Product introduction
Xingliu is a multimodal AI design and creation platform launched by LiblibAI, positioned as a "next-generation design Agent." Powered by its proprietary Star-3 Alpha image foundation model, it integrates text-to-image, image-to-image, video generation, 3D model creation, and audio synthesis into a single platform. It also features high-definition upscaling, intelligent outpainting, inpainting, and watermark removal tools for image refinement. Deeply adapted for the Chinese user environment, Xingliu allows users to input prompts or design requirements in natural Chinese. The built-in Xingliu Agent understands these instructions, automatically breaks down design tasks, and works with an infinite canvas to enable multi-turn conversational editing and layout, creating a complete workflow from inspiration to final product.
The platform hosts over 100,000 LoRA style models, covering realistic photography, anime, Chinese traditional illustration, e-commerce posters, and 3D rendering. Whether you are a beginner looking to quickly produce high-quality images or a professional designer needing a controllable commercial-grade environment, Xingliu offers suitable creative methods.
Key Features
- High-Quality AI Image Generation (Text-to-Image / Image-to-Image): Leveraging proprietary models like Star-3 Alpha, users can generate near-professional 4K HD images using Chinese or English prompts. It supports uploading reference images for image-to-image generation, line art coloring, or pose control, aiming for a "zero-AI-feel" realistic texture. A prompt optimization assistant automatically supplements descriptions.
- Xingliu Agent Intelligent Design Assistant: Describe your complete design needs in a single sentence of natural Chinese (e.g., "Help me create a Guochao-style opening poster for a milk tea shop, with bold black font for the main title"). The Agent automatically breaks down the task, handles generation, layout, and color matching. The right-side chat bar supports multi-turn modifications, allowing changes to text, backgrounds, and composition through conversation without switching to external software.
- Professional Image Editing Toolchain: Provides a full suite of retouching functions including localized inpainting (precisely modifying parts of an image while keeping the rest unchanged), intelligent outpainting (seamlessly extending image edges for aspect ratio conversion), HD upscaling (up to 2K/4K print resolution), watermark removal, color grading, and detail restoration. Edge blending precision leads the industry.
- Multi-modal Video, 3D, and Audio Generation: Integrates advanced video models like Kling and Hailuo. Inputting a sentence triggers automatic storyboard planning, scene generation, lip-syncing, voiceover, and background music, outputting MP4 files optimized for short-video platforms. It also supports text-to-3D and image-to-3D with GLB export, as well as generating background music and sound effects based on descriptions.
- 100k+ Style Model Library & Character Consistency: The platform integrates one of the world's largest LoRA enhancement model libraries (100k+ styles, updated daily with 200+ new ones), covering e-commerce, anime, Chinese traditional, and 3D categories. Multiple LoRAs can be stacked and mixed. Its unique "Character Reference" feature locks subject characteristics from uploaded images, maintaining high consistency in character appearance across different scenes.
- Infinite Canvas & Multi-format Export: All generated content can be dragged, laid out, and operated on multiple layers within the same infinite canvas, supporting mixed text and graphics. Exports include JPG, PNG, SVG (vector), MP4, and GLB formats, facilitating direct use in printing or digital media. Some membership plans include commercial usage rights.
Application Scenarios
- E-commerce Product Visuals & Marketing Materials: Online sellers can quickly batch-generate product main images, scene搭配 images, detail page atmosphere visuals, and promotional banners. The character reference function maintains consistency of models or products across different backgrounds, significantly reducing shooting and retouching costs.
- Brand Design & Advertising Poster Full-case Output: Brands or design teams can use the Xingliu Agent to generate initial Logo drafts, VI application examples, key visual (KV) designs for events, and offline materials with a single sentence. Vector SVG export facilitates subsequent printing extensions, shortening the cycle from brief to proposal.
- Social Media Content & Short Video Creation: Operators for WeChat Official Accounts, Xiaohongshu, or Douyin can generate cover images, infographics, and comic illustrations sized for each platform. They can also convert copy into short, fast-paced promotional videos with配乐 and voiceovers using video generation features, boosting content production efficiency.
- Illustration Art & Concept Design Exploration: Illustrators and concept artists can leverage the massive LoRA style library to explore visual directions such as Cyberpunk, Ink-wash Chinese style, or Dreamy Realism, quickly producing character designs, scene concepts, and picturebook storyboards for further development.
- UI/UX Design & Product Prototype Visual Assistance: Product managers and UI designers can generate APP splash screen illustrations, empty state placeholder images, icon assets, and even simple 3D display models, assisting in prototype demo building and internal review communication.
- Architectural Interior Effects & Spatial Design Pre-visualization: Designers can input space descriptions or upload sketches to generate indoor and outdoor rendering effect images for presenting design intent to clients. Intelligent outpainting can adjust aspect ratios to fit presentation layouts.
