How bad of an idea is this procedural game design?

This is my first attempt at building a game. I've made a good bit of progress but the further I get the more I keep doubting my design decisions:

I'm building a procedural turn based low-poly rpg that can have very large maps.
- Each tile is it's own game object.
- Each object on the tile is it's own game object.
- Right now I am not combining any meshes because I want the player to be able to interact with individual tiles /objects while in combat and in builder mode (where they can change the mesh of tiles/objects)
- I am using real time lighting for shadows right now...

I wrote code to chunk the map up based on current player location, but each chunk is 100 tiles, and I load the chunks adjacent to the player also so at minimum there are always 900 tiles loaded and a few hundred objects.

Performance is ok on my desktop. Still above 70 fps. WebGL is already a little laggy but playable since it is turn based.

Am I prematurely worried about optimization or is this design doomed? Any advice is appreciated!

 

Target platform is the question.

Be sure to selectively disable shadows on things that don't need them.

Seems you focused on technicals but I want to share one opinion I have about procedural design: I've never played a game with procedural levels that hooked me long enough for it to matter.

I think focusing on technically complex level design before classic, linear, hand placed level design is backwards way to go.

 

If I were only targeting pc would you say the design is acceptable?

 

Absolutely. Maybe chunk size can be messed with to find optimal solution, but that's simple matter right.

Always want to identify the factors that will be important and design them to be modular. This way you can easily play with options to find best solution.

 

Check out Unity DOTS. Traditional game objects are performance heavy.

 

Thanks for the replies.

Will do! The tiles have a monobehaviour but they don't even have an update method so something more lightweight would be great.

 

Since you are doing procedural generation, optimisation is important to consider right from the start, since you might end up calling it 100s, 1000s, 10000s of times a frame, so even the smallest improvement are multiplied depending on how much you are going to need to call it and anything saved can be used to calc, draw or just improve performance