Spaces:
Running
on
Zero
Running
on
Zero
File size: 4,192 Bytes
87dbdf7 197ad12 87dbdf7 2010fa1 87dbdf7 197ad12 87dbdf7 2010fa1 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 |
---
title: DreamO Video
emoji: ๐จ
colorFrom: purple
colorTo: yellow
sdk: gradio
sdk_version: 5.35.0
app_file: app.py
pinned: false
license: apache-2.0
short_description: A Unified Framework for Custom Image and Video generation
---
## English Description
**DreamO Video - AI-Powered Image and Video Generation**
DreamO Video is an advanced AI application that combines cutting-edge image generation with video synthesis capabilities. Built on the FLUX.1-dev model and enhanced with the DreamO pipeline, this tool allows users to create highly customized images and transform them into dynamic 2-second videos.
### Key Features:
1. **Dual Reference Image System**: Upload up to two reference images to guide the generation process, with flexible task options:
- **IP (Image Prompt)**: Use images as visual prompts to influence the overall composition
- **ID (Identity)**: Preserve facial features and identity from reference images
- **Style**: Transfer artistic style from reference images to generated content
2. **Advanced Image Generation**:
- Customizable resolution (768-1024 pixels)
- Fine-tuned guidance controls for precise output
- Seed-based generation for reproducible results
- Background removal and face alignment preprocessing
3. **Video Synthesis**: Transform any generated image into a 2-second animated video with natural motion and smooth transitions (Note: Full version supports up to 60-second videos)
4. **Professional Features**:
- Automatic watermarking for video outputs
- Advanced CFG (Classifier-Free Guidance) controls
- Negative prompting for refined results
- Gallery of preprocessing results for transparency
### Technical Specifications:
- Powered by FLUX.1-dev and DreamO pipeline
- GPU-accelerated processing using CUDA
- Integrated face restoration and background removal tools
- FFmpeg-based video post-processing
### Use Cases:
- Creative content generation for social media
- Character design and animation
- Product visualization and marketing materials
- Artistic style transfer and experimentation
- Personal avatar and portrait creation
---
## ํ๊ธ ์ค๋ช
**DreamO Video - AI ๊ธฐ๋ฐ ์ด๋ฏธ์ง ๋ฐ ๋น๋์ค ์์ฑ ๋๊ตฌ**
DreamO Video๋ ์ต์ฒจ๋จ ์ด๋ฏธ์ง ์์ฑ๊ณผ ๋น๋์ค ํฉ์ฑ ๊ธฐ๋ฅ์ ๊ฒฐํฉํ ๊ณ ๊ธ AI ์ ํ๋ฆฌ์ผ์ด์
์
๋๋ค. FLUX.1-dev ๋ชจ๋ธ์ ๊ธฐ๋ฐ์ผ๋ก ํ๊ณ DreamO ํ์ดํ๋ผ์ธ์ผ๋ก ๊ฐํ๋ ์ด ๋๊ตฌ๋ฅผ ํตํด ์ฌ์ฉ์๋ ๊ณ ๋๋ก ๋ง์ถคํ๋ ์ด๋ฏธ์ง๋ฅผ ์์ฑํ๊ณ ์ด๋ฅผ ์ญ๋์ ์ธ 2์ด ๋์์์ผ๋ก ๋ณํํ ์ ์์ต๋๋ค.
### ์ฃผ์ ๊ธฐ๋ฅ:
1. **์ด์ค ์ฐธ์กฐ ์ด๋ฏธ์ง ์์คํ
**: ์ต๋ 2๊ฐ์ ์ฐธ์กฐ ์ด๋ฏธ์ง๋ฅผ ์
๋ก๋ํ์ฌ ์์ฑ ๊ณผ์ ์ ์๋ดํ๋ฉฐ, ์ ์ฐํ ์์
์ต์
์ ๊ณต:
- **IP (์ด๋ฏธ์ง ํ๋กฌํํธ)**: ์ด๋ฏธ์ง๋ฅผ ์๊ฐ์ ํ๋กฌํํธ๋ก ์ฌ์ฉํ์ฌ ์ ์ฒด ๊ตฌ์ฑ์ ์ํฅ
- **ID (์ ์)**: ์ฐธ์กฐ ์ด๋ฏธ์ง์ ์ผ๊ตด ํน์ง๊ณผ ์ ์ ๋ณด์กด
- **Style (์คํ์ผ)**: ์ฐธ์กฐ ์ด๋ฏธ์ง์ ์์ ์ ์คํ์ผ์ ์์ฑ ์ฝํ
์ธ ๋ก ์ ์ก
2. **๊ณ ๊ธ ์ด๋ฏธ์ง ์์ฑ**:
- ์ฌ์ฉ์ ์ ์ ๊ฐ๋ฅํ ํด์๋ (768-1024 ํฝ์
)
- ์ ๋ฐํ ์ถ๋ ฅ์ ์ํ ๋ฏธ์ธ ์กฐ์ ๋ ๊ฐ์ด๋์ค ์ปจํธ๋กค
- ์ฌํ ๊ฐ๋ฅํ ๊ฒฐ๊ณผ๋ฅผ ์ํ ์๋ ๊ธฐ๋ฐ ์์ฑ
- ๋ฐฐ๊ฒฝ ์ ๊ฑฐ ๋ฐ ์ผ๊ตด ์ ๋ ฌ ์ ์ฒ๋ฆฌ
3. **๋น๋์ค ํฉ์ฑ**: ์์ฑ๋ ๋ชจ๋ ์ด๋ฏธ์ง๋ฅผ ์์ฐ์ค๋ฌ์ด ์์ง์๊ณผ ๋ถ๋๋ฌ์ด ์ ํ์ด ์๋ 2์ด ์ ๋๋ฉ์ด์
๋น๋์ค๋ก ๋ณํ (์ฐธ๊ณ : ์ ์ ๋ฒ์ ์ ์ต๋ 60์ด ๋น๋์ค ์ง์)
4. **์ ๋ฌธ ๊ธฐ๋ฅ**:
- ๋น๋์ค ์ถ๋ ฅ๋ฌผ์ ๋ํ ์๋ ์ํฐ๋งํน
- ๊ณ ๊ธ CFG (Classifier-Free Guidance) ์ปจํธ๋กค
- ์ ์ ๋ ๊ฒฐ๊ณผ๋ฅผ ์ํ ๋ค๊ฑฐํฐ๋ธ ํ๋กฌํํ
- ํฌ๋ช
์ฑ์ ์ํ ์ ์ฒ๋ฆฌ ๊ฒฐ๊ณผ ๊ฐค๋ฌ๋ฆฌ
### ๊ธฐ์ ์ฌ์:
- FLUX.1-dev ๋ฐ DreamO ํ์ดํ๋ผ์ธ ๊ตฌ๋
- CUDA๋ฅผ ์ฌ์ฉํ GPU ๊ฐ์ ์ฒ๋ฆฌ
- ํตํฉ๋ ์ผ๊ตด ๋ณต์ ๋ฐ ๋ฐฐ๊ฒฝ ์ ๊ฑฐ ๋๊ตฌ
- FFmpeg ๊ธฐ๋ฐ ๋น๋์ค ํ์ฒ๋ฆฌ
### ์ฌ์ฉ ์ฌ๋ก:
- ์์
๋ฏธ๋์ด๋ฅผ ์ํ ์ฐฝ์์ ์ธ ์ฝํ
์ธ ์์ฑ
- ์บ๋ฆญํฐ ๋์์ธ ๋ฐ ์ ๋๋ฉ์ด์
- ์ ํ ์๊ฐํ ๋ฐ ๋ง์ผํ
์๋ฃ
- ์์ ์ ์คํ์ผ ์ ์ก ๋ฐ ์คํ
- ๊ฐ์ธ ์๋ฐํ ๋ฐ ์ด์ํ ์ ์ |