissing is with the same seed, 33-4 (3304) | imgs generated with model-Ayodayno v2 commercial<pre><code>If you use the model on any platform, you must provide a link or reference to the original model. You must give credit to the licensor whenever you use the model.
The licensor does not provide any warranty and is not liable for any damages caused by the use of the model.</code></pre>contact: <code>raiko.horikawa</code> | <a target="_blank" rel="ugc" href="https://discord.gg/touhouai">discord server i am active in</a>

un-phony

<code>very uguu</code> might be helpful, added it manually to one source

Files with SHA256: 3251f182f8538abced6ed95254cb5ac0e450bf3f247632a66e968d19accf1d46