The comparison is very big, because Sora really falls short of Veo2, but I feel that Sora only gave us access to a very limited part of its system, as I remember that at the time of its release it showed us very realistic results and that it respected very well the physics of the characters and many details that made us see Sora as the leader in the generation of videos.
I think Sora is also very good, only that the model accessible to us as users, is a very limited model.
I will try to send the form to see if I get early access to Veo2.
RE: Google DeepMind Veo 2.0 | The most accurate Video Generation Model of the moment!