Ever since blocky and fudgey 3-D graphics hit video games, we've heard about how "cinematic" they began to look. In the rare places where I've read people talking about the visuals of video games as a medium, that's what they tend to compare them to -- especially movies, since they show moving pictures, but also painting or still photography.
But video games are interactive, and in particular the player has a fair degree of control over the camera. This makes it totally unlike movies, except for having a time arrow.
In movies, the visual team selects the set of images to show, in what order, at what tempo, etc. In a video game, the player takes over those roles. Even for a single still image within the larger flow of images, the visual team making a movie determines what the composition will be -- what's in frame, how the pieces relate to each other and the background, and so on. In a video game, that is all up to what the player decides to have the camera do at some point in the flow.
Naturally then the visual experience of video games will pale in comparison to watching a movie, as the movie experience is the result of more talented minds thinking over the outcome much harder, while the video game experience is the ad hoc choices of the average person. Cherry-picking the best-looking video game and the worst-looking movie does not disprove the overall point.
If they're not like movies, then the closest medium I can think of is landscape architecture. There too the visual designers have only minimal influence on what the experience will be like, mostly populating an empty space with the basic objects that the spectator will arrange into a stream as they please, and in relation to each other as they please. Will that tree be in their field of view or not? If so, will it be seen from a direction that includes that hill over there or not? This nearly total lack of control over the most fundamental aspects of composition must drive landscape architects crazy, and could explain why visual people go into fields where they have greater control.
This also explains why video games looked better in 2-D -- it gave the player less control, since an entire dimension of where to point the camera was not available. The designers knew that you'd be looking at the world head-on, so they could anticipate a good deal about what their compositions would look like. You could still scroll left-right or up-down, splitting up a field of objects that they intended to be seen together, or picking up two clusters at once that they intended to be seen in separate viewings. Still there was more influence from the people who had a better eye and who were working more tirelessly to put out a pleasing product.
Earlier I pointed out that video games will probably not endure as a narrative medium because they are too interruptible, like radio programs or comic books. This harms them as a visual medium too, since the flow of images is too interruptible, unlike the pre-ordained flow of images in a movie. Giving the audience member too much choice, as video games do, is another aspect of making them more interruptible: they have to decide what order the events of the narrative will unfold in, and they have to decide what the composition of a visual "shot" will be.
A choice here or there might not be so bad, but having to shoulder most of these responsibilities keeps the spectator from drifting into a dream-state where they can just absorb the video game. Just as they're about to get lost in the experience, the player keeps getting snapped awake by having to make yet another decision.