Seedance 2.5 turns text, images, and references into cinematic AI videos—text-to-video, image-to-video, flexible ratios, private generations. Independent freemium T2V/I2V product (not ByteDance).