<img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/469c7acc-2070-4847-ac97-93331ad1c828/original=true/469c7acc-2070-4847-ac97-93331ad1c828.jpeg" /><h2 id="nino-from-majo-no-tabitabi">💥 Nino from Majo no Tabitabi</h2>— Her appearance in Majo no Tabitabi starts quietly, almost like she is just another gentle maid passing through Elaina’s journey. But the more her situation is revealed, the more that quietness starts to feel painful instead of peaceful. Her story does not rely on a big dramatic speech; it hurts because of how helpless and trapped it feels. She is a brief character, but her episode leaves the kind of sadness that stays around after the scene ends.<h2 id="greatermajo-no-tabitabi-collectionless"><a target="_blank" rel="ugc" href="https://civarchive.com/collections/17217942">&gt;Majo no Tabitabi COLLECTION&lt;</a></h2><h1 id="turn-off-your-xxxx-filter-to-see-hidden-juicy-loras">⚠️Turn off your X/XXX Filter to see hidden JUICY Loras</h1><img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/412ebd2a-fe8a-4354-90ee-ee528c3ed08a/original=true/412ebd2a-fe8a-4354-90ee-ee528c3ed08a.jpeg" /><h2 id="how-i-generate-my-images">🖼️ How I Generate My Images</h2>I made a full article that explains how I generate my images — including my settings, prompting tips, and even my potato PC setup 😅👉 <a target="_blank" rel="ugc" href="https://civarchive.com/articles/17562">Check it out here</a> if you’re curious or want to try the same method.<h2 id="nsfw-gallery-(18+)">🔞 NSFW gallery (18+)</h2>Check my account <a target="_blank" rel="ugc" href="https://sociabuzz.com/zetsubou">here </a>to see the spicy showcase that I can't show here, if any.<h2 id="prompt-tips">🎨 Prompt Tips</h2>These are optional starting points for handling a few behaviors of this version. You may still need some retries depending on the checkpoint, prompt, and LoRA strength.<h3 id="pink-dress-seam-like-details">🔎 Pink dress seam-like details</h3>The original dataset does not have seam details on her pink dress, but generations may still add them sometimes. For a cleaner result closer to the source, try this:⚙️ CFG: 6–7 💪 LoRA weight: 0.9–1My usual CFG setting is 5, but raising it a bit can help the prompt follow the outfit more strictly. This should reduce the extra seam-like details, but it may still need a retry sometimes.<h3 id="removing-the-default-layered-outfit-bias">👕 Removing the default layered outfit bias</h3>Her default dress + shirt combination is somewhat baked in, so when prompting simpler shirt-only looks or other layered outfits, the model may still try to bring back the dress/vest structure.To reduce that behavior, try:🚫 Negative prompt: <code>(dress:1.5), (sleeveless_dress:1.5), (vest:1.5), (shirt_under_dress:1.5)</code>⚙️ CFG: 7💪 LoRA weight: 0.5–0.8Lower LoRA weight helps reduce the outfit bias, while higher CFG helps the prompt take control more strongly.<h3 id="cross-laced-chest-detail-and-neck-ribbon-leakage">🎀 Cross-laced chest detail and neck ribbon leakage</h3>If the cross-laced chest detail or neck ribbon keeps appearing on another outfit, lower the LoRA weight and negative those details directly.For example:🚫 Negative prompt: <code>neck_ribbon, cross-laced_dress</code>Like the other cases, this is controllable, but depending on your checkpoint and settings, it may still need a few retries.<h3 id="default-outfit-details-appearing-on-other-outfits">🧥 Default outfit details appearing on other outfits</h3>When a default outfit detail keeps appearing in a different outfit, add the exact leaking detail to the negative prompt.For example, if the cross-laced details appearing when you do not want it: 🚫 Negative prompt: <code>cross-laced_dress</code>Use the specific leaking tag instead of adding many unrelated negative tags.<h3 id="lora-strength-and-outfit-flexibility">⚖️ LoRA strength and outfit flexibility</h3>A weight of <code>1</code> can still work with a very different outfit, especially when the new outfit is clearly different from her default one.Only lower the LoRA weight gradually when outfit details keep leaking even after using clear outfit tags and negative prompts. Lowering it too much may also weaken her face, hair, and overall character identity<h2 id="lora-project">📊 Lora Project</h2>If you want to know what characters that I'm covering next you can check these link : 👉 <a target="_blank" rel="ugc" href="https://docs.google.com/spreadsheets/d/1oM6Xk0k5sBCw2CQfIFMzQQuuC6lHdOOnw3zES9phBm8/edit?gid=1864372839#gid=1864372839">Personal Project</a> 👉 <a target="_blank" rel="ugc" href="https://docs.google.com/spreadsheets/d/1k15ya9i0TjAHJnBpVfPSsjHsngpjkLsAZ11IcOT0hc0/edit?gid=0#gid=0">Commision Project</a>If this model sparked something in you, show some love with a like, review, or buzz. Now go— Be wild

Nino | Majo no Tabitabi/Wandering Witch: The Journey of Elaina/魔女の旅々

👗 Seam-like dress details<pre><code>This version may sometimes generate extra seam-like details on the pink dress, even though those details are not actually present in the dataset. If you want a cleaner result closer to the source, it usually helps to raise the CFG from my usual setting of 5 up to around 6–7, and use a LoRA weight around 0.9–1.

Even with that workaround, it can still need a retry sometimes.</code></pre>👕 Dress + shirt baked-in bias<pre><code>Her default layered outfit has a bit of baked-in bias. Because the dress + shirt combination appears very consistently in the dataset, the model may try to bring back vest/dress-like layering even when you want a simpler shirt-based outfit.

If you want to push the generation further away from that default layered look, lower the LoRA weight, increase CFG, and use more specific negative tags. I added the exact prompt tips below.</code></pre>🎀 Cross-laced + neck detail leakage<pre><code>The cross-laced chest detail and neck ribbon can also leak into other outfits sometimes. If that happens, lower the LoRA weight a bit and put the exact leaking details into the negative prompt, such as neck_ribbon and cross-laced_dress.</code></pre>As usual, the result can still vary depending on your checkpoint, prompt, LoRA strength, and settings, so a few retries may still be needed.

Files with SHA256: 94e42b9d23996fccf9b99794ae19041f3da899ed3d55433321b455534d694a1e