Thank you - excellent points and great idea. I've encountered similar issues when manually zooming - plus it makes some of the leaderboard information disappear when the text gets too large.
There are a few ways this could be implemented. One involves enlarging the entire VM window to a final limit of the screen width. Another approach involves maintaining a constant VM window size, but zooming in (centered on the vessel) such that part of the image around the sides will get pushed out of the visible bounds. The advantage to the latter approach is that you could zoom almost indefinitely.
In some of our early prototypes, we pre-zoomed such that the VM would show just the area immediately surrounding the target vessel. Ultimately, we reverted to a "full" view because our annotation experts explained that for very difficult vessels, the surrounding context can be useful. We had been inadvertently creating a "Forest from the Trees" issue. So even if we do implement a "centered zoom", we will likely reset the zoom for each new vessel image to ensure that people see the whole picture before they zoom in. For a full screen zoom we would likely maintain the zoom level once it was set.
In short, providing better options for magnifying the VM is on our to-do list! Thanks again.