Google正为其Google Photos和YouTube Shorts两大平台注入新的AI驱动功能，使用户能够轻松将静态照片转化为动态视频。这些新工具在技术上与近期在Google Gemini应用中推出的基于Veo 3的照片转视频功能相似，但它们由Google的早期Veo 2视频模型驱动，因此在功能和限制上有所不同。Google Photos中的照片转视频功能主要限于生成六秒的短片，而YouTube Shorts版本则允许用户自定义生成视频的时长。然而，与Gemini或直接使用Veo 2不同的是，这两个新工具均不支持用户自定义提示词来引导生成结果，用户只能从预设的选项中进行选择，例如“Subtle movements”或“I’m feeling lucky”。这些功能正逐步在美国地区向Google Photos用户推出，而YouTube Shorts用户则将在未来一周内在美国、加拿大、澳大利亚和新西兰地区体验到这些新能力。

✨ Google Photos和YouTube Shorts正在整合新的AI驱动照片转视频功能，用户可以将静态照片转化为动态内容。这些功能基于Google的Veo 2视频模型，虽然与Gemini的Veo 3功能相似，但存在一些限制，例如无法自定义提示词，用户只能从预设选项中选择，并且Google Photos版本限制为六秒短片。

🎬 Google Photos还将推出一项“Remix”工具，允许用户将照片转换为不同的艺术风格，如动漫、漫画、素描和3D动画。照片转视频和Remix工具将被整合到一个新的“Create”标签页下，与拼贴、精彩集锦等其他创意功能并列，方便用户查找和使用。

🚀 YouTube Shorts的相机功能也新增了生成式AI效果，可以根据涂鸦创建图像，或为自拍视频添加特效，如复制用户或模拟水下游泳效果。这些AI工具将集中在新的“AI Playground”中心，用户可以通过点击创建视频时的右上角星形图标访问，目前已在美国、加拿大、澳大利亚和新西兰向Shorts用户开放。

⚠️ Google提醒用户，这些Remix和照片转视频功能仍处于“实验”阶段，可能产生不准确的结果。用户可以通过点赞或点踩来提供反馈，帮助Google改进这些工具。所有AI生成的视频和照片都将包含Google不可见的SynthID数字水印，而Photos生成的视频还将添加可见水印，以便用户一眼识别。

⚖️ 与Apple的Image Playground工具不同，Google的AI功能旨在让用户发挥创意，而Apple则倾向于限制生成照片级真实感的图像，以规避深度伪造和错误信息等潜在风险。这反映了两者在AI内容生成领域的不同发展策略。

Google is adding new AI-powered features to Google Photos and YouTube Shorts that allow users to transform their photos into videos. The tools are similar to the Veo 3-powered photo-to-video capabilities that were added to Google’s Gemini app earlier this month, only they’re powered by Google’s older Veo 2 video model instead and have more limitations.

The photo-to-video AI generation in Google Photos is restricted to making six-second clips, while the YouTube Shorts version allows users to select what clip length to generate. Unlike in Gemini or using Veo 2 itself, however, neither tool allows users to enter their own prompt descriptions to guide the results. Instead, users can only select from a list of provided prompts, such as “Subtle movements” or “I’m feeling lucky” in Google Photos.

The feature is starting to roll out today in the US for Google Photos on Android and iOS devices, and over the next week for YouTube Shorts users in the US, Canada, Australia, and New Zealand.

A Remix tool for Google Photos will also be available to Android and iOS users in the US in the next few weeks. This tool transforms photos into a different style, including anime, comics, sketches, and 3D animation. The photo-to-video and Remix tools will be housed under a new Create tab in the Photos app that’s launching in the US next month. The Create tab places the tools alongside other creative features like collages, highlight video, and more, so that they’re all in one place and easier to find.

Google warns users that the Remix and photo-to-video features are “experimental” and may produce inaccurate results. You can thumbs up or thumbs down on generated images and videos to provide feedback that Google can use to improve the tools going forward. All videos and photos generated will include Google’s invisible SynthID digital watermark. Videos generated in Photos will also carry a visible watermark that makes them easier to identify as AI-generated at a glance.

The YouTube Shorts camera is also getting new generative AI effects that can create images based on doodles and apply video effects to selfies that duplicate the user or make them appear like they’re swimming underwater. YouTube is making these effects and other generative AI tools easier for Shorts users to find with a new AI Playground hub, which users can access by tapping the sparkle icon in the top right corner when creating a video. AI Playground is available now for Shorts users in the US, Canada, Australia, and New Zealand “to start,” according to Google.

This update adds to the generative AI capabilities in Google Photos that were already miles ahead of anything that Apple provides in the iOS Photos app. The two companies have different approaches, however: while Google lets your imagination run wild (sometimes to concerning limits), Apple’s Image Playground tool won’t generate photorealistic images, in order to mitigate concerns over things like deepfakes and misinformation.