Hi Sarasir,
Based on the pictures and your description it's difficult for me to find a plausible explanation for the noise difference and exclude the influence of unknown, but real, external noise sources.
If all the noise was generated inside the Owon the oil can would block most of it from escaping from its confines. However, I don't see how it could prevent common mode currents inside the Owon from being converted to differential mode currents when faced with an impedance imbalance on the coax. So in my mind, a plausible explanation is the influence of the oil can over external noise sources, in other words, its ability to block most of the external radiations and prevent them from reaching the Owon. With an unshielded Owon the influence of external noise sources would account for an increase of common mode currents within the Owon and as a result more visible noise. The reverse would be true when the Owon is inside the oil can. On the other hand, the 475, which by itself registers very little noise, is built inside a metallic cabinet with probably better shielding characteristics than the oil can. This seems to be consistent in my mind with the theory that some of the noise comes from external sources.
There are a couple of other things that in my mind may play a part in the big picture. Inside the oil can the Owon's screen, a known emitter of near field radiation, is prevented from spilling it onto the environment. However, looking at the pictures, I don't see any of the coax cables close enough for this to be a mayor player on the experiment results. The other thing that comes to mind is the role that the points where you chose to ground the shield may have on existing common mode currents inside the Owon. By redirecting some through a potential shorter path there may be fewer available at the coax when faced with an impedance imbalance. However, without a way of visualizing these currents choosing the best grounding points may be a daunting task.
Anyway, let me know what you think!