A naukri.com initiative
Siliconangle
9h
17
Image Credit: Siliconangle
Midjourney debuts new V1 video generation model
- Midjourney Inc. unveiled its new V1 AI model, capable of generating videos up to 21 seconds long.
- The company initially focused on AI image generators and now caters to around 21 million users.
- Midjourney generates revenue through a subscription-based cloud service allowing users access to its models.
- Its latest service includes a gallery feature with an option to animate images using the V1 model.
- V1 can extend the initial 5-second clip up to 21 seconds, surpassing the capabilities of Google's Veo 3 and OpenAI's Sora models.
- Users can let V1 autonomously animate images or customize the process by providing prompts.
- V1 offers settings to align with the user's prompt closely or add creative elements for a unique touch.
- The model allows customization of motion levels for different scene types, from low to high motion.
- V1 arrives after the release of Midjourney's V7 AI image generator, known for faster speeds and enhanced image quality.
- AI generators like V1 typically utilize a diffusion approach, reconstructing images to create new ones.
- Video generators like V1 feature temporal modules to maintain object consistency across frames and frame display order tracking.
- Midjourney plans to focus on training AI models for interactive 3D simulations, with V1 being a stepping stone in this long-term effort.
Read Full Article
1 Like
For uninterrupted reading, download the app