UnitySpectrum_XL is a LoRA designed to generate natural interactions between two people, with stable poses, believable physical contact and no limb fusion.It was trained on 12 main poses, each with multiple variations, to ensure consistency and generalization:friendly_hugside_hugselfie_togetherhandshakeshoulder_toucharm_holdclose_poseleaning_posecouple_posesupportive_posesoft_embracetogether_stance👉 generalizes perfectly to any combination (white/Black/Asian/Latina etc.).⭐ Recommended checkpointLoRA has been optimized for:👉 Brave Studio XL — Simple / Normal schedulerIt also works on SDXL Base, but Brave Studio guarantees:more stable geometrymore natural eyesmore consistent skinless noise⭐ Recommended weightLoRA works best between:👉 0.7 and 1.00.7 → softer, more natural1.0 → more precise on the poseNot recommended to exceed 1.1.<h3 id="example-prompt:">Example Prompt:</h3>1_usxl, white man and Black woman, friendly_hug, realistic photo, 50mm lens, natural light, detailed skin, soft shadows, clean iris texture, sharp eyes2_usxl, Asian woman and Latina woman, selfie_together, candid moment, smartphone angle, natural lighting, realistic skin texture, face in focus3_usxl, Black man and white man, handshake, business casual, indoor lighting, realistic photo, 50mm, detailed hands, clean geometry4_usxl, white woman and Black woman, side_hug, warm moment, soft light, realistic skin, balanced lighting, sharp focus<h3 id="negative-prompt:">Negative prompt:</h3>blurry eyes, glossy eyes, distorted pupils, deformed hands, fused fingers, extra limbs, overexposed skin, low detail, bad anatomy<h2 id="the-cover-images-were-created-on-mage.space-for-convenience-however-magespace-does-not-preserve-metadata-so-i-include-example-prompts-here-to-guide-proper-usage">The cover images were created on mage.space for convenience; however, MageSpace does not preserve metadata, so I include example prompts here to guide proper usage</h2>

UnitySpectrum_XL

UnitySpectrum ZIT is the enhanced version of the original SDXL LoRA, designed for higher precision, cleaner details, and more stable identity rendering. It maintains the same artistic philosophy as the SDXL version, but delivers sharper facial structure, improved micro‑details, cleaner poses, and more consistent character fidelity.The LoRA performs best between 0.6 and 0.9:<ul><li>0.6–0.7 → softer, more natural realism</li><li>0.8–0.9 → stronger identity, more defined features, slightly more “cinematic” look</li></ul>Everything else remains fully compatible with the SDXL version — same style, same versatility — but with ZIT‑level clarity and accuracy.

Files with SHA256: 446fcdfc033f26aae49749b7aee62b43770768a8be972ab2cd5e81d16c2d96b8