ForRealXL Lightning v1全能模型

使用介绍

关于v1版本

如果你看到 V0.5 的发布，你就可以想象我为了创建一个在各个方面都真正更好的版本有多难。

这大概是 100 个不同版本的结果。

更好的面孔（和更多的差异）（不完美）
对于人和面孔，你不应该使用 DDPM。使用 DPM++ 3M SDE 或 Euler

更多细节

低 cfg 时颜色更佳

推荐设置：

采样方法：DDPM（我最喜欢的）、Euler、DPM ++ 3M SDE

采样步骤：10（更多也很少会产生影响。）

CFG 等级：1-4（越低，越现实，我更喜欢 1-1.8）

Upscaler：SwinIR _ 4x（这是我一直以来最喜欢的）

（Hires steps：15；Denoise：0.28，upscale by 1.5 – 4（取决于您的 GPU））

📌使用sdxl _ vae，下载已包含

使用提示：

CLIP SKIP: 2

所有图像均在SD Forge中使用Nvidia RTX 4090生成

请测试其他设置并与社区分享，这样我将了解设置并进行优化

请尝试不同的重量。Loras 确实改变了这个模型！

建议LoRA

Detail Tweaker XL细节增强

Midjourney v5/v6模仿画风

黑暗氛围/明暗对比摄影增强XL

动漫风格背景XL/风景/场景LoRA

作品参照

Game art, Border Collie, Spring, soft focus, Hopeful, Bastardcore, Starlight, Depth of field 270mm, Cold Colors, art by Eleanor Fortescue-Brickdale, Ferdinand Hodler, close-up, magical, cinematic light, luxurious, dynamic dramatic atmosphere, illustrious quality, luxurious sharp focus, flowing, stunning detail
Negative prompt: autumn, low contrast, photorealism, modernist, minimalist, moody, dark, lonesome, night, monochromatic, washed out, full body shot, bokeh, blurry, blur, photograph, figurativism,
Steps: 17, CFG scale: 1.4, Sampler: DDPM, Seed: 2954785090, VAE: sdxl_vae.safetensors, Size: 720x1280, Model: ForRealXL-V1.0, Version: f0.0.17v1.8.0rc-latest-278-gbfee03d8, Model hash: ff0240ddf7, Hires steps: 14, Hires upscale: 1.5, Hires upscaler: SwinIR_4x, Variation seed: 2631023952, Denoising strength: 0.16, Hires negative prompt: [object Object], Variation seed strength: 0.86, Clip skip: 2

photograph, portrait, close up of a wearing Cashmere Wide-leg jumpsuit and sandals, (Teuta:1.2) , Alien, Freckles, (designed by Elke Vogelsang:1.3) , commercial, close-up, <lora:aidma-Image Upgrader-v0.1:0.65> photo, professional, realistic, detailmaximizer, max details, <lora:aidmaFluxStyleXL-v0.2:0.55> flux style
Negative prompt: blurry, low quality, deformed, boring
Steps: 17, CFG scale: 2, Sampler: DPM++ 3M SDE, Seed: 103834213, VAE: sdxl_vae.safetensors, Size: 720x1280, Model: ForRealXL-V1.0, Version: f0.0.17v1.8.0rc-latest-278-gbfee03d8, Model hash: ff0240ddf7, Hires steps: 14, Hires upscale: 1.5, Hires upscaler: SwinIR_4x, Variation seed: 3143257008, ADetailer model: face_yolov8n.pt, ADetailer version: 24.6.0, Denoising strength: 0.16, ADetailer mask blur: 4, ADetailer model 2nd: hand_yolov8n.pt, ADetailer confidence: 0.3, ADetailer dilate erode: 4, ADetailer mask blur 2nd: 4, Variation seed strength: 0.86, ADetailer confidence 2nd: 0.3, ADetailer inpaint padding: 32, ADetailer dilate erode 2nd: 4, ADetailer denoising strength: 0.4, ADetailer inpaint only masked: True, ADetailer inpaint padding 2nd: 32, ADetailer denoising strength 2nd: 0.4, ADetailer inpaint only masked 2nd: True, Clip skip: 2

,professional car porn photo, low angle view, super tuning car, Lamborghini, shiny green, back view
<lora:aidma-Image Upgrader-v0.1:0.55>
Negative prompt: blurry, low quality, deformed, boring
Steps: 18, CFG scale: 1.4, Sampler: DDPM, Seed: 3538381193, VAE: sdxl_vae.safetensors, Size: 720x1280, Model: ForRealXL-V1.0, Version: f0.0.17v1.8.0rc-latest-278-gbfee03d8, Model hash: ff0240ddf7, Hires steps: 14, Hires upscale: 1.5, Hires upscaler: SwinIR_4x, Variation seed: 3837785901, ADetailer model: face_yolov8n.pt, ADetailer version: 24.6.0, Denoising strength: 0.16, ADetailer mask blur: 4, ADetailer model 2nd: hand_yolov8n.pt, ADetailer confidence: 0.3, ADetailer dilate erode: 4, ADetailer mask blur 2nd: 4, Variation seed strength: 0.86, ADetailer confidence 2nd: 0.3, ADetailer inpaint padding: 32, ADetailer dilate erode 2nd: 4, ADetailer denoising strength: 0.4, ADetailer inpaint only masked: True, ADetailer inpaint padding 2nd: 32, ADetailer denoising strength 2nd: 0.4, ADetailer inpaint only masked 2nd: True, Clip skip: 2

建议LoRA模型