I was asked to give critique for this painting: iyulian.deviantart.com/art/the…. I see a paintover as one of the best ways to show suggestions. In fact some of my most valuable learning experiences involve people doing paintovers on top of my own paintings. And the help came from people that I don’t even know. So I thought it’s time for me to do the same!
For realistic results one always needs to have proper reference. Searching for “mushroom” in DeviantArt stock photos I found this: maadobs-garden.deviantart.com/…. It gives a nice reference for the “building” and also for the vegetation around it. And the vegetation is needed to give the sense of scale. My version has a larger scale as the original is really tiny if it fits among the grass. On the other hand there is also smaller grass growing on the building itself, so I’m a bit confused about the intended scale.
To avoid a flat image one needs distinct foreground, midground, and background. They can be separated by using atmospheric perspective — that is using lighter values and less contrast in the background. The original painting has the darkest values in the background, which is fine when depicting darkness in the background (such as night), but here the foreground is also lit although the visible lights are in the midground. Speaking of lights, I think there a too many of them competing for the viewer’s interest. I left only the ones that were off the dead center. A weaker warm light could be added for one of the windows to show someone is inside at the moment. For a natural outdoors look there needs to be temperature contrast. In daylight it would be warm sunlight vs cold shadows that reflect the sky color. In this case it would be warm lamps contrasting the dimmer blue light (again the sky) from above. For added depth I had the lamps go in perspective from the foreground to the midground.
My version is way too rough for a real painting, but I think a quick sketch like this demonstrates the points mentioned above. Still doing this had me realise my own weakness, which is the usage of random brush strokes to depict some detail that I’m just too lazy to do correctly. But the fact is, spending a bit more time to determine the correct brush strokes would in the end save time, since it wouldn't need any adjustments or repainting later. There is no excuse for doing just random scribble in hopes it would turn out good. Every probability says it won’t.