Not always perfect.. better than nothing Perfect SEApony is an SDXL LoRA that can turn almost any character (or pony) into a pony-mermaid.<h2 id="quick-tip-(fast-setup)">Quick tip (fast setup)</h2>Download the LoRA, write your prompt as usual, and add this at the end (A1111 / Invoke / most UIs):<code>perfect_SEApony, seapony, merpony, (double_fin_wings:1.5), (central_dorsal_fin:1.0), &lt;lora:perfect_SEApony_R1:0.75&gt;</code>If it doesn’t change much, increase the LoRA weight. Recommended range: <code>0.15–1.0</code> (PlantMilk / YiffyMix, etc.) For PonyDiffusion you may prefer ~1.0–1.5, but feel free to try lower. Long description (you can skip) Main activator(paste into prompt): <code>perfect_SEApony, merpony, seapony</code> Optional boosters (extra tags): <code>central_dorsal_fin</code> — adds the central dorsal fin<code>fin_ears_v1</code> — canonical fin-ears (recommended / most stable)<code>fin_ears_v2</code> — shell-like ears (can be less stable; try <code>fin_ears_v1</code> first)<code>double_fin_wings</code> — double wing-fins That’s it! You can play with weights. Example: if <code>double_fin_wings</code> is missing, try <code>(double_fin_wings:1.25)</code> → <code>(double_fin_wings:1.75)</code> (or higher). <code>central_dorsal_fin</code> can sometimes create extra fin/hair-like shapes on the tail. Quick workaround: put <code>central_dorsal_fin</code> into negative, then do a small inpaint to fix the fin. If you want to support me or this model, you can do it via my Patreon: <a rel="ugc" href="https://www.patreon.com/cw/TidalSeashell">link</a>. (I don’t post there very often yet, but if you support me, I’ll be able to release new versions of this LoRA much faster — so it’s worth it.) This model is designed as a modular system. You’re currently on the page for the base LoRA. More LoRAs will be released later as [PerfectSEApony] addons… P.S. If you want a specific character with this LoRA, you’ll need a separate character LoRA. Try mixing them like [character ~0.65] + [Perfect SEApony ~0.45], then experiment with the weights. I tested this LoRA for over two weeks before release. This is my main LoRA that I use for most of my work

Files with SHA256: 441bcf5f5e80280b210b5557ac953b012d34141ab2524b32b98359ea96eb2f19