I have worked on 3D shopping - this seems like a huge step up from the world of photogrammetry to NeRFs to veo. When it comes to shopping - the realism matteres the most. Its not just about 3D model but shadows of light, texture, how it looks in different settings (for e.g. if it's furniture item in a larger scene). That's where the rubber meets the road. Regardless, I'd be curious - how fast can we create 3D models from existing imagery/videos.