I recently found setosa.io . It got a page with an excellent visual explanation of Markov chains.
I guess that Markov chains are just a probability selector among different states. From this I wonder if I could generate some idea.
I wonder if many states together with their probabilities could form a non random image. That could perhaps look like an ocean from top view.
Then with image processing tools I guess it would be possible to enhance that model. Sharpen the image with google super type image resolution and other image tools like noise filters.
So basically. Every parameter cluster of a model should ?perhaps be a non random looking image.