Endless Mode’s End…?

Bad news bears: my computer died! I make daily off-site backups (something I’d recommend to any dev) so Bleed 2 wasn’t destroyed — phew. Still, a frustrating amount of time was spent resurrecting my computer and restoring files, so I didn’t manage to finish the sixth level. Next week for sure!

In the meantime: an update on Endless Mode, and why it may fall by the wayside. Really uplifting blog this week, I know! I gave Endless Mode a few months of work after pondering it in December, and I thought I’d show how far it came and discuss some of its issues.

For all who don’t know, Endless Mode is me jumping on the randomly-generated bandwagon for Bleed 2. I created a bunch of bite-sized level chunks to be randomly stitched together into full maps. In December I had five chunks, and now I’ve raised the count to thirty.

Some of the created chunks!

I tried to design the chunks with variety in mind: mixing up the enemies, hazards, ceiling heights, visual elements, direction they lead, etc etc etc. Mathematically, having thirty chunks makes for a very high number of combinations — over a trillion, in fact! Wowie zowie, gee whiz!

A sample of some of the levels that are created.

So that sounds impressive, but the reality is a bit of a letdown. Chunks are tiny, so the effect of their varied design is super minimal. A chunk that changes the action by making you travel down, for example, is over in seconds and barely registers. Other aspects just straight-up weren’t executed successfully, like varying the visuals — some chunks have pipes in the foreground, some have them in the background, but most chunks are visually busy and have the same general patterns in the placement of the elements, background features, etc.

In short, I can randomly generate trillions of levels that aren’t very exciting and all blur together. They feel like running through a big tube that has occasional kinks. They aren’t very challenging either — I’m zooming blindly through them, effortlessly destroying everything in these gifs.

And it’s not like I’ve played this before. It’s random!

Randomly-generated content is still a fascinating puzzle to me, and I’d love to keep figuring it out. The first solution I’d try is to be more extreme with the chunks — if one travels down, make it go WAY down; if another features an open, free-form space, make it REALLY open. I’d also develop the algorithm that creates the levels, because right now it’s very simple. Once I do solve (or at least improve) all this, I’ll have to repeat the process many times so additional environments have their own sense of personality and design. It’s an unknowably large amount of work.

In the end, I doubt I should invest the time if I hope to finish the game relatively soon. Best-case, I could try and develop a single, generic environment to ship in the release, but even that’s uncertain. It’s still something I’m interested in, so I may come back to it in the future! Right now I’m going to focus on what I believe is primarily desired from Bleed 2 — the main campaign. Hopefully I’ll have a happier post about completing another level of it next week!