User LoRA
3D Cyberpunk Style SDXL-by Happy Ai/赛博朋克/游戏人物场景/ゲーム/게임/サイバーパンクスタイル/사이버펑크/室内设计/环境设计/3D渲染
- Uploaded at
- Jul 10, 2024, 7:33 AM
- Uses
- 2.44k
- Reviews
- Excellent (1)
- Permissions
- Allow image generation & sharing
- Allow users to download this model
- Allow commercial use of generated images
Description
❤️Hello all friends~ ~大家好みなさんこんにちは안녕하세요 여러분BonjourприветHolaनमस्तेمرحبًاWelcome to Happy Ai Art Studio❤️IntroductionThis LoRA is specifically designed to create cyberpunk effects for game scene, cinematic scene, environmental design or those who crave the cyberpunk art. Unlike some SDXL models which often render blurry scenes with a 2d artificial or lines effect, resembling a faked flat shoot rather than realistic render environments, this LoRA model addresses this limitation. It is good at cyberpunk effects of 3D Model, and Unreal Engine 5 render for environmental design. It also does well in dystopian atmosphere, urban city landscapes and sci-fi interiors. This model not only creates a more immersive atmosphere but also provides greater flexibility and give something genuinely new and engaging to creators.I sincerely hope you like the LORA and enjoy using it.NO Triggers(Although no trigger word is required, if the result is not towards the style you like, recommend to adjust/increase the Lora weight or add some style words that you may like into Positive Prompts,for example: "3D, 3D Render, cyberpunk, cyberpunk aesthetic, dystopian atmosphere, urban city, sci-fi, futuristic...etc ."And if needed, add some words into the Negative Prompts to avoid flat style, like for instance :" 2D, Anime, Flat anime, 2D illustration....etc")Tips:LoRA Weight: Between 1-2 range, recommended to start with 1.4 for game scenes, and 1.6 for human.(DON'T USE weight 1 or below 1, it doesn’t work. If over 1.7 , some weird photos may appear) (pls find below some comparison photos for different lora weights)recommend to download the model into your computer to generate images instead of creating it on Civitai only, because due to different settings bettwen Internet and home computer.Base Models:Although various base models can function, models with realistic styles like the SDXL 1.0 series base models, DreamShaper XL are preferred for optimal performance, but you can also try pairing it with other base models to explore different styles like anime style...etc.Resolutions:10241024 Square8321216 Portrait1216832 Landscape1344768 Landscape(recommended)1536640 Landscape(recommended)1408704 LandscapeOther SDXL Res will work fineRecommended Setting for base model SDXL1.0Sampler: Euler a: 30 to 60 stepsCFG Scale: 7Resolutions: 1344*768Clip Skip: 1VAE: SDXL VAEFor business inquiries, custom models, and consultations, please get in touch at chloey2019@163.comIf you need to use this LoRA in your project, please refer to my civitai website: https://civitai.com/user/lovelydream2023logv1: Initial releaseThis is my first model published in CIVITAI, i am so happy to share it with all of you.If you like this model, please have fun to share your artwork in gallery,and consider giving it a ❤️ and leaving a good review. Thank you and wish you all happy eyery day.Enjoy exploring the LoRA!Comparison Example between LEFT: SDXL Original and RIGHT:Lora weight 1.5below is the translation of Chinese language❤️大家好欢迎来到 Happy Ai 艺术创作室❤️简介这款 LoRA 专为游戏3D渲染建模场景、电影场景、环境设计或那些喜爱风的创作者制作赛博朋克的效果而设计的。不同于某些SDXL模型倾向于出平面2D的图,这个Lora是特别擅长3D 立体建模,虚幻引擎 5 渲染效果的赛博朋克风的环境设计。这个模型不仅能创造出一种更具沉浸感的氛围,更高级的质感,而且还提供了更大的灵活性,并为创作者提供了真正新颖和引人入胜的东西。我诚意希望你们会喜欢这个lora,和乐于分享它。不需要触发词(虽然不需要触发词,但如果结果不符合您的喜好,建议调整/增加 Lora 权重或将一些您可能喜欢的风格词添加到正向提示中,例如:“3D、3D 渲染、赛博朋克、赛博朋克美学、反乌托邦氛围、城市、科幻、未来主义……等等。”"3D, 3D Render, cyberpunk, cyberpunk aesthetic, dystopian atmosphere, urban city, sci-fi, futuristic...etc ."如果需要,可以在负向提示中添加一些词以避免扁平风格,例如:“2D、动漫、扁平动漫、2D 插画……等等”)2D, Anime, Flat anime, 2D illustration....etc")温馨提示:LoRA 权重:介于 1-2 之间,建议使用1.4的权重。(请不要使用权重 1 或者低于 1 ,是没有效果的。如果超过 1.7 ,可能会出现一些奇怪的图)详细可见底部的权重对比图。基本模型:搭配不同的SDXL类别的基本模型都行,写实模型例如 SDXL 1.0 系列基本模型、DreamShaper XL都不错,但您也可以尝试和其它不同风格的模型搭配使用,以探索不同的风格,如动漫风格等。(因为这是基于SDXL底模训练的,所以请不要使用1.5的模型哦)图片规格:10241024 (方图)8321216 (竖图)1344768(横图)1536640(横图)1408704(横图)其他 SDXL 的常规规格都可以基本模型 SDXL1.0 的推荐设置采样器:Euler a:30 步CFG 值:7分辨率:10241024CLIP Skip:1VAE:SDXL VAE如需业务咨询、定制模型和咨询,请联系邮箱chloey2019@163.com声明:这个模型禁止用于任何形式的商业用途,仅供交流学习使用。日志v1:初始版本这是我在CIVITAI上发布的第一个模型,很开心与大家分享O(∩_∩)O如果您喜欢这个模型,请多多返图哦!请点赞❤️,留下好评,多多转载给您的朋友们。祝大家每日开心~~