Currently, this model is a merge of: MIDNIGHT Illustrious https://civitai.com/models/981618/midnight-illustrious Nova Unreal XL https://civitai.com/models/982047/nova-unreal-xl iLustMix https://civitai.com/models/1110783/ilustmix and a 2.5D style LoRA of mine. (THIS INFORMATION IS SUBJECT TO CHANGE AS I USE THE MODEL MORE) *I recommend V2 for non-female subjects.* Quality tags: masterpiece, best quality, hires, absurdres, newest, aesthetic, (photorealistic, realistic), *IF YOUR PROMPT IS LONG PUT AT START AND END* Without the photorealistic/realistic style tags in positive you wont get the same style as the previews. **For V3, I recommend using (realistic) if your prompt is short and (photorealistic, realistic) if your prompt is longer. The default style (no style tags) of the model is a mix between 2.5D and Anime (Anime face, 2.5D body, realistic lighting). The purpose of the merge is was to achieve the style shown in the preview images. The anime-screencap style tag is powerful if you're looking for a more flat toned anime/manga styling. Example prompt: masterpiece, best quality, hires, absurdres, newest, aesthetic, (photorealistic, realistic), detailed background:1.2, rating_safe, 1girl, dynamic pose (rating_safe, rating_questionable & rating_explicit all work well in keeping gens aligned to your preference. I encourage you to try them) Recommended Embedding: https://tensor.art/models/793668072050352375 Parameters Steps: 5 - 40 (I recommend at least 15 steps but I have had decent results on as low as 5) CFG: 4 - 10 (If you get blown out lighting, lower your CFG scale and try again) Hires Upscaler: 4x-UltraSharp, 4x_NMKD-Superscale-SP_178000_G, 4x_foolhardy_Remacri, R-ESRGAN 4x+ Anime6B Hires Steps: ≥10 Sampler: DPM++ 2M Karras, DPM++ 2M SDE Karras, Euler A, Heun Adetailer: face_yolov8m, hand_yolov8s