<h1 id="club3-style">[Club3 Style]</h1>["The go-to anime illustration style, especially strong for NSFW work"]<hr /><h2 id="quick-info">Quick Info</h2><pre><code>Type: LoRA
Model: Anima
Trigger Words: none
Recommended Weight: 1.0 ~ 2.0
Target: portraits and illustrations of anime-style characters</code></pre>Full model details are in the "Versioning" section below, and are also mirrored in the model version notes.<hr /><h2 id="introduction">Introduction</h2>This model recreates the artwork style of @club3, with a focus on capturing that "finished piece" feel rather than a rough sketch look.It's built as a study tool — a way to generate reference images while learning this particular art style. It's aimed at art enthusiasts and students who want hands-on material to study from.Why this one?Other creators are racing to ship fast. We're taking the slower route to get quality right. We're not promising speed — we're promising flexibility and reliability. As of this writing, only one other Anima LoRA targets this same style, and comparing results, our more conservative learning rate kept composition intact while still nailing the target aesthetic<h3 id="roadmap:">Roadmap:</h3>Version - Name - Status - Description:<ul><li>1.0 — Definitive [released]<ul><li>First baseline proof of concept, using only a fraction of the available data</li></ul></li></ul><hr /><h1 id="versioning:">Versioning:</h1>The next section is specific for the version previously mentioned.<hr /><h3 id="v1.0-definitivestatus:-released">v1.0 — Definitive Status: ✅ Released</h3><img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/3e06b623-6fa9-4944-95ab-0da7dae3087d/original=true/3e06b623-6fa9-4944-95ab-0da7dae3087d.jpeg" /><h3 id="details:">Details:</h3>This LoRA is plug-and-play — no keyword required.<h3 id="recommended-settings:">Recommended Settings:</h3><ul><li>Sampler: Euler a, ER SDE</li><li>Steps: <code>20 ~ 50</code></li><li>CFG Scale: <code>4 ~ 7</code></li><li>Clip Skip: <code>1</code></li><li>LoRA Weight: <code>1.0 ~ 2.0</code></li></ul><h3 id="known-issues:">Known Issues:</h3><ul><li>Text-heavy images may generate with a faint watermark near the bottom.</li></ul><h3 id="geek-info:">Geek Info:</h3><ul><li>Trained on a curated set of 205 images using dual captioning (booru-style tags + natural language prompts).</li><li>Training ran on the neme-anima trainer at a max resolution of 1024² — not the ideal resolution for Anima-style models, but a clear step up from our earlier 512² attempts.</li><li>Tags were generated with the WaifuDiffusion (WD 1.4) tagger; natural language captions came from Qwen3-VL-8B, run locally via LM Studio and automated through the neme-anima script.</li></ul><hr /><h2 id="usage-tips">⚙️Usage Tips</h2><ul><li>Prompting style: No departure from standard Anima conventions — dual captioning still gets the best results here.</li><li>Negative prompts: Standard Anima negatives work fine. If odd markings show up at the bottom of an image, adding <code>unknown artist signature</code> or <code>watermark</code> to the negative prompt usually clears it up.</li><li>Good pairings: Works best on base checkpoints without a strong, hard-baked style of their own. Heavily stylized checkpoints will fight it.</li><li>Things to avoid: The source artist's work is NSFW, so keep the "safe" tag in your prompt when generating SFW output, and be specific/detailed to steer clear of unwanted results.</li></ul><hr /><h2 id="faq">❓ FAQ</h2>Q: Can I use this commercially?A: Yes, as long as it doesn't negatively affect the original artist this model is based on, and you follow the licensing terms (no re-uploads/re-prints, credit required).Q: Why do my results look different from the showcase images?A: Most likely the generation UI. CivitAI's built-in generator, A1111, A1111 Forge Neo, and ComfyUI all handle generation slightly differently, for instance, A1111-style UIs generate initial noise on the GPU while ComfyUI does it on the CPU. Even with an identical seed, that alone can shift the output noticeably.Q: Will there be an SDXL / Flux / Pony / Illustrious version?A: Not in the early future.<hr /><h1 id="conclusion">🙏 Conclusion</h1>This project started on a whim — we kept running into Anima's shaky grasp of this artist's style, and the heavy watermarking that came with it. So we trained our own fix, v1.0 should be the only version of this LoRA you'll need for reliable club3-style generations.This ran long, so if you're still reading — genuinely, thank you. I hope it helps you learn, study, and create. Drop your thoughts, questions, or generations in the comments; I read all of them, even if I'm slow to reply.<hr /><h2 id="support-this-model">💖 Support This Model</h2>If you enjoy this model and want to help it grow:<img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/5539db94-d235-4be7-b888-ba9298d5713d/original=true/5539db94-d235-4be7-b888-ba9298d5713d.jpeg" /><ul><li>⭐ Leave a review — it genuinely helps visibility on CivitAI.</li><li>📸 Share your generations in the gallery — I love seeing what people create, and it also helps us collect enough Buzz to train on-site without frying our GPUs.</li><li>💖 Follow this profile — so you won't miss when a new release drops, and so we can climb the leaderboard and reach more people.</li><li>🖌️ Just use the model — simply putting it to work is support in itself. Knowing people are using it keeps us motivated to keep doing this.</li><li>⚡ Buzz donations — if you've got a dozen Buzz to spare, we'll gladly take them. Our hardware isn't powerful enough to run video models reliably, so generating animated cover images for the character models costs us a bit (xd) — even a couple of Buzz can make a difference now and then.</li></ul>Thank you for checking out Club3 Style — happy generating! 🎉<hr />Last updated: [19/07/2026]

Club3 Style

<pre><code>Type: LoRA
Model: Anima
Sampler: Euler a, ER SDE
Steps: 20 ~ 50
CFG Scale: 4 ~ 7
Clip Skip: 1
Trigger Words: none
Recommended Weight: 1.0 ~ 2.0
Target: portraits and illustrations of anime-style characters</code></pre>The weight has range +1 : +2 to let the user choise a more accurate style at the cost of the composition, but it'll be ultimately up the user

Files with SHA256: 539704eef638575d9dfa404ad33e5595ea1b1ef05a8f798f8b70ca90c0a4ee08