软餐获悉，Meta 拥有的 AI 图像生成工具 Imagine 新增了 “Reimagine” 功能，现在它可以为生成的图像创建补充内容。当用户在 Meta 的消息服务群聊中使用 Imagine 工具时，用户将能够基于已生成的照片创建新的图片，只需选择相关图片并输入附加文本提示，就会生成一张新图片。Imagine 工具基于 Meta 的 Emu 模型开发而成，并且已在 WhatsApp、Messenger 和 Instagram 等应用的英语版本中提供。现在，该工具还可通过自己的网站作为独立应用使用，该网站初期仅对美国用户开放。Meta 表示，未来几周内将向使用此模型生成的图像添加一个不可见水印，这种水印能够抵御 “常见图像篡改技术（如裁剪、调整颜色和截屏）。
Meta’s AI image generation tool Imagine has added a new feature called “Reimagine”, which allows users to create supplementary content for the generated images. When users use the Imagine tool in Meta’s messaging service group chat, they will be able to create new images based on the already generated photos by selecting relevant images and entering additional text prompts. The Imagine tool is developed based on Meta’s Emu model and is already available in English versions of applications such as WhatsApp, Messenger, and Instagram. Now, this tool can also be used as a standalone application through its own website, initially only open to users in the United States. Meta stated that within the next few weeks, an invisible watermark will be added to images generated using this model to resist common image manipulation techniques such as cropping, color adjustment, and screenshots.
微信扫一扫