<h2 id="winter-wyvern-auroth">Winter Wyvern — Auroth</h2>Bring Auroth from Dota 2 into your own scenes and styles. The Illustrious version focuses on her appearance, leaving room for your checkpoint, prompt and style LoRAs to shape the final look.<h2 id="v1.0-or-illustrious">v1.0 | Illustrious</h2><ul><li>Trigger: <code>auroth</code>. For best results, also include <code>wyvern</code> and/or <code>dragon</code> in your prompt.</li><li>Strength: start at <code>0.7</code>; try <code>0.6–1.0</code> depending on your checkpoint and prompt.</li><li>Starting settings: CFG <code>5.5</code>, <code>30–50</code> sampling steps.</li></ul>Trained from scratch with an expanded character dataset and a separate set of regularization images. Higher strength can bring out more character details, but is not always better for anatomy or prompt flexibility.<h3 id="one-character-different-looks">One character, different looks</h3>Watercolor, lineart and a 3D-styled example. These use different prompts and checkpoints to show a range of looks; they are not a controlled style-only comparison.<img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/30bf900a-2359-4584-9460-0de13f1e23fe/original=true/30bf900a-2359-4584-9460-0de13f1e23fe.jpeg" /><h3 id="finding-the-right-strength">Finding the right strength</h3>The same prompt, seed and settings, with only the Auroth weight changing: off, 0.6, 0.8 and 1.0. Start lower and increase only if you need a stronger likeness.<img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/4b05c85a-59f3-49d1-806a-2dd93bb9eede/original=true/4b05c85a-59f3-49d1-806a-2dd93bb9eede.jpeg" /><a target="_blank" rel="ugc" href="https://civarchive.com/models/503815/nova-furry-xl?modelVersionId=2943166">NovaFurryXL_ilV180A</a> · seed 93103 · DPM++ 2M SDE / Karras · 30 steps · CFG 5.5.<h3 id="combining-with-style-loras">Combining with style LoRAs</h3>Auroth can be paired with a style LoRA. Below, Auroth stays at 0.5 while <a target="_blank" rel="ugc" href="https://civarchive.com/models/1486815/hooves-art-style-or-pony-illustrious">Hooves Art Style v2</a> changes from off to 0.6 and 1.0. The prompt and seed stay the same, including the style words <code>3d, hoovesart_style</code>.<img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/2589da37-9043-4269-a58b-ec78ac098647/original=true/2589da37-9043-4269-a58b-ec78ac098647.jpeg" /><a target="_blank" rel="ugc" href="https://civarchive.com/models/503815/nova-furry-xl?modelVersionId=2943166">NovaFurryXL_ilV180A</a> · seed 77715 · DPM++ 2M SDE / Karras · 30 steps · CFG 5.5.Appearance and anatomy still depend on the scene and checkpoint. These examples show tested combinations, rather than a guarantee that every style or pose will behave the same way.<h2 id="v0.1-or-pony">v0.1 | Pony</h2>This LoRA is designed to generate Winter Wyvern from Dota 2. Assembling the dataset was challenging, so the results may not always match expectations perfectly.Key Details:<ul><li>Trigger Word: "winter wyvern"</li><li>Recommended Settings: CFG scale around 6, start with 20-30 sampling steps</li></ul>

Winter Wyvern (Auroth) Dota2

Retrained for Illustrious with an expanded dataset and regularization images. Designed to capture Auroth’s appearance while keeping the base model’s style flexible. Trigger: auroth. Start at strength 0.5; increase to 0.8 for a stronger likeness.

Files with SHA256: 0618c84d02496a7ba7096dc2822ade7d976bf57bd492bee9ba8b1a885784b8bf