<h2 id="inspiration">Inspiration</h2>🎾 Beata Love is a fictional character. Initial inspiration came from <a target="_blank" rel="ugc" href="https://civarchive.com/images/3373177">this AI-generated image</a> by <a target="_blank" rel="ugc" href="https://civarchive.com/user/AleksPls">AleksPls</a> rendered using the <a target="_blank" rel="ugc" href="https://civarchive.com/models/101888/beautyfool?modelVersionId=124055">BeautyFool v1.5</a> checkpoint. Further refinement—especially toward a more realistic style—and a training dataset were then built using <a target="_blank" rel="ugc" href="https://deepmind.google/models/imagen/">Google Imagen 4</a> in the <a target="_blank" rel="ugc" href="https://labs.google/fx/tools/flow">Google Flow</a> creative studio.Beata (pronounced beh-AH-tah) is a 20-year-old French native of Southern European and Mediterranean descent. While she can seem a bit guarded at first, she quickly reveals a captivating blend of warmth, wit, and independence. An agile presence on the tennis court, Beata relishes nothing more than balancing a high-energy match with a deeply relaxing spa treatment afterward.<hr /><h2 id="z-image-turbo">Z-Image —Turbo</h2><ul><li>Diffusion Model: <a target="_blank" rel="ugc" href="https://civarchive.com/models/2168935/z-image-turbo">Z-Image Turbo</a>, or try <a target="_blank" rel="ugc" href="https://civarchive.com/models/2342797/z-image-base">Z-Image [Base]</a> —or try a finetuned model, with very good results paired with <a target="_blank" rel="ugc" href="https://civarchive.com/models/2218365/cyberrealistic-z-image-turbo?modelVersionId=2780276">CyberRealistic Z-Image Turbo v3.0</a></li><li>LoRA strength: 1–1.6</li><li>Sampler: dpmpp_2s_ancestral</li><li>Scheduler: simple or beta</li><li>Steps: 4–12</li><li>CFG Scale: 1, or try 1.2–3; CFG Norm: 1–1.12</li><li>Shift: 5–7 (first sampler) and 0.5 (second sampler)</li><li>Trigger word (for superior results): ATRX_BEATA_ZIMAGE</li><li>Negative prompt (as needed): "Asian woman"</li></ul>

Beata Love

Prompts must include trigger word for superior results: ATRX_BEATA_ZIMAGE. With proper settings, produces decent images using the original diffusion model, but some of the newer finetunes may produce superior results: <a target="_blank" rel="ugc" href="https://civitai.com/models/2218365/cyberrealistic-z-image-turbo?modelVersionId=2780276">CyberRealistic Z-Image Turbo v3.0</a> is quite nice! Can produce artistic nudes pretty well on its own, but can be further assisted when paired with a NSFW LoRA. With structured prompting, other characters described in the scene should somewhat retain their own distinctive features.The dataset for training consisted of 75 high-quality photorealistic images AI-generated using Imagen 4 in Google Flow, with references drawn from <a target="_blank" rel="ugc" href="https://civitai.red/images/3373177">this image as initial inspiration</a> and resampled rendered images of the same character. SeedVR2 in ComfyUI was used to upscale images from 1024x1024 to 1280x1248. Most captioning was borrowed from work on a previous character using Qwen Chat, but then checked and tweaked through human curation.No known reference to any real person was included in any part of the training.

Files with SHA256: dec28d375c43d661d578a7345d993c0df15039077b6b6025a1089adb428e8309