0xSoul 发表于 2023-6-7 22:56:44

[SD 2.0 768][TextualInversion]djz Cyber Samurai V2 0

<p>This new technique will surely come in handy to many people, the images in the dataset are all solid red .PNG images. <br /><br />The Filenames are a selection of Prompts from <a target="_blank" rel="ugc" href="http://Lexica.art">Lexica.art</a> <br />Trigger word is Case Sensitive: "<strong>Cyber-Samurai</strong>"</p><p></p><p>This demonstrates the concept that we can choose "good prompts" &amp; "Inverted Text" without use of any Image data whatsoever. <br /><br />note: this technique requires a prompt, using the trigger alone will produce wildly abstract results as the AI needs an anchor in the latent space to allow the elements from the prompt selection to resurface.</p><img src="https://imagecache.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/0902fdba-66fb-4171-dcb3-53bd01736000/width=525" /><p><strong>MODELS MATTER:</strong><br />Images below show the outputs using the same prompt &amp; seed with the default released 2.1-768 (top) <br />versus my own fine tuned 2.1-768 model (bottom)<br /></p><img src="https://imagecache.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/f668aaa4-2789-4b8e-7edc-dc1896069800/width=525" /><img src="https://imagecache.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/c4d537b6-d1ea-4044-608a-b5696be36400/width=525" />
页: [1]
查看完整版本: [SD 2.0 768][TextualInversion]djz Cyber Samurai V2 0