Comfyui-Flux写实人像摄影风格探索

在一些小伙伴的建议下,我最近开始着手整理ComfyUI的相关内容。其实之前就一直在关注这个工具,但由于工作繁忙,一直没能抽出时间去总结。

与SD webui不同的是,comfyui有着极高的自由度和灵活性,支持高度的定制化和工作流复用,同时对系统配置的要求较低,并且能够加快原始图像的生成速度。

然而,由于它拥有众多的插件节点,以及较为复杂的操作流程,学习起来相对困难。 另一方面,WebUI 特点是拥有固定的操作界面,使得其易于学习和快速上手。经过一年多的发展,它已经建立了一个成熟且稳定的开源生态系统。但相比 ComfyUI,在性能方面可能稍逊一筹,且在工作流程复制方面存在局限性,用户需在每次操作时手动进行设置。

本次使用Comfyui是官网的版本,由于comfyui安装节点和模型有大量的坑。我已经将所需的节点、插件、模型都打包到一起,解压即可使用。(📢唯一需要注意的是可能需要安装cuda12.1)

工作流是网上找的Flux工作流。介于之前在群里分享的FLUX整合包显存要求太大了,所以这次使用FP8量化后的低显存的Flux模型。

这个版本与其他ComfyUI版本(如秋叶版本)可以共存,相互独立运行,不会产生冲突。可以放心安装使用,无需卸载已有的ComfyUI版本。

配置要求

Windows 8G显存 N卡

MAC (没做🤦🏻‍♀️)

快速上手

解压整合包,双击整合包目录下的启动器.exe。

出现启动信息,等待一段时间,会弹出网页。

Flux.json文件拖拽到网页内

可以看到工作流已经导入

这里调整生成步数。

在这里选择画面尺寸

来到这个地方输入提示词

然后点击执行队列按钮,即可开始生图。

生成好的图片会在下方,点击图片右键另存为可以保存到本地。

总的来说操作步骤不难。

提示词参考

这里放上我觉得还不错的提示词供大家参考。

The image captures a close-up portrait of a young woman bathed in warm afternoon light, highlighting her natural features and genuine expression. She's wearing a cozy cream knit cardigan that softly drapes around her shoulders, with a simple white t-shirt underneath, creating an effortless everyday look. Her hair frames her face in gentle waves, catching the golden sunlight.

The frame focuses on her upper body as she sits near a window, the warm light creating a soft glow on her skin and adding depth to her features. Her expression is relaxed and authentic, with a subtle smile playing at her lips. The background is intentionally soft and minimal, allowing her natural beauty and the interplay of light and shadow to take center stage.

The image captures a delicate half-length portrait of a young woman in gentle afternoon sunlight. She's wearing a light pastel blue silk blouse with subtle floral details, its soft fabric creating elegant drapes around her shoulders. The neckline features a small bow or ruffle detail, adding a feminine touch without being overstated.

Her pose is natural and graceful, perhaps with her head tilted slightly, as warm light cascades across her features. Her hair is styled in soft waves, a few strands catching the golden light. Her expression is serene and tender, with a gentle gaze that conveys warmth. The background is softly blurred, creating an intimate atmosphere that enhances the overall gentle mood of the portrait.

The image captures a young woman in a serene forest setting, where dappled sunlight filters through the green canopy above. She's wearing a flowing midi-length dress in sage green or soft ivory, with delicate long sleeves and a subtle floral pattern that harmonizes with the natural surroundings. The dress moves gently with the forest breeze, creating elegant movement in the frame.

The portrait is framed as a three-quarter shot, showing her natural pose among the trees. Soft afternoon light creates a magical atmosphere, casting gentle spots of sunshine around her and highlighting her features. Her hair is styled in loose, natural waves that catch the golden light. Her expression is peaceful and dreamy, with a gentle smile that mirrors the tranquil forest setting. The background features a soft bokeh effect of the trees, creating depth while maintaining focus on her serene presence.

整合包获取

👇🏻👇🏻👇🏻下方下方下方👇🏻👇🏻👇🏻

关注公众号,发送【Comfyui人像】关键字获取整合包。

如果发了关键词没回复你!记得看下复制的时候是不是把空格给粘贴进去了!

制作不易,如果本文对您有帮助,还请点个免费的赞或在看!感谢您的阅读!

相关推荐
春日见6 分钟前
RL精华知识
人工智能·机器学习
东方佑28 分钟前
波动力学语言模型(Wave Dynamics Language Model, WDLM)
人工智能·语言模型·自然语言处理
John_ToDebug36 分钟前
CLAUDE.md 与 Skills 的区别:一张表彻底分清
人工智能·经验分享·ai
程序员cxuan37 分钟前
我花了两天时间,终于把 Codex 额度掉太快的问题整明白了!!
人工智能·后端·程序员
IT_陈寒39 分钟前
Vue这个动态响应坑把我整不会了
前端·人工智能·后端
feasibility.1 小时前
ROS2+Gazebo+VLM服务:纯仿真环境下的具身智能闭环系统| 大脑-小脑分离控制
人工智能·机器人·ros·仿真·具身智能·vla·vlm
lqqjuly1 小时前
自动驾驶仿真平台:理论、架构与实践
人工智能·机器学习·自动驾驶
“码”力全开1 小时前
解耦异构算力与多协议接入:基于Docker与源码交付的开源企业级GB28181/RTSP边缘计算AI视频管理平台架构深度解析
人工智能·docker·开源
J2虾虾1 小时前
Spring AI Alibaba文档
java·人工智能·spring
Mr数据杨1 小时前
【CanMV K210】传感器实验 U 型光电传感器遮挡检测与 LED 提示
人工智能·硬件开发·canmv k210