<h3 id="important:-read-below-iq8qtil5a"><a target="_blank" rel="ugc" href="https://ko-fi.com/ctian"><img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/cf614846-63c8-4596-a5fd-f5962d69ae75/width=525/cf614846-63c8-4596-a5fd-f5962d69ae75.jpeg" /></a>IMPORTANT: READ BELOW</h3>Trigger word(s): ctiancelesti4, blue eyes, Clothing: blue necktie, white capelet, long sleeves, underbust, blue skirt, emblem, black thighhighs, frills, frilled skirt, Battlesuit: unitard, black thigh boots, white breasts, white sleeves, black gloves, bare thighs, latex, (white buruma),commission: <a target="_blank" rel="ugc" href="https://ko-fi.com/ctian">https://ko-fi.com/ctian</a> 😼 Example Positive prompt ✅:<pre><code>score_9, score_8_up, score_7_up, source_anime, masterpiece, 1girl, ctiancelesti4, blue necktie, white capelet, long sleeves, underbust, blue skirt, emblem, black thighhighs, frills, frilled skirt, looking at viewer, outdoors, city, depth of field, day, sunny,</code></pre>Negative prompt ❌:<pre><code>sketch, 3d, pouch,</code></pre>Recommended resolution, aspect ratio for sdxl, pony 📷768 x 1344: Vertical (9:16)832 x 1216: Portrait (4:5) **1024 x 1024: Square (1:1)1182 x 886: Photo (4:3)1254 x 836: Landscape (3:2)1365 x 768: Widescreen (16:9)Work flow, procedure 🔧1. Collect dataset (booru, anime screencaps, search)2. Clean up dataset using upscalers, inpainting and photoshop3. Tag dataset using booru <a target="_blank" rel="ugc" href="https://github.com/starik222/BooruDatasetTagManager">github</a>4. Train LoRa in kohya ss locally <a target="_blank" rel="ugc" href="https://github.com/bmaltais/kohya_ss">github</a>5. Test epochs using xyz grid in controlnet to ensure LoRa is not overfitted/underfitted.6. Submit model to CivitAI <a target="_blank" rel="ugc" href="https://civarchive.com/user/ctian">profile</a>do not ask me training parameters, check lora metadata for training parameters<h1 id="faq:-s7lewsido">FAQ:</h1>Q: why not use character name as trigger word? A: I use unique token to prevent other things from being prompted when using triggerword Q: do you have other account in other websites? A: yes, tensor, pixai Q: what external generator do you use? A: a111 mostly, comfyui for img2vid

Celistia Ralgris セリスティア・ラルグリス PONY - Undefeated Bahamut Chronicle バハムート (2 CLOTHING)

the battle outfit is undertrained, not enough data.

Celestia_Ralgris_pony_ct.safetensors

v1.0

<h3 id="celistia-ralgris-pony-undefeated-bahamut-chronicle-d9kp7f4gy">Celistia Ralgris セリスティア・ラルグリス PONY - Undefeated Bahamut Chronicle バハムート</h3><h3 id="ko-fi-commission-important:-place-&quot;pouch&quot;-in-negative-prompts-kcjnfson6"><a target="_blank" rel="ugc" href="https://ko-fi.com/ctian"><img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/7a25ef69-a98e-4d14-9e8b-8b68ff0c3339/width=525/7a25ef69-a98e-4d14-9e8b-8b68ff0c3339.jpeg" /></a>Ko-fi commission IMPORTANT: PLACE "POUCH" IN NEGATIVE PROMPTS</h3>Trigger word(s): ctiancelesti4, blue eyes, Clothing: blue necktie, white capelet, long sleeves, underbust, blue skirt, emblem, black thighhighs, frills, frilled skirt, Battlesuit: unitard, black thigh boots, white breasts, white sleeves, black gloves, bare thighs, latex, (white buruma),commission: <a target="_blank" rel="ugc" href="https://ko-fi.com/ctian">https://ko-fi.com/ctian</a> 😼 Example Positive prompt ✅:<pre><code>score_9, score_8_up, score_7_up, source_anime, masterpiece, 1girl, ctiancelesti4, blue necktie, white capelet, long sleeves, underbust, blue skirt, emblem, black thighhighs, frills, frilled skirt, looking at viewer, outdoors, city, depth of field, day, sunny,</code></pre>Negative prompt ❌:<pre><code>sketch, 3d, pouch,</code></pre>Recommended resolution, aspect ratio for sdxl, pony 📷768 x 1344: Vertical (9:16)832 x 1216: Portrait (4:5) **1024 x 1024: Square (1:1)1182 x 886: Photo (4:3)1254 x 836: Landscape (3:2)1365 x 768: Widescreen (16:9)Work flow, procedure 🔧1. Collect dataset (booru, anime screencaps, search)2. Clean up dataset using upscalers, inpainting and photoshop3. Tag dataset using booru <a target="_blank" rel="ugc" href="https://github.com/starik222/BooruDatasetTagManager">github</a>4. Train LoRa in kohya ss locally <a target="_blank" rel="ugc" href="https://github.com/bmaltais/kohya_ss">github</a>5. Test epochs using xyz grid in controlnet to ensure LoRa is not overfitted/underfitted.6. Submit model to CivitAI <a target="_blank" rel="ugc" href="https://civitai.com/user/ctian">profile</a>do not ask me training parameters, check lora metadata for training parameters<h1 id="faq:-0rgtib68c">FAQ:</h1>Q: why not use character name as trigger word? A: I use unique token to prevent other things from being prompted when using triggerword Q: do you have other account in other websites? A: yes, tensor, pixai Q: what external generator do you use? A: a111 mostly, comfyui for img2vid

&lt;p&gt;the battle outfit is undertrained, not enough data.&lt;/p&gt;

Files with SHA256: 5455323054e1f1bc5ae88617324209b7865758908d9cd7e2ff66a7f9150437cc