This is a difficult question to answer in a historical context, because it is asking why something didn't happen, which is always trickier than asking why something did. I'll try to answer that question in a general sense in a follow-up post, but I think we first have to ask another question:"how do I know when a painting is realistic?"

To many people, this feels like a strange question. We often think of "realism" as something inherent and objective to certain works of art. We often think that certain paintings just "look like real life" while others don't. But in truth, the act of painting a three-dimensional scene in a two-dimensional medium always involves trade-offs and compromises, and our idea of a "realistic" painting is culturally and historically constructed. It's also a much more complicated, and often self-contradictory idea than we realize. I would argue that when we think a painting is "realistic" we are often using one of the following metrics:

1. The Wil-E-Coyote test. According to this model of realism a painting should literally trick the viewer into thinking there is no painting there at all. In practice however, this kind of trickery is only possible under very specific conditions. Modern, "correct," linear perspective (more on this later) assumes a viewer at a single, fixed, position. So, maybe, if the roadrunner viewed Wil-E's painting at exactly the right point, he might be fooled into thinking he was just looking at the road continuing, but as he moved from left to right, or approached to canvas, he would immediately understand that something was wrong.

You can get around this a little bit by painting things with very shallow depths of field-like this trompe-l'oeil painting by [Evert Collier] (https://en.wikipedia.org/wiki/Evert_Collier#/media/File:Edward_Collier's_trompe_l'oeil_painting.jpg). When depicting things that are already almost-flat, you can trick the eye more easily. But of course, this limits your subject matter quite a bit, and this sort of explicit trickery is more often treated as a kind of curiosity. You might be able to trick a viewer into thinking he saw some letters on a rack rather than an oil painting, but there is simply no way to make someone looking at a painting think they are actually observing the bombardment of Algiers. So clearly, "realism" in painting means something more than just tricking the eye. Which beings us to metric #2.

2. The perspective is "correct." One of the hallmarks of the types of painting that we consider "realistic" is that the representation of space adheres to certain rules of perspective that were more or less formalized in the renaissance by people like Alberti, Uccello, Piero della Francesca and others. If you google "linear perspective" you can get any number of explanations of these rules and how they work, usually with the implication that if you follow these rules you can depict space as the eye actually sees it. Unfortunately, this is bullshit. The way a human being perceives space is fundamentally different from how any painting, even one employing the most "correct" perspective sees it. For one thing, we have the problem stated above, that a painting in perspective assumes a viewer in a specific, static position, which may not be the position in which the viewer is actually standing.

Secondly, most humans have two eyes, rather than one, and this causes all sorts of problems. Of course, having binocular vision means we have depth perception, which is something a painting can't approximate, unless you did duplicate paintings and set them up in some kind of stereo viewer. But binocular vision does all kinds of funny things. While I was thinking about this question this morning, I was staring out of my window as I made coffee. As I did so, with my eyes focused on the yard outside the window, I started to see the mullions in my window in double: each eye was seeing them seperately as I focused on the scene behind them. What's more, the pairs mullions appeared strangely transparent, as my brain superimposed the different views taken from two different eyes on top of one another. If I wanted to paint that scene "realistically", should I show one set of opaque mullions in the window, or two sets of transparent ones? The first might more accurately reflect the material reality of the scene, but the second might more accurately reflect my perception in that moment.

Linear perspective gets other things wrong too, of course. This gets a little technical, but the geometry of linear perspective assumes that the images in the scene are being projected onto a flat plane: the canvas. However, in the human eye, scenes are projected onto a curved plane: the back of the eye. The result of this is, among other things, that straight, parallel lines projected onto the back of our eye would actually appear to curve. In practice, we don't really notice this because our brain does all sorts of calculations, corrections, and fudging with the hard input we get from our eyes. There are many other quibbles to bring up with linear perspective, but the gist is that linear perspective is a set of rules, which captures some truths about how human visual perception works, but fails to capture even more. These rules also tend to rationalize or standardize the messiness of human perception into something more mathematically precise than it actually is. With this out of the way lets get to the third metric of "realism:"

3. It looks like a photograph. During the 20th century, the photograph has become kind of the high-water mark for realism in a painting, and the term "photorealism" has come to be used as a kind of synonym for "realistic." The path to this happening is an interesting one, and a historian of photography could no doubt say a lot about this. For the time being, however, I will point out that photographs have a lot of the same problems that I mentioned above: they assume a fixed and unmoving vantage point, they project images onto a flat plane, rather than a curved one, they are generally monocular rather than binocular (although stereo photography has a long and fascinating history). Also, as with paintings, photographs differ from human perception in so many different ways. Think about light, for one. As I write this, I can turn and look out a window at the snow-covered roof of my garage. The light reflected of that snow is currently so bright that it is painful to look at for more than a few seconds, and when I look away I see purple-ish "afterburn" in my field of vision. Thats a very real part of my visual perception of my current moment, but a photograph wouldn't capture that.

Also, the idea that photographs are impartial, perfect replications of a given scene is also nonsense. A photographer taking a picture of a scene makes all sorts of decisions: he or she sets their focus, their aperture, and their exposure length. They can develop the photo for longer or shorter periods, using different chemicals. Different film stock will give different perceptions of colors. Two photographers could take two different photos of the exact same scene, and end up with images that look nothing like one another. How would you know which one is more "realistic?" Is color film more "realistic" than black and white?

.....Anyway. I could go on and on about this. The big point is, is that there is no objective standard for what "realism" is in painting. The kind of paintings we consider "realistic" are ones that capture certain realities about human visual perception, while ignoring or distorting others. We have been trained to accept some things as important to "realism" while glossing over others. That said, we can acknowledge that, culturally constructed or not, most people have a certain idea of what "realism" is in painting. The question then becomes, why doesn't art from all periods of history meet this standard? To answer that, we need to go to part two: