Gamedev tools constantly change. 10 years ago, if you wanted to make a game, you had to make your own game engine first. That was a solid gatekeeper, allowing only strong teams to create games (even mediocre games, were technical achievements)

Nowadays, game engines are free^, plugins are free, tutorials are everywhere. Many assets are free - definitely enough to make any kind of prototype. But now we have reached the point you can literally generate free assets instantly on demand! Solo gamedev becomes extremely viable.

The making of an oldschool RPG

What art assets do you need to make such a game? Various backgrounds to fit the dungeons?
Here is a consistent generation of a single phrase, enough to fill a single dungeon:

Want cel shading? Here you go:

Artstyle doesn't match your game? Choose your artstyle and background, here is other various artstyles possible out of millions:

As you can see, you can have any kind of background you wish. So, what else is required for an oldschool RPG? Obviously, enemies!

Each enemy (type) in an oldschool RPG needs only a single front image. Just type Concept art of [MONSTER] and just copy-paste it into your game.

What remains? Art assets for your heroes/protagonists!

Images of your characters. You can make entire character sheets:

Portraits? Many games require those. Here is a bunch:

"You need to change character faces depending on emotion or dialogue"

lol look at this: https://rentry.org/faces-faces-faces#emotions

And by extension, yes, you can make your own original characters and use them for dialogues and cutscenes and combat.

Isometric Asset Creation

All art in this page being generated by AI == the solodev doesn't create art but edits it.

Basically, artists will slowly fall in the place of correcting and editing art, instead of creating original art. An accurate analogy is animators, who instead of making the keyframes from scratch, are hired to make interpolations between keyframes, and also various touchups.

Gamedev 2025 - 3D Model Generation

The next step for Neural AI Generation is generating 3D models from images. So you give it 3 images of the same sword (in the far future, just 1), and it generates a 3D sword as displayed.

Alternatively, you can just type what you want 'curved sword' and generate many variations of it. Of course, image inputs are better, since you want details in complex objects like houses.

I believe this 3D Model Generation will be used in production firstly for generic objects like barrels, rocks, vases, tables etc etc, but also for generating textures to apply on armors or other surfaces. Only after quite a while it will consistently get proper depth for complicated renders like insanely detailed armors.

What I don't believe will happen until 2030+ is AI being used in production to generate animations or bones.

With Neural AI generating any kind of art asset, indie gamedev will reach its final form, because a single dedicated skilled person can basically make any kind of game. I know I can at least, and by extension others as well. You can basically get infinite assets, all made in the same style of your game. But 3D won't be free like 2D art, because the computational power to create high-quality 3D models dwarfs the computational power required for 2D. So it is safe to say 3D artists will not be 'deprecated' (rip 2D artists), for many decades.

June 2023 Update: Animations used to need specialization. Now, AI-assisted animations make quality animations possible for everyone https://www.youtube.com/watch?v=b3OCTbT1NP4