前言
在9月23日,我在网上无意中看到了Flux.1结合真人LoRa生成的图像,效果已经可以媲美真人照片。于是在下班后,我利用公司的算力资源进行了测试。( 附录有所需资源下载地址! )
评测环境
- 系统 Ubuntu 24.04.1 LTS
- 显卡 NVIDIA GeForce RTX 4090 24G
- 模型 FLUX.1-dev
测试流程
- 下载FLUX.1-dev模型放到 ComfyUI/models/unet/
- 下载 CLIP 模型放到 ComfyUI/models/clip/
- 下载 VAE 模型放到 ComfyUI/models/vae/
- 下载lora放到 ComfyUI/models/loras
- 将工作流程拖入ComfyUI(需进行模型配置选择)
测试结果
Promot
This is a high-resolution photograph, A radiant model with clear, dewy skin and expressive eyes stands in a sun-dappled forest, her figure adorned in a vibrant, bohemian dress, her hand gently touching a tree as she exudes a sense of harmony with nature,Young Asian cute gril.
一个容光焕发的模特站在阳光斑驳的森林里,她的皮肤晶莹剔透,眼神富有表现力,她的身材装饰着充满活力的波西米亚风格的衣服,她的手轻轻地抚摸着一棵树,散发着与自然和谐相处的感觉
结果展示
测试提示词
This is a high-resolution photograph, a fashionable beauty with clear, dewy skin and a symmetrical face stands in a tranquil library, her figure adorned in a soft, cable-knit sweater and a flowing skirt, her expression thoughtful as she peruses the shelves, her posture elegant and poised,Young Asian woman.
这是一张高分辨率的照片,一个时尚的美女站在一个安静的图书馆里,她的皮肤晶莹剔透,匀称的脸,她的身材穿着柔软的针织毛衣和飘逸的裙子,她仔细阅读书架时的表情若有感触,她的姿势优雅而镇定
This is a high-resolution portrait of a modern style, featuring a young woman. She is wearing a tight dress printed with English and patterns, the pattern of which resembles a newspaper. Her hair is black, falling over her shoulders, and she has one hand resting on the back of her head while the other hangs naturally by her side. Her eyes are fixed on the camera, and her expression is calm. She appears to be between the ages of 20 and 30, slim in figure, with delicate features, and fair skin. Her eyes are brown, and her lips are tinted with a subtle shade of red lipstick. Her hair falls to her shoulders, with a deep brown color, styled in loose waves without bangs. The background is dark, giving a mysterious feel,Young Asian woman.
这是一幅现代风格的高分辨率肖像画,主角是一位年轻女子。她穿着一件印有英文和图案的紧身连衣裙,图案像报纸。她的头发是黑色的,披在肩上,一只手放在后脑勺上,另一只手自然地垂在身体两侧。她的眼睛盯着镜头,表情平静。她的年龄在20到30岁之间,身材苗条,五官精致,皮肤白皙。她的眼睛是棕色的,嘴唇上涂着淡淡的红色唇膏。她的头发披在肩上,呈深棕色,波浪状,没有刘海。背景是黑色的,给人一种神秘的感觉
This is a high-resolution photograph, A radiant beauty with flawless skin and delicate features poses elegantly in a chic, urban coffee shop wearing a trendy velvet jumpsuit, her expressive blue eyes sparkling with enthusiasm as she gently sips her latte,Young Asian woman.
这是一张高分辨率的照片,一个皮肤无暇、五官精致的光彩照,穿着时髦的天鹅绒连衣裤,在一家别致的城市咖啡店里优雅地摆姿势,她富有表现力的蓝色眼睛闪烁着热情,轻轻地喝着拿铁咖啡
This is a clear mobile selfie photo with a daily style. A fashion-forward woman with a glowing complexion and sharp features lounges on a vintage chaise longue in a sunlit conservatory, her form draped in a luxurious silk robe, her expression one of serene contemplation,Young Asian woman.
这是一张日常风格的清晰手机自拍。在阳光明媚的暖房里,一个面色红润、五官鲜明的时尚前卫的女人坐在一张老式躺椅上,她的身体披着一件奢华的丝绸长袍,她的表情是一种安详的沉思
总结
Flux.1 模型在图片生成方面的细节处理已经达到了令人难以辨认的地步,以至于我们几乎无法区分出这些图片究竟是真实的摄影作品还是AI技术的产物。这种高度逼真的图像生成能力不仅展示了深度学习和人工智能领域的最新进展,也引发了关于真实性、版权和创意所有权等一系列复杂问题的讨论。
Flux.1 模型利用了先进的算法和大量的训练数据,能够捕捉到照片中微妙的光影效果、纹理细节以及色彩的丰富层次。这使得生成的图片在视觉上与真实拍摄的照片几乎没有差别,甚至在某些情况下,AI生成的图片可能在细节上更加完美无瑕。
然而,这种技术的突破也带来了一些挑战。首先,随着AI生成图片的质量越来越高,人们越来越难以判断一张图片的真实性。这可能会对新闻业、社交媒体和其他依赖视觉证据的行业造成影响,因为观众可能会对图片的真实性产生怀疑。其次,AI生成的图片可能会侵犯真实摄影师的版权,因为他们的作品可能被用作训练数据而没有得到适当的授权或许可。此外,如果AI生成的图片被用于商业目的,那么创意所有权的问题也将变得复杂起来。
尽管如此,Flux.1 模型在图片生成方面的成就仍然是值得称赞的。它为我们提供了一种新的工具来创造视觉内容,无论是用于艺术创作、游戏设计还是虚拟现实等领域。同时,它也推动了人工智能技术的发展,让我们对未来的可能性充满了期待。