SD3 has been released. For the tutorial on local software installation, deployment, and use, please go to: 【STablediffusion3 Local Installation and Deployment Tutorial】
https://v.douyin.com/i2PtvVgd/
Stability has released SD3 and claims that its image generation capability has surpassed DALL·E 3 and Midjourney V6, but it has not disclosed specific parameters and details, which means the model may not be free and open source. The model's biggest feature is its text rendering capability; generating text in images has long been impossible, and now it is possible.
How to Use Text Rendering
Simply describe the text you want to render in double quotation marks “XXX”; below is an example provided by the forum moderator.
Prompt: A chinese girl with short white t-shirt , a red words "NIUBOYIAI" write in the t-shirt
Currently, SD3 can be used directly on the official website or called externally via API. Using it directly on the official website requires queuing, so it will be a bit slower; using your own API will be faster. The API is not for unlimited use; new users get a certain free quota when applying, and can top up after it is used up. Below are the addresses for the two methods.
1. Use Directly on the Official Website
https://sdxlturbo.ai/zh-CN/stable-diffusion3
2. Use Externally via API
Step 1: Go to the official website https://platform.stability.ai/account/keys to register an account, then click the create api key button in the upper right to obtain an API
At present, the experts in China have not yet made an installation package. It is believed that friends who have installation packages from Qiuye and other experts will later be able to obtain the SD3 version through upgrades. |