Browse production-facing AI capability cases or jump straight into browser-based interactive demos to understand outputs, interaction style, and runtime shape.
Platform Guide
Browse production-facing capability work
Review real examples across image, video, audio, docs, and automation with modal breakdowns and standalone detail pages.
Try realtime interaction directly
Open interactive demos, grant camera or audio access when needed, and experience realtime AI feedback running in the browser.
Capability Cases
15
Interactive Labs
7
Homepage Guide
The homepage now presents both production-facing capability showcases and realtime interactive labs in one discovery flow. Start with outcomes and case structure, or jump directly into a playable demo.
Path 01
Capability Showcase
Best when you want to evaluate output quality, case structure, and delivery format first. Supports filtering, modal breakdowns, detail pages, and sharing.
Card preview / modal breakdown / detail page
Path 02
Interactive Labs
Best when you want to feel realtime interaction, instant feedback, and playful engagement first. Open and start immediately.
7 live labs / instant play
Capability cards still support previews, modal breakdowns, and standalone detail-page sharing. If you want to review delivery structure and output quality first, start with the category grid below.
High-fidelity face replacement for image content, supporting multi-person swaps and style variations.
A complete pipeline view of source face, template video, and generated outputs across multiple side-by-side cases.
Supports multiple visual styles such as Disney-inspired, retro, and realistic comic looks for portraits and photo upgrades.
Transforms portrait photos into Pixar-like animated characters for avatars, posters, and IP character creation.
Voice-Driven Image Avatar
单张图片驱动口播视频，首页优先展示完整人物形象。
Generate a speaking video avatar from a single portrait image, suitable for explainers, announcements, and virtual hosting.
Zero-Shot Voice Cloning
Reference voice and zero-shot clone output preview.
Reference
Short Voice Sample
Output
Emotion Variants
Short-reference, zero-shot voice cloning with emotion variants, suitable for character dubbing and reusable voice assets.
AI Song Voice Conversion
Compare user voice, source track, and converted output.
Target
Target Voice
Compare
Source / Result
Convert an entire song using a target voice or user reference voice, with clear before-and-after comparison for demos and voice transfer showcases.
Expressive TTS
A/B listening across commentary, dialogue, and long-form reading with TTS 3.0 and TTS 4.0.
A multi-scene TTS showcase covering sports commentary, everyday dialogue, and long-form reading with direct TTS 3.0 vs TTS 4.0 comparison.
Automatic foreground cutout optimized for e-commerce and content workflows, with support for products, portraits, and complex edges.
Remove watermarks across short-video platforms while preserving core content quality for reuse and redistribution.
Detect 106 facial landmarks and identify beautification regions for photo enhancement, portrait analysis, and preprocessing demos.
Detect 360 facial landmarks for finer portrait structure analysis, feature localization, and preprocessing demos.
Transfer expressions, lip motion, and head movement from a singing driving video onto a target portrait video for performance-style content.
Reduce eye lines, forehead lines, and nasolabial folds while improving skin tone balance for portrait retouching demos.
Video Matting
Foreground extraction, frame preservation, synced audio output
Output
MP4
Audio
Kept
Demo
24s
Automatically extract human foreground and deliver preview-ready matted video for live backgrounds and short-form editing.
Keep Exploring
The site now supports two discovery paths in one place: browse capability cases for output and delivery context, or open browser-native labs for hands-on interaction.
Image
7
Video
5
Audio
3
Capability pages explain the cases and delivery direction, while interactive labs make the experience value tangible right away.
Labs Spotlight
These demos are now surfaced as first-class in-site destinations so visitors understand immediately that this product includes both capability cases and playable realtime experiences.
Invite users to swing both hands to the beat and slice incoming targets for a more immersive, high-energy interaction.
Core capability
Combines hand recognition with rhythm-based feedback, making it a strong fit for warm-up moments and youth-focused experiences.
Turn simple jump motions into an easy-to-learn dodge challenge that gets people moving right away.
Core capability
Uses motion detection to recognize jump and dodge behavior, making it ideal for low-friction body-controlled mini games.
Preview different hair colors in real time so users can compare styling changes and visual moods more intuitively.
Core capability
Combines hair-region recognition with instant preview feedback, making it useful for beauty, hair-color, and personalized recommendation flows.
