1109 nai学习-画面自然语言叙事化描述

最近在b上看到了一些nai的图片,作者也是很有开源精神的把prompt给公开了,故此专门学习了一些些关于nai v5的玩法

作者是https://space.bilibili.com/265841740 我这边就拿他的图举例了

这是我这次主要学习的图片:
image

提取了正向提示词 然后大概分析一下

1girl,alternate costume,year 2024, year 2025, offical outfit ,a produce vendor leaning across a neighborhood market stall to pass a small paper bag of tomatoes to a customer, one hand cupping the bottom of the bag and the other holding the folded rim, the customer’s fingers entering from the right foreground, the vendor looking down to check the handoff, wooden crates, scallions and a small mechanical scale clustered along the counter, eye-level waist-up framing, clear sunny morning, unobstructed sunlight coming from the upper right through the open side of the stall, warm ivory light on the vendor’s cheek and hands, narrow highlights along tomato skins, a sharply defined awning shadow dividing the sunlit counter from the cool blue-charcoal interior, muted daylight bounced from the pavement revealing the shaded eyes and lettering on the scale, matte paper texture, restrained reflections, lively neighborhood setting, horizontal composition, painterly anime illustration, best quality

nai v4之前我作图都是靠的tag堆砌,导致抽卡稳定系数一般般,最近也是借助一些群u免费蹭到了这个v5的api,所以大概也研究了一下

首先naiv5是可以支持自然语言描述的

这也是我这次学习的核心部分 用自然语言叙事化表达一个画面

我们刨去tag部分 可以获得一个很长的一个主体叙事描述

a produce vendor leaning across a neighborhood market stall to pass a small paper bag of tomatoes to a customer
这里要尽量的精简 但是一定要点明

  • 主体:produce vendor
  • 动作:leaning across
  • 事件及对象:pass a small paper bag of tomatoes to a customer

接下来是主体的细节描述

one hand cupping the bottom of the bag and the other holding the folded rim, the customer’s fingers entering from the right foreground, the vendor looking down to check the handoff

四句话交代了四个人物细节:

  • 一只手托住纸袋底部one hand cupping the bottom of the bag
  • 另一只手扶住折叠的袋口and the other holding the folded rim
  • 画面右侧前景伸入顾客的手指the customer’s fingers entering from the right foreground
  • 摊贩视线下看于"handoff"the vendor looking down to check the handoff
    这就是自然语言相对于孤立 Tag 的一个优势:它可以把动作和动作的目的联系起来

然后是一个很长的场景描写

wooden crates, scallions and a small mechanical scale clustered along the counter
三件物品勾勒出场景 甚至用clustered along the counter指出了分布关系

再后面是镜头描写 eye-level waist-up framing

最后才是最关键的,我目前还在学习的光影描写

clear sunny morning, unobstructed sunlight coming from the upper right through the open side of the stall, warm ivory light on the vendor’s cheek and hands, narrow highlights along tomato skins, a sharply defined awning shadow dividing the sunlit counter from the cool blue-charcoal interior, muted daylight bounced from the pavement revealing the shaded eyes and lettering on the scale
  • clear sunny morning整体环境条件
  • unobstructed sunlight coming from the upper right through the open side of the stall光线的主体关系 即大致走向
  • warm ivory light on the vendor’s cheek and hands光线的色彩倾向和受光区域
  • narrow highlights along tomato skins高光
  • a sharply defined awning shadow dividing the sunlit counter from the cool blue-charcoal interior阴影处理
  • muted daylight bounced from the pavement revealing the shaded eyes and lettering on the scale反射光以及反射光照射方向

因为是作者风格的关系吧,这个光影描写拔高了整体的观感质量 但是我目前学习到的是场景的叙事化描述方式

一个画面,先分主次,然后找细节,最后打上光影

  1. 核心动作 决定画面主体在做什么 点明

  2. 姿势细节 让姿势更加明确

  3. 细节修饰 丰富画面细节

    以及

  4. 镜头 镜头与构图

  5. 光影 光线与曝光

说实话这个部分我真得补补课了,这俩是典型的摄影学基础,我学的有点晕头转向,确实是需要系统性学习了

另外很喜欢作者的标题: 构图之道 就在其中

posted @ 2026-11-09 07:28  伊芙宁西普  阅读(10)  评论(0)    收藏  举报