Comment by vikmex

Hey! Yes, it's using image-to-video. I went with a model by KlingAI as it seemed to produce the best results from my tests.