I recently got a chance to try Sora 2, OpenAI’s newest AI video generation model, and I’m honestly still blown away.
Compared to the first version, Sora 2 feels way more realistic — smoother motion, better lighting, and far fewer of those weird flickering details that usually make AI videos look fake. The new model really understands camera depth, motion, and texture in a way that feels closer to actual film production.
What impressed me most is how easy it is to turn a simple text prompt into something that looks like a real scene. You can describe a shot — “a golden retriever running through a field at sunset” — and Sora 2 builds it frame by frame. It even handles shadows and reflections naturally now.
The only catch? It’s still invite-only, so not everyone can try it yet. But if you ever get your hands on a code, it’s worth exploring — especially if you’re into creative filmmaking or visual storytelling.
After experimenting with a few clips, I ran them through HitPaw VikPea Video Enhancer, and it made the results look even cleaner and more cinematic — like a finishing touch for AI-generated footage.
Can’t wait to see what people create once Sora 2 becomes public. Anyone else curious how this might change indie video production?