NeRF models are trained on several views with known location and viewing direction. This model takes one image (and you don't need to train a model for each object).

But if it takes only one image, isn't it likely to hallucinate information?

Not just likely, it does. Try out the demo and see, e.g. what the backside of their Pikachu toy looks like. Or a little simpler, the paper has an example (the demo also has this) of the back of a car under different seeds.

Not hotdog.