procedural generation
124 subscribers
5.18K photos
1.66K videos
4 files
11K links
Created by @r_channels
Download Telegram
Wish fulfilling ball in binary forest (In the novel "Piknik na Obochine" (Roadside Picnic) by the Strugatsky brothers, the mysterious and enigmatic "Zone" contains various artifacts left behind by advanced alien visitors. One of these artifacts is a sphere called the "Wish Fulfilling Ball" or "Gold)
https://redd.it/1522lua
@proceduralgeneration
How do you think this was made?

I've found this artwork very inspiring and I am struggling to understand how it was made
It kinda looks like some kind of datamosh but the subtle blends and whispers makes me think there some other techniques involved.

https://vimeo.com/360870079

Any idea?

https://redd.it/152kjdc
@proceduralgeneration
How to ensure validity of generated content

Hi all,

I guess this question is mainly addressed to researchers in the field of PCG, particularly the ones working on game design.

I am currently working on a PCG problem using reinforcement learning, and where I have to create new levels of a given game - let's say that the game some form of Mario bros.

My agent can position building blocks in the environment, following some trivial constraints. But of course, these trivial constraints cannot be perfect nor complete. Combining building blocks, which is what makes the levels more interesting, comes at the cost of making it harder to ensure validity of the levels.

By validity I mean, in the context of Mario, whether the level can actually be completed by a player.

My question is the following: what are typical methods in the research field to tackle this problem of ensuring generation validity? I understand very well that every game will have a specific way of tackling this (e.g. solving this problem for PCG on Maze is a different problem than solving it for Mario), but I was curious to know if there are common strategies, or some paper that review this specific issue in PCG research.

Hope that was clear, and thanks in advance!

https://redd.it/1538l1r
@proceduralgeneration
This media is not supported in your browser
VIEW IN TELEGRAM
Multivariate colormaps based on satellite imagery and digital elevation model
https://redd.it/153aq84
@proceduralgeneration
Various Unity noise libraries performance comparison

I've been doing a lot of benchmarking lately to try to optimize some of my heavier code, so I figured I'd post the results here in case others are curious.

​

# 2d Gradient Noise - 256x256

|Library|Computation Time|Uniformity|
|:-|:-|:-|
|FastNoiseLite Perlin|\~29 ms|Low|
|FastNoiseLite Simplex|\~28 ms|High|
|FastNoiseLite SimplexS2|\~29 ms|Moderate|
|Perlin Library|\~15 ms|Low|
|Mathf.PerlinNoise|\~16 ms|Moderate|
|Unity.mathematics Perlin|\~160 ms|Moderate|
|Unity.mathematics Simplex|\~140 ms|High|

It's important to note that Unity.mathematics was intended to be used with the burst compiler, so these results don't necessarily match it's optimal use.

​

# 3d Gradient Noise - 100x250x100

Run in a burst compiled, multithreaded job

|Library|Computation Time|
|:-|:-|
|Perlin Library|\~28.5 ms|
|Unity.mathematics Perlin|\~62.2 ms|
|Unity.mathematics Simplex|\~68 ms|

Interestingly, the Perlin library (which is almost a decade old at this point) seems to be 2x as fast as the newer, more highly optimized Unity.mathematics library. If anyone has an explanation I'm very curious as to why that is.

https://redd.it/1554zd2
@proceduralgeneration