文字生成图片
文字生成图片，为你的想象而做
描述一个主体、一处场景，以及你想要的感觉。在一个简单的工作台里生成新图片，下载它，或者直接带进下一步编辑。
三步把想法变成图
- 1
描述画面
先写主体，再补上风格、光线和构图。
- 2
选择规格
选择 PixJelly V2.0 或外部图像模型，然后设定画面比例。
- 3
生成并下载
跟着任务进度看，确认结果后保存图片。
把方向留给你自己
想探索就写一句短提示，想要更明确就把细节补足。这里没有必须遵守的公式。
美感藏在细节里
把心里想的材质、光线和构图写出来。不断试不同的版本，直到这张图真的贴合你的项目。
一个专注的图像工作台
上手需要的控制项，都在这里。
灵活的画面比例
方形、竖版、横版任选。
官方与外部模型
既可以用我们调好的工作流，也可以直接选 Nano Banana、GPT Image 2 或 Seedream。
结果随时下载
在结果面板里直接保存生成的图片。
The rest of the image toolkit
Making a picture is one job. These are the others.
Image to Image
Start from a photo you already have and describe a new style, setting or mood for it.
Annotate & Edit
Keep an image and change one part of it. Draw an arrow at the spot and write the instruction there.
Remove Background
Separate a subject as a transparent PNG. That tool runs on your own device rather than on a server.
接着往下做
开始一件视觉作品
美术方向
- 试探一种氛围
- 试一版构图
社交配图
- 做一张竖版图
- 搭一个视觉概念
静物
- 写清物体和材质
- 换着光线试试
What is a text to image generator?
An AI image generator earns its place when the picture you need does not exist yet: an illustration for an article, a concept to show someone, a placeholder that fits your layout. Here is what happens to what you type.
Your prompt
The workspace accepts up to 5,000 characters. A subject on its own is enough to begin, and adding material, lighting, framing and mood narrows what comes back. There is no keyword syntax to learn and no magic words to collect; ordinary sentences work.
Your model
PixJelly V2.0 is our curated workflow, currently running on Nano Banana Pro. Nano Banana 2, GPT Image 2 and Seedream 5 Pro are available too, and each one reads the same prompt in its own way. Running one brief through two of them is a cheap way to learn which suits your kind of picture.
Your frame
Pick 1:1, 3:4, 4:3, 16:9 or 9:16 before generating. The ratio shapes the composition, so it is worth matching to wherever the image will end up. Switching models resets the choice back to square, which makes the format the last thing to set.
Your result
Generation happens on the provider servers, not in your browser. The finished image appears in the result panel with a download button, and starting another generation replaces it, so save anything you want to keep before the next run.
先生成，再打磨
先用文字生成一张起始图。下载之后打开标注编辑，用箭头指挥具体的修改。
关于图片生成的常见问题
PixJelly V2.0 是什么？
PixJelly V2.0 是我们调校过的图像工作流，目前底层由 Nano Banana Pro 驱动。它把一套稳定的视觉需求描述和底层模型组合在一起，本身不是单独训练的基础模型。
需要准备参考图吗？
不需要。这个工具直接根据你的文字提示生成图片。
生成需要多久？
取决于所选模型和服务方的负载。结果面板会显示进度。如果等待超时，回去查看原来那个任务就好，不用再发起一次生成。
图片生成是在本地完成的吗？
不是。你的提示词会发送给图像服务方。在本地设备上运行的是移除背景。