Spaces:
Runtime error
Runtime error
| 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 ์ ์ |