Pika Labs has recently announced their new Image2Video model, which is now available for free on their website. This model offers significantly improved quality for videos created from both images and prompts. Typically, users will get a decent-quality video on the first attempt. All videos are still 3 seconds long, and users can apply all the editing features previously available on Pika Labs.
BREAKING: Pika Labs released a new text-to-video model. All generations have a very very good quality from the first prompt 🔥
— TestingCatalog News 🗞 (@testingcatalog) June 11, 2024
"Cyberpunk hacker" example I've just made. All vids are 3 seconds long but can be enhanced further. https://t.co/JdFyEs7kOP pic.twitter.com/kSdHyLR0jO
People have already started generating many cool videos with this new model. While some say that the overall quality is only slightly better than before, others report that the quality of videos generated directly from text prompts is now much, much better. You can try it anytime at pika.art 🔥