http://kenji.blog/de/tags/bildgenerierung/Recent content in Bildgenerierung on kenji.blogHugo -- gohugo.iodekenjinoteThu, 06 Apr 2023 00:43:19 +0900http://kenji.blog/de/p/aistablediffusion%E3%82%92%E4%BD%BF%E3%81%A3%E3%81%A6%E3%82%A4%E3%83%A9%E3%82%B9%E3%83%88%E7%94%BB%E5%83%8F%E7%94%9F%E6%88%90%E3%81%99%E3%82%8B%E6%96%B9%E6%B3%95/Thu, 06 Apr 2023 00:43:19 +0900http://kenji.blog/de/p/aistablediffusion%E3%82%92%E4%BD%BF%E3%81%A3%E3%81%A6%E3%82%A4%E3%83%A9%E3%82%B9%E3%83%88%E7%94%BB%E5%83%8F%E7%94%9F%E6%88%90%E3%81%99%E3%82%8B%E6%96%B9%E6%B3%95/<img src="http://kenji.blog/p/aistablediffusion%E3%82%92%E4%BD%BF%E3%81%A3%E3%81%A6%E3%82%A4%E3%83%A9%E3%82%B9%E3%83%88%E7%94%BB%E5%83%8F%E7%94%9F%E6%88%90%E3%81%99%E3%82%8B%E6%96%B9%E6%B3%95/img.png" alt="Featured image of post Wie man Illustrationsbilder mit KI (Stable Diffusion) generiert" /><h1 id="was-ist-stable-diffusion">Was ist Stable diffusion </h1><p>Stable diffusion ist eine KI, die von einem Forschungsteam der Universität München, Deutschland, entwickelt wurde und Bilder aus eingegebenen Textinformationen generiert. Indem sie mit verschiedenen Bildern trainiert wurde, kann sie eine Vielzahl von Bildern generieren, von Fotos bis hin zu Illustrationen.</p> <p>Dieses Mal werde ich vorstellen, wie man Illustrationsbilder mit den vortrainierten Daten von Stable diffusion generiert.</p> <h1 id="was-sie-benötigen">Was Sie benötigen </h1><ul> <li>Google-Konto</li> </ul> <p>Nur das</p> <h1 id="generierungsverfahren">Generierungsverfahren </h1><ol> <li>Öffnen Sie <a class="link" href="https://colab.research.google.com" target="_blank" rel="noopener" >https://colab.research.google.com</a></li> <li>Wählen Sie <code>Datei</code> oben links und dann <code>Neues Notebook</code></li> <li>Wählen Sie <code>Bearbeiten</code> und dann <code>Notebook-Einstellungen</code></li> <li>Ändern Sie den <code>Hardwarebeschleuniger</code> auf <code>GPU</code> <img src="http://kenji.blog/p/aistablediffusion%E3%82%92%E4%BD%BF%E3%81%A3%E3%81%A6%E3%82%A4%E3%83%A9%E3%82%B9%E3%83%88%E7%94%BB%E5%83%8F%E7%94%9F%E6%88%90%E3%81%99%E3%82%8B%E6%96%B9%E6%B3%95/img_2.png" width="627" height="324" srcset="http://kenji.blog/p/aistablediffusion%E3%82%92%E4%BD%BF%E3%81%A3%E3%81%A6%E3%82%A4%E3%83%A9%E3%82%B9%E3%83%88%E7%94%BB%E5%83%8F%E7%94%9F%E6%88%90%E3%81%99%E3%82%8B%E6%96%B9%E6%B3%95/img_2_hud7bbdf6a24e33b2b36528d9e25b016c8_23961_480x0_resize_box_3.png 480w, http://kenji.blog/p/aistablediffusion%E3%82%92%E4%BD%BF%E3%81%A3%E3%81%A6%E3%82%A4%E3%83%A9%E3%82%B9%E3%83%88%E7%94%BB%E5%83%8F%E7%94%9F%E6%88%90%E3%81%99%E3%82%8B%E6%96%B9%E6%B3%95/img_2_hud7bbdf6a24e33b2b36528d9e25b016c8_23961_1024x0_resize_box_3.png 1024w" loading="lazy" alt="img_2.png" class="gallery-image" data-flex-grow="193" data-flex-basis="464px" ></li> <li>Fügen Sie den folgenden Code ein und führen Sie ihn aus</li> </ol> <div class="highlight"><div class="chroma"> <table class="lntable"><tr><td class="lntd"> <pre tabindex="0" class="chroma"><code><span class="lnt">1 </span></code></pre></td> <td class="lntd"> <pre tabindex="0" class="chroma"><code class="language-fallback" data-lang="fallback"><span class="line"><span class="cl">!pip install diffusers==0.8.0 transformers </span></span></code></pre></td></tr></table> </div> </div><ol start="6"> <li>Fügen Sie den folgenden Code ein und führen Sie ihn aus</li> </ol> <div class="highlight"><div class="chroma"> <table class="lntable"><tr><td class="lntd"> <pre tabindex="0" class="chroma"><code><span class="lnt">1 </span></code></pre></td> <td class="lntd"> <pre tabindex="0" class="chroma"><code class="language-fallback" data-lang="fallback"><span class="line"><span class="cl">from diffusers import StableDiffusionPipeline </span></span></code></pre></td></tr></table> </div> </div><ol start="7"> <li>Fügen Sie den folgenden Code ein und führen Sie ihn aus</li> </ol> <div class="highlight"><div class="chroma"> <table class="lntable"><tr><td class="lntd"> <pre tabindex="0" class="chroma"><code><span class="lnt">1 </span><span class="lnt">2 </span></code></pre></td> <td class="lntd"> <pre tabindex="0" class="chroma"><code class="language-fallback" data-lang="fallback"><span class="line"><span class="cl">pipe = StableDiffusionPipeline.from_pretrained(&#34;gsdf/Counterfeit-V2.5&#34;) </span></span><span class="line"><span class="cl">pipe.to(&#34;cuda&#34;) </span></span></code></pre></td></tr></table> </div> </div><ol start="8"> <li>Fügen Sie den folgenden Code ein und führen Sie ihn aus</li> </ol> <div class="highlight"><div class="chroma"> <table class="lntable"><tr><td class="lntd"> <pre tabindex="0" class="chroma"><code><span class="lnt">1 </span><span class="lnt">2 </span><span class="lnt">3 </span><span class="lnt">4 </span><span class="lnt">5 </span></code></pre></td> <td class="lntd"> <pre tabindex="0" class="chroma"><code class="language-fallback" data-lang="fallback"><span class="line"><span class="cl">prompt = &#34;((masterpiece,best quality)),1girl, solo, animal ears, rabbit, barefoot, knees up, dress, sitting, rabbit ears, short sleeves, looking at viewer, grass, short hair, smile, white hair, puffy sleeves, outdoors, puffy short sleeves, bangs, on ground, full body, animal, white dress, sunlight, brown eyes, dappled sunlight, day, depth of field&#34; </span></span><span class="line"><span class="cl">n_prompt = &#34;EasyNegative, extra fingers,fewer fingers&#34; </span></span><span class="line"><span class="cl"> </span></span><span class="line"><span class="cl">image = pipe(prompt, negative_prompt = n_prompt).images[0] </span></span><span class="line"><span class="cl">image </span></span></code></pre></td></tr></table> </div> </div><p>Der hier verwendete <code>Prompt</code> basiert auf dem <code>Prompt</code> von <a class="link" href="https://huggingface.co/gsdf/Counterfeit-V2.5" target="_blank" rel="noopener" >https://huggingface.co/gsdf/Counterfeit-V2.5</a>.</p> <h2 id="generierte-ergebnisse-einige">Generierte Ergebnisse (einige) </h2><p><img src="http://kenji.blog/p/aistablediffusion%E3%82%92%E4%BD%BF%E3%81%A3%E3%81%A6%E3%82%A4%E3%83%A9%E3%82%B9%E3%83%88%E7%94%BB%E5%83%8F%E7%94%9F%E6%88%90%E3%81%99%E3%82%8B%E6%96%B9%E6%B3%95/img_1.png" width="512" height="512" srcset="http://kenji.blog/p/aistablediffusion%E3%82%92%E4%BD%BF%E3%81%A3%E3%81%A6%E3%82%A4%E3%83%A9%E3%82%B9%E3%83%88%E7%94%BB%E5%83%8F%E7%94%9F%E6%88%90%E3%81%99%E3%82%8B%E6%96%B9%E6%B3%95/img_1_hu1ec19e515fd7538067cc14f2c3c8bddf_594392_480x0_resize_box_3.png 480w, http://kenji.blog/p/aistablediffusion%E3%82%92%E4%BD%BF%E3%81%A3%E3%81%A6%E3%82%A4%E3%83%A9%E3%82%B9%E3%83%88%E7%94%BB%E5%83%8F%E7%94%9F%E6%88%90%E3%81%99%E3%82%8B%E6%96%B9%E6%B3%95/img_1_hu1ec19e515fd7538067cc14f2c3c8bddf_594392_1024x0_resize_box_3.png 1024w" loading="lazy" alt="img_1.png" class="gallery-image" data-flex-grow="100" data-flex-basis="240px" ></p> <p><img src="http://kenji.blog/p/aistablediffusion%E3%82%92%E4%BD%BF%E3%81%A3%E3%81%A6%E3%82%A4%E3%83%A9%E3%82%B9%E3%83%88%E7%94%BB%E5%83%8F%E7%94%9F%E6%88%90%E3%81%99%E3%82%8B%E6%96%B9%E6%B3%95/img_3.png" width="512" height="512" srcset="http://kenji.blog/p/aistablediffusion%E3%82%92%E4%BD%BF%E3%81%A3%E3%81%A6%E3%82%A4%E3%83%A9%E3%82%B9%E3%83%88%E7%94%BB%E5%83%8F%E7%94%9F%E6%88%90%E3%81%99%E3%82%8B%E6%96%B9%E6%B3%95/img_3_hu67e673668e1b02812ead4fb15912ac36_533749_480x0_resize_box_3.png 480w, http://kenji.blog/p/aistablediffusion%E3%82%92%E4%BD%BF%E3%81%A3%E3%81%A6%E3%82%A4%E3%83%A9%E3%82%B9%E3%83%88%E7%94%BB%E5%83%8F%E7%94%9F%E6%88%90%E3%81%99%E3%82%8B%E6%96%B9%E6%B3%95/img_3_hu67e673668e1b02812ead4fb15912ac36_533749_1024x0_resize_box_3.png 1024w" loading="lazy" alt="img_3.png" class="gallery-image" data-flex-grow="100" data-flex-basis="240px" ></p> <p><img src="http://kenji.blog/p/aistablediffusion%E3%82%92%E4%BD%BF%E3%81%A3%E3%81%A6%E3%82%A4%E3%83%A9%E3%82%B9%E3%83%88%E7%94%BB%E5%83%8F%E7%94%9F%E6%88%90%E3%81%99%E3%82%8B%E6%96%B9%E6%B3%95/img_4.png" width="512" height="512" srcset="http://kenji.blog/p/aistablediffusion%E3%82%92%E4%BD%BF%E3%81%A3%E3%81%A6%E3%82%A4%E3%83%A9%E3%82%B9%E3%83%88%E7%94%BB%E5%83%8F%E7%94%9F%E6%88%90%E3%81%99%E3%82%8B%E6%96%B9%E6%B3%95/img_4_hu2b826ec3ad4d7685d995efaa40aac0c5_597506_480x0_resize_box_3.png 480w, http://kenji.blog/p/aistablediffusion%E3%82%92%E4%BD%BF%E3%81%A3%E3%81%A6%E3%82%A4%E3%83%A9%E3%82%B9%E3%83%88%E7%94%BB%E5%83%8F%E7%94%9F%E6%88%90%E3%81%99%E3%82%8B%E6%96%B9%E6%B3%95/img_4_hu2b826ec3ad4d7685d995efaa40aac0c5_597506_1024x0_resize_box_3.png 1024w" loading="lazy" alt="img_4.png" class="gallery-image" data-flex-grow="100" data-flex-basis="240px" ></p> <h2 id="referenzen">Referenzen </h2><ul> <li><a class="link" href="https://huggingface.co/gsdf/Counterfeit-V2.5" target="_blank" rel="noopener" >https://huggingface.co/gsdf/Counterfeit-V2.5</a></li> <li><a class="link" href="https://www.youtube.com/watch?v=l8-fVSM2PVQ" target="_blank" rel="noopener" >Ich habe in 15 Minuten ein Bildgenerierungsprogramm mit Künstlicher Intelligenz (KI) erstellt 【Live-Programmierung】</a></li> </ul>