星流是 LiblibAI(哩布哩布)旗下的一站式 AI 设计 Agent,他也被称为全球爆款设计工具 Lovart 的国内官方版。 你只需要用大白话跟它对话,它就能帮你出图、出视频、出 3D,甚至出一
star stream isOne-stop AI design agent under LiblibAI, he is also known as the global hot design toolThe official domestic version of Lovart。
you just needTalk to it in plain English, and it will help you produce pictures, videos, 3D, and even a complete set of brand visual solutions.。
Although everyone calls it an AI painting tool, after using it, I found that Xingliu’s ability goes far beyond “making a picture”. Let’s see what it can do:
The base is LiblibAI's self-developed Star-3 Alpha large model, which claims to have surpassed Midjourney and DALL·E in multiple professional tests. After my actual test, the most amazing thing is actually its ability to understand Chinese prompt words. Whether it is the cross-dimensional grafting of "Cyber Dunhuang Style" or the very metaphysical regional aesthetic of "New Chinese Dream Core Fishing Port", it can actually capture it without the mechanical feeling of "translating English into Chinese and then applying it mechanically".
Moreover, Xingliu is connected to more than ten top models such as F.1, Kling, Qwen, Hailuo02, etc. If you really want to differentiate your style, you can also choose the model yourself.
I have to mention this feature separately - it's true! of! good! use! The working interface of Xingliu is a "canvas" that can be expanded infinitely in any direction. All generated pictures and video content can be freely arranged and combined, making it feel like dragging it in Figma.
The editing tools are also very hard-core: high-definition enlargement (up to 8k support), intelligent image expansion, background removal, partial redrawing, and element layering, which basically cover 90% of daily design retouching needs.
In my opinion, this function is the "king bomb" of Xingliu. You can set a character or a set of visual specifications, and AI can help you output dozens or hundreds of pictures or a complete set of design materials in batches while maintaining a highly consistent style. For example, in the past, making a set of emoticon packs required drawings one by one. Now you only need to make a request, and it can bring you "9 emoticon packs + peripheral design + promotional video" in one go.
That’s right, this “drawing tool” can also create videos and 3D models. Enter a product description, and it can handle everything from script to storyboard to finished film. 3D character modeling is also supported. I have tried to generate plush-textured dolls. Although the output effect is not at the level of professional modeling software, it is enough for quickly producing concept drafts.
I have to talk about this carefully, because I found that the various opinions on the Internet are a bit confusing, so I will help everyone sort it out.
free mode: Register to get initial points, and log in every day to get free points. Get 30 free points every day.
Paid model: Different tiers of 99 yuan/month correspond to different points amounts and access to advanced functions.