News
xAI upgrade Imagine Video1
2 min read
Source: Telegram AI频道
xAI upgrade Imagine Video1.5: Added image voice reference and native 1080p video generation. Currently, the text-to-video and 1080p video generation functions are available on the Grok Imagine web version, iOS version, and Android version. The image and voice reference functions have been first opened to SuperGrok Heavy and SuperGrok Plus users in the United States, and plans to cover more users in the future. This update provides users with a more flexible video generation process. Creators can generate videos directly through text descriptions, or upload reference images to control characters, products or scenes. They can also combine character pictures and voice samples to allow the AI to maintain consistent appearance and sound in different shots. xAI said that Imagine Video1.5 supports up to seven visual references in a single generation, which can be used to fix people's faces, product features or environmental elements respectively, thereby achieving more refined content control. In multi-reference mode, users can keep the identity of the character and replace the background, keep the scene the same and change the character, or just adjust the action while maintaining the overall visual setting. The voice reference function further solves the problem of character continuity in AI video production, allowing the same character to maintain stable facial features and voice performance in multiple clips. For developers, the xAI API already supports calling image references, text-to-video and native 1080p video generation capabilities through the grok-imagine-video-1.5 model. Developers can complete video generation by passing in parameters such as prompt words, reference image URL, video duration, aspect ratio, and resolution through the API. The voice reference function needs to be accessed through requests. Imagine Video 1.5 was released last month. xAI has said that this version has been optimized in terms of motion performance, physics simulation and audio generation. This upgrade further strengthens the model's ability to control identity, scenes and product details, making AI video generation gradually move from simple picture conversion to a multi-modal creation tool closer to the professional film and television production process. via AI News (author: AI Base)