XinYu.

XINYU AI

User Guide

Learn how to use the XinYu AI canvas to create high-quality AI images and videos.

Overview

XinYu AI's core philosophy is "canvas as workflow". Each creative project is an independent canvas where you can freely place different node types (prompt, image, video, audio), define reference relationships through connections, and let AI generate content based on your arrangement.

Node-Driven

Different node types correspond to different creative elements, freely combined to build workflows.

Connection References

Use connections to tell AI which assets to reference, precisely controlling generation results.

Multi-Modal Generation

Supports text-to-image, text-to-video, image-to-video, video enhancement, audio generation, and more.

New users receive free credit allowance upon registration, so you can start creating immediately.

Quick Start

From zero to your first AI creation in 5 steps:

1

Sign Up & Log In

Visit xinyuai.app and register with your email. Log in after email verification.
2

Create a Project

After logging in, enter the workspace, click "New Project", name it, and open a blank canvas.
3

Add a Prompt Node

Click the ⊕ button on the left dock to open the add menu, select "Prompt" to create a prompt node. Type your desired content description.
4

Choose a Model & Generate

In the prompt node panel, choose generation type (image/video) and specific model, set parameters (size, ratio, duration, etc.), and click generate.
5

View & Export Results

After generation, results appear in a new node. Click to preview full-size images/play videos, and use the toolbar download button to save locally.
Double-click on empty canvas space to quickly open the add-node menu — faster than using the left dock.

Node Types

Every element on the canvas is a "node". Different node types carry different content and functionality:

Prompt Node

Core driver node. Enter text descriptions, choose AI model, generate images or videos. Connect reference images/videos/audio as input conditions.

Image Node

Holds AI-generated or manually uploaded images. Connect as reference to prompt nodes, influencing style and content of subsequent generations.

Video Node

Holds AI-generated or uploaded videos. Supports video preview, quality enhancement, duration extension, and other post-processing.

Audio Node

Holds AI-generated or uploaded audio files. Connect to prompt nodes as audio-driven input.

Prompt nodes are the only node type that can trigger AI generation. Image, video, and audio nodes don't generate on their own but can be connected as reference inputs to prompt nodes.

Connections & References

Connections are the core interaction mechanism in XinYu AI. They define reference relationships between nodes — when you connect a media node to a prompt node, AI automatically references that asset's content, style, or structure during generation.

How to Create Connections

  • 1.Hover over a node, the ⊕ handle appears on the right
  • 2.Press and drag from the ⊕ handle to pull out a connection
  • 3.Drop on the target node's left handle to establish the connection

Reference Type Guide

Image → Prompt: Image serves as first-frame or style reference, influencing visual style and composition
Video → Prompt: Video as dynamic reference for style transfer, motion continuation, or content extension
Audio → Prompt: Audio-driven video generation, video rhythm and visuals match the audio content

Deleting Connections

Hover over a connection line, a scissors icon appears — click it to delete the connection.

A prompt node can receive multiple reference inputs simultaneously. For example: connect an image for style reference + an audio for rhythm reference, and the generated video will match both.

AI Generation

XinYu AI integrates multiple cutting-edge AI models, supporting various generation modes. All generation is triggered through prompt nodes.

Text-to-Image

Enter a description in the prompt node, select an image model (GPT Image, Seedream, etc.), set dimensions and ratio, and generate.

Supported models: GPT Image 2, Seedream 4.0, etc.

Text-to-Video

Enter a video scene description, choose a video model (Seedance 2.0), set duration and aspect ratio, and generate. Generation typically takes 1-5 minutes.

Supported models: Seedance 2.0, Seedance 2.0 Fast, etc.

Image-to-Video

Connect an image node to a prompt node — the image becomes the first-frame reference. Add motion descriptions and select a video model to generate.

Video Post-Processing

After selecting a video node, the toolbar offers "Enhance" for quality and "Extend" for continuing content.

Do not close or refresh the page during generation. If accidentally closed, in-progress tasks will automatically resume when you reopen the project.

Canvas Controls

The canvas supports two interaction modes, toggled via the first button in the bottom-left toolbar:

Select Mode (Default)
  • • Left-drag empty space → Selection box
  • • Middle / Right-drag → Pan canvas
  • • Space + drag → Pan canvas
Pan Mode
  • • Left-drag → Pan canvas
  • • Shift / Cmd + drag → Selection box

Other Canvas Operations

•Scroll / Trackpad — Zoom in/out
•Double-click empty area — Open add menu
•Right-click canvas/node — Context menu
•Drag files to canvas — Direct upload
•Bottom-left zoom slider — Precise zoom
•Fit view button — Show all nodes

Keyboard Shortcuts

Delete selected
⌫
Select all
⌘A
Copy / Paste
⌘C/⌘V
Undo / Redo
⌘Z/⌘⇧Z
Fit view
⌘1
Add to selection
⇧+ drag
Pan canvas
Space+ drag

Asset Management

Material Library

Click the material library icon on the left dock to open the global library. Save frequently-used images, videos here for reuse across projects. Right-click a node → "Save to Library" for quick bookmarking.

Project Assets

Click the folder icon to view all generated images, videos, and audio in the current project. Convenient for quick browsing and locating historical results.

Upload & Download

Upload: Drag files onto canvas, or use the ⊕ menu. Supports JPG/PNG/WebP images, MP4/WebM videos, MP3/WAV audio. Download: Select node → toolbar download button.

Billing

XinYu AI uses "Xins" (心点) as its unified credit unit. Different models and operations have different costs.

Image generation3-10 Xins/image
Video generation (Seedance 2.0)~20-23 Xins/sec
Video enhancementBilled by resolution
Generation failureAuto-refunded

How to Top Up

Click the avatar button at the bottom of the left dock → Select "Top up" → Choose a credit pack → Pay securely via Stripe. Minimum top-up is 500 Xins ($5).

Follow our Discord community and social media for occasional free credit events and creative contests.

FAQ

Still have questions?

Contact Us