I started off simply cribbing all the tiles from Ultima IV for the Apple II, then gradually adding some rudimentary new tiles as the need arose. Starting with a pixel "rock" and "stick", changing the clothes on existing characters, then eventually gaining a bit of confidence and launching off on more complicated things. Eventually coming back and redoing all the "borrowed" tiles, and launching off into new, more detailed, characters and items.

"Constraints hide your flaws" got me a long way. I've relaxed those constraints a bit as I got better at shading, so it's easy to tell which tiles were drawn at what point in my "career"

[1] https://valtima4.com/, the Survival Crafting RPG you would have played on you Apple II in the '80s. It's essentially Valheim crammed into Ultima IV's interface.

Single player works up through the first couple bosses, but it's not really ready to ship into early access yet.

Quite the opposite. The fewer pixels, the more each one has to be perfectly in place. Honestly should've been obvious in hindsight. If I have any games left in me after my current one's finished, I'll just use as high a resolution as I'm comfortable with.

Unless the sprites are truly tiny, like 16x16 with 2 or 3 frame animations, I don't know if pixel art makes a good shortcut to an aesthetically appealing game. Then again, it might be easier than six years of every day practice.

Contrast seems to be also one of those things I see even very experienced pixel artists get wrong, especially in the context of games where visibility can be crucial. There's many games with beautiful spritework but actually playing them is very tiring because all of the sprite work uses the same range of the palette. You want to consider if things are in foreground or background, interactable or not, dangerous or friendly and then limit how much the range of their colors intersect. Creating contrast through hues is more common (, see red hostile vs blue friendly), but differentiation through saturation and value/luminosity is much more effective and readable at a glance while also being more accessible to color blind people by default.

Full stop. There are quite a few coders with artistic talents. And even if some specific individual does not have such talent, they are allowed to have their own taste - we do not need to train ourselves to mimic other people's preferences.

Another thing that helped me was to experiment with different canvas sizes and styles. I was surprised how changing these affects my process, speed and results. Then again, this can be difficult in an ongoing project.

[1]: https://apps.apple.com/app/nonoverse-nonogram-puzzles/id6748...

Also, just like in coding: Constraints don't hide your flaws (per se); you fuck up, people will (let you) know. And pieces in constrained environments can be much, much harder to pull off.

I had hoped for something closer to the intersection of pixel art and graphics programming. Well, maybe in the future.