The reason we tend to get rid of obvious striping patterns is because they are obvious when there's multiple of the same tree on-screen. The easiest solution is to figure out a good workflow that lets you create several of these trees in short time. Then you spend the time making as much variations as possible. That or you create the perfect tiling tree, which is nearly impossible and even really REALLY good artists, like the ones that did the forests of, say, Chrono Trigger, just barely make it work. We tend to emulate older art and thus drag with us certain limitations that don't exist anymore. Keeping an indexed palette has a bunch of merits. Conserving tile space just stifles tile sets, even though it can be fun training exercise to maximize creativity within bounds. I don't recommend it for commercial projects, just like I wouldn't recommend making a NES-compatible game (versus one that looks and feels like it's NES-compatible).
This "tile sets lack variety for no good reason" nail is one I'll be hitting a lot, apologies in advance
. It hurts my brain when people try to make the perfect tiling tile, when the solution is usually variety instead. Even very restricted SNES games have at least a few different grass tiles for example. Trees are a very good example, because it's such a daunting exercise to create a tree that handles on-screen copying well. By being so large, any small striping detail (like the dark horizontal band you have now) will look ridiculously obvious.
So how many trees do you need? A general average that I see a lot is 3 free-standing trees and some sort of tileset that allows one to create "thickets", aka dense forest, for use as impassible terrain. If I were to create a general purpose tileset, this is the list of assets I'd create:
- One basic tree, with extreme stripe removal
- One variation, with more character
- One with a different color to splash a secondary color
- One big tree, as a place-of-interest
- A set of tiles to create thickets, that mostly match the basic tree
