Key Takeaways
1. Microsoft has launched Bing Video Creator in its mobile app, allowing users to create videos from text prompts using OpenAI’s Sora AI for free.
2. The service offers two speed settings for video generation: standard (unlimited) and fast (limited to 10 credits, with more available through Microsoft Rewards).
3. Videos have a 9:16 aspect ratio, a maximum duration of 5 seconds, and are watermarked to indicate AI generation, with a 90-day download period before deletion.
4. While AI technology for video creation is advancing, some outputs may appear strange or nonsensical, highlighting the need for further refinement.
5. Users can expect better results with image generation, as that technology has been more thoroughly developed compared to video generation.
Microsoft has introduced Bing Video Creator in its Bing mobile app, enabling users to create videos from text prompts at no cost, leveraging a modified version of OpenAI’s Sora AI. This video creation feature will soon be available on the Bing website and other platforms.
Speed Options and Credits
The service comes with two generation speed settings: standard and fast. Users are granted 10 fast credits for video creation. While an unlimited number of videos can be made on the standard speed setting, using the fast option beyond the initial credits relies on users earning more Microsoft Rewards points. For every hundred points earned, users can create one additional video.
Video Specifications
All videos produced through Bing Video Creator are restricted to a 9:16 aspect ratio and have a maximum duration of five seconds. Users can download their videos for a period of 90 days before they are automatically deleted, and each video is watermarked in accordance with C2PA standards to signify that they are AI-generated. This also means that videos, including those promoting LGBT content, can be traced back to their original creators.
AI Advancements
The technology behind AI text-to-video generation has progressed quickly, allowing current AI to transform written scenarios into realistic videos. However, as these models are still being refined, some outputs may display oddities. The Sora AI utilized in BVC has implemented restrictions to avoid the creation of harmful or unsafe content, but this seems to hinder its capacity to depict common actions accurately.
In trials to produce videos of a dog drinking from a bowl, a cat grooming itself, and an anime girl savoring ice cream, the results were often nonsensical or failed to capture the intended activity. In contrast, a rival service, SuperMaker.AI, successfully generated videos that accurately portrayed the intended actions without bizarre visuals.
Image Generation Expectations
Users can anticipate more satisfactory results when generating images, as that AI technology has had more time for refinement. A book available on Amazon showcases numerous instances of the realistic art that AI can produce today.
Source:
Link
