Pika Labs launches its AI video model Pika 2.0, allowing video generation from custom pics

Pika Labs has unveiled Pika 2, a significant upgrade to its AI video model, enhancing motion, consistency, and customization. This update allows users to have more control over video elements by uploading "scene ingredients," which facilitates the integration of custom characters, objects, clothing, and settings into their videos. Pika 2's advanced image recognition capabilities enable it to understand the roles of reference images, creating cohesive shots with smooth transitions while accurately following text prompts and capturing complex instructions.

The model demonstrates a strong grasp of physics, delivering both realistic and fantasy motion, such as natural human movements and imaginative scenarios like humans flying or alien environments.

Pika Labs highlights the model's ease of use, catering to both casual and professional creators, marking it as a strong contender in the AI video model market, especially now that OpenAI has recently launched the highly anticipated Sora.