ComfyUI-Minitools
ComfyUI Mini Toolkit – Helps reduce node redundancy. Let’s see if there’s time to keep it updated regularly.
Nodes (16)
An LLM Prompt Optimizer That Speaks Bernini's Task Language
A dropdown that hands you a latent, width and height — no calculator required
Chinese ⇄ English prompt translation that also knows which way you typed it
From a Color Picker String to Three Separate Numbers
Get Your Cutout Out of All That Empty Space
Every Color, Squeezed Into One Number
Actually move, rotate and scale layers by hand before ComfyUI keeps going
Any Image Into a Mask, With a Brightness Knob
The 22-port hose that turns one H3 pipe into wires you can actually use
The node that writes MiniMax H3 prompts so you don't have to learn the shot-timestamp grammar
One node from rough idea to MiniMax H3 conditioning — text, first-frame, or all-reference
One config node to feed every LLM-powered node in the pack — any OpenAI-compatible API will do
Swap the words your model won't touch for ones it will, one random pick at a time
Stop typing the same system prompt twice — Bernini's task roles, one dropdown away
A Prompt Translator That Makes You Edit the Source Code
Two Sliders That Just Pass Numbers Through
comfyui-minitools
comfyui小工具集,避免节点冗余,看有没时间持续更新吧~
安装方法:
git clone命令克隆本项目,即可完成所有节点的安装。
1、color2rgb 用于配合comfyui layer style中颜色选择器节点,转换成RGB,且分别输出R、G、B值
2、promptsTranslateEN 提示词翻译节点,将输入的内容转换成英文,用的百度翻译api,所以需要实现准备好appid(可在百度翻译官方免费申请)
节点演示:
3、RGB or HEX Convert to DEC 将十六进制颜色码和RBG转换成十进制
4、CropTransparentEdges 按透明区域裁剪
5、ImageToMaskWithAlpha 将图片转换成带有alpha通道的遮罩
6、NumericSlider数字滑块,包含浮点数和整数
7、PromptOptimizer 基于字节Bernini官方仓库中提示词增强优化而来,适用于提示词反推和优化、扩写(官方演示)
支持单图反推:可用于出图、出视频、首尾帧视频、视频编辑
多图反推:多图、视频反推,用于出视频、首尾帧视频、视频编辑
反推内容会自动根据Bernini官方提示词预设模板完成补全和优化,也就是说不只是用于Bernini,可以完全独立出来作为一个提示词生成器来用。 节点依赖阿里百炼大模型api,默认使用的是多模态模型,如果想节约token,可以分纯文本大模型和带有VL标识的视觉模型或多模态大模型,点这里可注册阿里云百炼后创建apikey,就可以免费获得不同模型100万token使用额度 <img width="1674" height="1056" alt="image" src="https://github.com/user-attachments/assets/10a13d69-81df-437d-af6d-cfa7e74a57b3" />
8、Minimax H3 Studio 海螺H3工作台 利用这个节点只用一套工作流完成文生视频、图生视频(含首尾帧)、全能参考;支持所有兼容openai api协议的接口,包括但不限于本地ollama和阿里百炼api。 支持在提示词输入框中输入@调用上传的图片,支持点击图片自动填入图片名到提示词。 在不适用llm接口的情况下,可以直接当作提示词编辑器离线使用,不走api来跑视频。 <img width="858" height="756" alt="image" src="https://github.com/user-attachments/assets/a597e3fb-457d-4122-9958-eac93ff00f27" />