Xingliu AI: a design tool that turns one sentence into a layout

Xingliu AI is a one-stop AI design and creation tool from LiblibAI, covering image, audio and video generation plus practical retouching such as high-definition upscaling, outpainting and object removal. Its centrepiece is the Xingliu Agent: describe what you need in a sentence and the design agent produces an editable piece of work.

In design work the expensive part is rarely the idea; it is turning the idea into a deliverable file, with composition adjusted, sizes resized, assets swapped and formats exported. Xingliu AI folds that chain into an infinite canvas where images can be retouched, replaced and recomposed directly, and a few turns of conversation carry the revision forward without hopping between applications.

The model layer draws on Xingliu Star-3, Nano, Qwen-Image and others, while output covers jpg, png, svg, mp4 and glb, meaning image, video, audio and 3D sit in one flow. For projects that need both a static layout and motion assets, that removes a good deal of format conversion and asset housekeeping.

Main capabilities

Several design disciplines

Image, audio and video requests all start in the same platform, from a static poster to a motion asset, so there is no separate tool for each content type.

Automated design flow

The Xingliu Agent reads the instruction and generates a design on its own, compressing describe, draft and revise into a conversational loop that suits the early stage when options are being spread out.

Infinite canvas editing

Multi-turn understanding lets retouching, asset swapping and recomposition happen on the canvas itself. Without a fixed page size, layout and extension behave more like a real design process.

Chinese-language tuning

The agent was built for Chinese-speaking users, so it reads Chinese prompts and Chinese typesetting needs closer to local habits, a difference that shows up when headlines and slogans are involved.

Model selection

Xingliu Star-3, Nano, Qwen-Image and others are available and can be switched by subject and style, so a project is not capped by one model's ceiling.

Multiple output formats

jpg, png, svg, mp4 and glb cover illustration, cartoon, visual design and more; the vector and 3D options let a deliverable move straight into later production.

Image post-processing

High-definition upscaling, outpainting and erasing handle the usual problems: too little resolution, a composition that needs extending, unwanted objects in frame.

Inspiration gallery

Browsing by illustration, cartoon or visual design gives a reference point for subject and tone before generating anything.

How to use it

1. Register and sign in on the official site; free members receive 150 Xingliu points a day, enough to try basic generation.
2. Open the AI creation entry and choose a model. Models differ in realism, illustration and text rendering, so a small test image saves points.
3. Describe the scene in the text box; if there is a reference image, upload it through image-to-image, then set the size, images per batch and total count.
4. Press generate, review the results on screen and download the ones worth keeping.
5. For deeper changes or a complete design, talk to the Xingliu Agent in the right-hand panel and let it adjust composition, swap elements or extend the canvas directly.
6. Export according to the deliverable: jpg or png for flat layouts, svg when later editing is needed, mp4 for motion and glb for 3D assets.

Pricing and membership tiers

• Free membership: 150 Xingliu points per day after signing in.
• Standard: 469 yuan a year, 4,000 points a month, roughly 4,000 generator images or 23 agent tasks (46 in the first month), 100 extra points daily, accelerated generation for 2,000 images a month, member models, commercial-safe use and multi-format export.
• Advanced: 949 yuan a year, 12,000 points a month, roughly 12,000 images or 69 agent tasks (138 in the first month), 100 extra points daily, accelerated generation for 10,000 images a month.
• Super: 1,899 yuan a year, 24,500 points a month, roughly 24,500 images or 141 agent tasks (282 in the first month), 100 extra points daily, accelerated generation for 21,000 images a month.

Who it suits

• Brand and product design teams that need to spread out logo directions, posters and packaging quickly before converging on one.
• Social media and content teams producing covers, story images and short video assets in bulk at platform-specific sizes.
• Advertising and marketing staff drafting posters, brochures and animated ads to shorten the proposal cycle.
• UI and UX designers generating interface concepts, prototype hints and motion references.
• Illustrators and artists using the gallery for direction and multi-model comparison to explore style boundaries.

Limits to know first

• How points translate into agent tasks: the source only gives approximations (4,000 points for about 23 agent tasks), so real consumption moves with complexity.
• The exact scope of the commercial-safe label, the permitted scenarios and any attribution requirement: not disclosed, so confirm item by item before commercial use.
• Ownership of generated work and the boundary of the platform's usage rights: not disclosed.
• How many images the 150 free daily points actually produce, and when the allowance refreshes: not disclosed.
• Specific model versions, parameter ceilings and concurrency limits: not disclosed, and queues are possible under load.
• Agent output is still a first draft; brand guidelines, font licensing and print colour all need human review.
• Prices and point policies can change, so check the live page inside the platform before paying.

FAQ

How is this different from an ordinary AI drawing tool?

An ordinary tool focuses on producing one image. Xingliu AI works through the design process: a first draft from one sentence, multi-turn revision on an infinite canvas, then export to svg, mp4 or glb as needed, which suits people who must hand over files rather than look at pictures.

Can I use it without prompt-writing skills?

Yes. The Xingliu Agent is designed so a single sentence produces a draft, and later changes happen in conversation. If describing is hard, start from the inspiration gallery and upload a reference through image-to-image.

How are points consumed?

Points pay for generation and agent tasks. The source estimates 4,000 points as roughly 4,000 generator images or 23 agent tasks, which shows an agent task costs far more than a single image, so budget before starting a full design.

Can the output be used commercially?

The tiers mention commercial-safe terms and member models, but the exact licence scope, permitted scenarios and attribution requirements are not disclosed, so verify per project, particularly when a brand mark or a typeface is involved.