Generative AI

• python==3.12.3
• torch==2.8.0(+cu128)
• triton==3.4.0
• torchvision==0.23.0(+cu128)
• torchao==0.13.0
• torchaudio==2.8.0
• flash-attn==2.8.3

---

ComfyUI

ComfyUI
git clone https://github.com/comfyanonymous/ComfyUI
cd ComfyUI/custom_nodes

Custom Nodes:

• ComfyUI-Manager git clone https://github.com/ltdrdata/ComfyUI-Manager
  

• ComfyUI-WanVideoWrapper git clone https://github.com/kijai/ComfyUI-WanVideoWrapper
  

To run ComfyUI
cd ~/ComfyUI
python main.py

GUI: http://127.0.0.1:8188

---

1. Generative Image

NanoBanana

Prompt:

請生成年輕女性
臉部特徵與妝容
​五官: 擁有清秀、柔和的五官，臉型屬於鵝蛋臉或偏瓜子臉。豐滿上圍凸出
​眼睛: 眼神清澈且專注，是單眼皮或內雙眼皮，帶有自然的東方美感。
​膚質/妝容: 膚色白皙、透亮，妝容非常自然、輕薄，呈現出**「偽素顏」或「裸妝」的效果，強調肌膚的光澤感和無瑕疵**。
​💇 髮型
​髮色/髮質: 髮色是深棕色或自然黑，髮質看起來柔順且有光澤。
​造型: 髮型是半紮式馬尾（或公主頭），將上半部的頭髮向後梳起，展現出俐落感；同時保留了幾縷髮絲自然地垂落在臉頰兩側，增添了柔美的氣息。
​👚 服裝與整體風格
​服裝: 穿著一件米色或淺裸色的上衣，材質似乎是輕薄的針織或有細紋理的布料，
​風格: 整體風格是清新、自然、優雅，模特兒的形象。
​🌟 簡潔重點總結
​她是一位外型清新、氣質溫柔的女性。擁有白皙透亮的自然裸妝，搭配柔順的深棕色半紮髮，整體散發出優雅而專注的氛圍。在咖啡廳

---

Imagen4

Prompt:

A beautiful young woman with long, voluminous, wavy brown hair and hazel eyes, looking thoughtfully to the side.
She is illuminated by soft, natural light coming from a nearby window with sheer curtains. She is wearing a simple, beige off-the-shoulder top.
The mood is serene and pensive. The style should be a photorealistic portrait with a shallow depth of field, creating a soft, blurred background.

---

Grok.imagine

Prompt:

Photorealistic close-up portrait of a young East Asian female singer (K-pop idol aesthetic) on a dark stage.
big eyes with thick lips wearing a black off-the-shoulder top with spaghetti straps. Her light brown hair is styled in wavy and curry.
She is holding a professional stage microphone and singing with a focused, emotional expression.
Dramatic, high-contrast volumetric lighting, strong spotlight isolating the subject, deep shadows, blue/black background.
Cinematic shot, 8k, hyperdetailed, shallow depth of field, aspect ratio 9:16

---

Z-Image Turbo

Model: Comfy-Org/z_image_turbo

<iframe width="514" height="289" src="https://www.youtube.com/embed/qHqRSXB4xPQ" title="ComfyUI 結合 Z-Image Turbo 全攻略｜不可思議的6B本機生圖模型，20秒出圖、支援中文！本機部署實測" frameborder="0" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture; web-share" referrerpolicy="strict-origin-when-cross-origin" allowfullscreen></iframe>

---

RED-Zimage 1.5

Model: https://civitai.com/models/958009/redcraft-or-redzimage-or-updated-dec03-or-latest-red-z-v15
Blog:RED-Zimage 1.5 Review: Finally, Real AI Photos?

<iframe width="533" height="289" src="https://www.youtube.com/embed/DOkeQdxNjaU" title="50.ComfyUI完整指南：ComfyUI 新王者！REDZimage 1.5 全面超越 Zimage Turbo，寫實效果再進化！" frameborder="0" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture; web-share" referrerpolicy="strict-origin-when-cross-origin" allowfullscreen></iframe>

---

2. Generative Video

Sora2

Veo3.1

---

Wan2.2 T2V & I2V

ComfyUI - WAN2.2

5B TI2V workflow

14B T2V workflow

14B I2V workflow

---

Wan2.2 Animate

ComfyUI - Wan2.2 Animate

14B Animate workflow

---

Wan2.2 S2V

ComfyUI - Wan2.2 S2V

14B S2V workflow

---

InfiniteTalk

ComfyUI - Wan2.1 InfiniteTalk 讓圖片、影片生成的人物完美對應口型

1. Text-to-Image : Grok.com/imagine
   
2. Lyrics-to-Song : Suno.com
   
3. Image-to-Video : ComfyUI + Wan2.1 Video + InfiniteTalk

---

I2T -> T2I -> I2V

ComfyUI_Qwen3-VL-Instruct

---

3. Generative Song

DiffRhythm

Jamify

https://www.kaggle.com/code/rkuo2000/jamify

---

ACE-step

ComfyUI ACE-step

---

Suno

Suno小標籤提示詞1.pdf

---

4. Generative Speech

Spark-TTS

https://kaggle.com/code/rkuo2000/Spark-TTS ~ spark.mp3

Index-TTS2

https://kaggle.com/code/rkuo2000/Index-TTS2 ~ gen.mp3

FireRedTTS2

audio.mp3

---

5. Talking Avatar

First Order Motion Model for Image Animation

---

LipSync-Avatar

---

TalkMateAI

---

6. Image-to-3D

Hunyuan3D

Paper: Hunyuan3D 2.5: Towards High-Fidelity 3D Assets Generation with Ultimate Details
Code: https://github.com/Tencent-Hunyuan/Hunyuan3D-2.1

ComfyUI - Hunyuan3D 2.1