99
Created by @r_channels
hydraulic erosion, even though that does appear to be potentially capable of what I'm aiming for (which I don't know for sure is the case, to be totally honest) because by my estimation it will require about an order more compute to produce a viable result than a proper lattice-based method should. Simulating a bunch of particles that are randomly accessing memory, and only affecting wherever they are, just doesn't seem like it makes as efficient of use of available compute (as well as hardware cache) as well as a clean uniform lattice solution. (i.e. the particle based simulation will be slower and more expensive). The project I'm tasked with at the moment entails generating a square of terrain that's 31.2km to one side at a resolution of 2048^2 for the simulation heightmap - so 25 meters to one pixel.
This seems like a perfectly solvable problem using a lattice-based method, and apparently Musgrave isn't the solution I'll be using, unless someone has managed to get it to do more than eroding rocks into sand and is willing to share what they did. :P
I forgot to mention that the primary aim of the project is for the formation of a proper watershed - rivers and ponds and lakes, along with the mountains and hills that are formed in tandem with the watershed. Musgrave definitely does not appear to be capable of forming proper rivers - it seems to only simulate eroding rock/sandstone into puddles of sand, and that's about it.
If anyone has successfully formed V-channel valleys and river systems using a lattice-based hydraulic erosion method I'd love to hear about the details, if you don't mind sharing. Or, if someone has implemented Musgrave and was able to generate land formations such as this: https://imgur.com/NqTj2Ep I'd love to hear about your implementation.
Anyway, I thought I could've figured this out by now but its turning into a much greater time sink than I had planned for, so I thought I'd turn to reddit to crowdsource the hivemind for some insight/halp/etc. before I resort to more drastic measures.
Thanks! :]
https://redd.it/1gqcl3e
@proceduralgeneration
A procedurally generated creature navigating a procedurally generated planet playing procedurally generated animations (sorry)
https://redd.it/1gq2924
@proceduralgeneration
Planet generator for a spherical strategy prototype (wip)
https://redd.it/1gpxmgj
@proceduralgeneration
curly scribbles
https://redd.it/1gpp5io
@proceduralgeneration
Meet the Hell Demon Slime from Legends of Elementia! This fiery enemy transforms from its harmless-looking slime form into a fearsome Hell Demon, unleashing devastating fire attacks. Beware its unpredictable nature and explosive power! 🔥
https://redd.it/1gpfj8i
@proceduralgeneration
Modular Tiles [javascript + canvas]
https://redd.it/1gp0fea
@proceduralgeneration
Making my Procedural 2D RPG Immersive
https://youtu.be/bxJ4KuLk37E?si=QOEfMdPXNnjcV4nd
https://redd.it/1gouo8w
@proceduralgeneration
Adding procedural gegenrated alien_like grasslands to my game :) will add mor procedural generated obstacles and creatures to make it more immersive soon :)
https://redd.it/1goegp0
@proceduralgeneration
I’ve been developing a sci-fi space exploration and strategy game called Beyond Astra for five years, where you build your own civilization, manage cities, and lead real-time battles across the galaxy. All worlds are procedurally generated. What do you think?
https://redd.it/1go2zwr
@proceduralgeneration
Bird
https://redd.it/1gnhc4v
@proceduralgeneration
How to deal with floating point errors on terrain generation.
PLEASE READ THE POST BEFORE ANSWERING.
I know how to implement Floating Point Origin and I moved most of my generation code to work in local space.
Together with Floating Origin I can get pretty high travel distances.
But there's just one thing that I can get it working nicely.
To simplify the problem, let's say that I have some 2D perlin noise that represents the terrain. To get continuous generation I make use of the vertex world position to determine which perlin height value should be there.
The problem is that when using the world position i end up with floating point errors since the x, z coordinates have really high values. How should I approach it? If I use local vertex position all chunks will look the same, but as soon as I ass the origin to the point sample i get the floating point errors.
https://redd.it/1gn6xg3
@proceduralgeneration
Making a procedural manta ray
https://redd.it/1gmv1rn
@proceduralgeneration
I'm working on a procedural planet builder called Lagrange, here it is!
Hello! Some friends of mine have been suggesting that I post my project here, sooo... here it is!
Current 0.3 logo. Every update will have it's own, because why not lmao
# What is Lagrange ?
Lagrange is a procedural planet generator that I've been wanting to create for years, and I'm really proud of what I've achieved so far! Essentially, it allows you to create your own planets, including surface terrain, clouds, atmosphere & biomes, with many more features to come! You can also share them with others by exporting them and importing them back into the app.
This is mainly a way for me to learn more about shader programming & threejs, but it also could be useful for all sorts of worldbuilding, lore-making, TTRPG stuff, etc.
I've been inspired by a few projects I've stumbled upon years ago, such as the 2013 PlanetMaker Chrome experiment, and more recently, NASA's Eye on Exoplanets app (oh, and Ratchet & Clank :3).
# How does it work ?
Procedural generation relies on GLSL code with implementations of fractional Brownian motion, as well as atmospheric scattering using Rayleigh & Mie theory. The web presentation layer uses VueJS, and the 3D engine is supplied by threejs. GLSL code is available in the src/assets directory on GitHub if you'd like more details (link below).
Controls on the UI are directly linked to material uniforms from threejs, and are updated in real-time as soon as a value is changed. This keeps things snappy and smooth-ish... as long as you have WebGL2 capabilities however (WebGL1 is not supported by threejs).
Sections of the project rely on DataTexture instances to more efficiently relay colour data to the GPU, such as the surface & biomes.
As for storage, I'm using IndexedDB to store all app information, so everything runs locally in your browser. You can change language, keybinds & themes at will in the in-app settings. (also there's an easter-egg somewhere... can you find it ? :3)
Current milestone in progress is version 0.4, which will add rings & additional surface noise options to create more varied terrain (as well as gas giant-ish planets). These include adjustable domain warping & curl noise options, as well as XYZ warping/stretching.
# Where can I try it ?
Source code, app information & link here: https://github.com/EepyBerry/lagrange
Important: I'm changing licenses for future versions; you can look at branch dev/0.4 for the new license. I don't want bad actors using my project for ill intent.
Thanks for reading <3
https://redd.it/1gmij0m
@proceduralgeneration
Procedural Generation of 3D Tomato Plant Models
Hi everyone,
I’m currently working on a project that requires generating multiple 3D models of tomato plants, each representing different stages of growth. I need to create a wide range of models at various ages and not just a few.
I've done quite a bit of research, including reading academic papers, but I haven't found a clear or straightforward method to achieve this.
Does anyone have any experience or know of any tools that could simplify this process? For example, there are programs that can easily generate trees in various growth stages. I’m looking for something similar but for tomato plants.
Any advice or pointers would be greatly appreciated!
Thanks in advance!
https://redd.it/1gly5uk
@proceduralgeneration
The latest version of my procgen terrain allows the user to change it completely in an instant
My Terrain
https://preview.redd.it/1qp5cssmjhzd1.png?width=1599&format=png&auto=webp&s=3c05f92455c03e5e5eb9d2d0e3c54392a9e7a2b0
https://redd.it/1glr1i6
@proceduralgeneration
Lattice-Based Hydraulic Erosion For River/Valley/Canyon Formation?
I implemented Musgrave's hydraulic erosion (and a few other variants derived from the genuine article) but the very best that I ccould get them to do makes everything look like rock being eroded into sand, like this: https://imgur.com/a/DzVfxhE
I cannot, for the life of me, get it to erode a channel and then continue eroding that channel into a "V shape". I've spent a week (or two) fiddling with parameters; water velocity, erosion/deposition, rain volume/count/evaporation, etc.. I've discovered a few interesting things the algorithm can do but none of them result in the formation of mountainous terrain and flowing rivers - just more variations on eroded rock. Musgrave is very sensitive/finicky about parameters too and it's been a chore just getting it to do anything that looks decent in the first place. That's fine, it's just not what I'm going for with the project I'm currently tasked with.
I admit that I did make the mistake of assuming that the classical V-shaped channels that have eroded into the earth resulted from wide shallow erosion that progressively eroded the center more over time as a product of moving faster than the edges, and I only recently discovered that happens to not be the case (doh!) Such formations occur by first eroding a small narrow channel and then the sides erode inward, and the center keeps eroding deeper, and the sides continue eroding in - resulting in the clean 45(ish) degree slopes on the sides that can be seen all over in nature as the thing widens and deepens simultaneously.
It seems like Musgrave should be able to do this except that in my implementations, once a channel forms there's no erosion happening inward from the sides, or it doesn't erode any deeper. It either just gets deeper but never wider, or wider but never deeper.
I've arrived at the assumption that this is how it behaves for everyone who has implemented it - though I suppose it's totally feasible that I somehow missed or overlooked something, but after messing around with it for a few weeks, going over the original paper and others based on it, I'm really not inclined to not think that my implementation is the problem. It's just a limited algorithm in its concept. If anyone has implemented Musgrave and had it not look like rock eroding into sand, and had it form mountains and rivers, you are who I am looking to hear from! :]
Musgrave does erode along the sides of formed channels, but it doesn't deepen in the center at all - it just widens the existing depth of the initially formed channel where fluid has initially flowed, because the accumulated fluid from higher up that's flowing through it is already loaded up with sediment, and it won't take anymore while en-route to an area of lower velocity where it deposits the sediment. The sides erode from fluid flowing in from the sides - which is what you want for V-channel formation, except that the center never grows deeper where that's happening. The entire width of the flow grows deeper uniformly, if at all (because that entails not already being loaded up on sediment) like ribbons of erosion. Increasing the erosion factor just loads the fluid up with sediment up faster which produces virtually the same result, and decreasing the erosion factor just causes it to basically not do anything until it coalesces with enough from other neighboring channels to erode - and then it's only either eroding the sides of some other channel downstream that ends up being shallow too or eroding deeper the channel it's already in. The parameters seem to only control the width of the initial channel depth (affecting the flow of subsequent rainfall) and then it can erode the channels wider or deeper, but seemingly never both.
I've varied the parameters by orders of magnitude in all directions and tried every possible permutation of them. I'm left to conclude that Musgrave (and its descendents) are simply not capable of eroding a V-channel and forming a proper mountain river.
I'd rather not settle for particle-based
Audio Reactive 3D IFS: Satisfaction
https://redd.it/1gpyn72
@proceduralgeneration
Voxel planet generation for a Spore successor I'm developing.
https://www.reddit.com/gallery/1gpuagv
https://redd.it/1gpul1e
@proceduralgeneration
i built a room generator with up to 10^40 unique variations in webgl
https://www.youtube.com/watch?v=xKV0-ttXiH0
https://redd.it/1gpna4u
@proceduralgeneration
INNOVATIVE ELEMENTBENDING COMBO MOVES IN LEGENDS OF ELEMENTIA
https://redd.it/1gp14e7
@proceduralgeneration
Rainflow Circuit /// a colorful trippy musicvideo mograph animation made completely procedural in blender geometry nodes (and shader nodes)
https://www.youtube.com/watch?v=43CJxus2VTU
https://redd.it/1goxkd0
@proceduralgeneration
Gravity Well
https://redd.it/1gogb9w
@proceduralgeneration
A procedurally generated radio station
Hey guys, check out my proof of concept procedural radio station, that uses several AI models to generate conversations with the hosts and it plays AI generated songs. Hope you like it.
https://github.com/00Julian00/AI-radio
https://redd.it/1go8ju8
@proceduralgeneration
Gravity as a pencil
https://youtu.be/uXI5ZI_y34A?list=TLGGOxEOcXR-m8oxMDExMjAyNA
https://redd.it/1gnp3g3
@proceduralgeneration
Audio reactive geometric craziness Pt2
https://redd.it/1gn7oso
@proceduralgeneration
Procedural Interiors - Tips?
I am working on building procedural "dungeons", I have done some work in generating the shape and path of the dungeon including generating rooms, but now I have hit a snag...
My dungeons are man made, and some sort or order is expected, but I cant seem to find a way that gives me "good enough" results.
Lets assume rooms are certain area (they are not rects, or squares, but are grid based). I now want to generate Decor and facilities areas but I am struggling to make some semblance of structure inside those rooms.
I have tagged the rooms with a theme, which selects what can show up in the room, but the actual placement is what I cant seem to get right.
I have thought of using trying to use Wave Function Collapse to see if I can generate something since the rooms are 3D, but the grid like structure seems very rigid, and in reality my domain is very extensive, which I know WFC has trouble with.
I am not looking for a solution per se, but more of ideas on how to handle this.
https://redd.it/1gn2hbv
@proceduralgeneration
New trailer for the 1.0 release of the game I'm developing with procedurally generated planets and spaceships
https://redd.it/1gmovyf
@proceduralgeneration
Simple way to generate items?
Hi everyone, I want to make a sort of facebook marketplace simulator, where you can resell items and such, however, I don't want to spend a lot of time making items and still have repeats, so I want to find a way that I can procedurally generate things to be listed. Any tips would be great, thanks!
https://redd.it/1gmbh3p
@proceduralgeneration
asynchronous - python + gimp
https://redd.it/1glsr2a
@proceduralgeneration
How to go about map generation without making all rooms connect?
Hello all, i have a small issue. currently my map generation creates maps in a branching way. where one hallway connects 2 parts of the map, which is nice and all but I don't know if that's okay.
https://preview.redd.it/ktm44h7uzgzd1.png?width=987&format=png&auto=webp&s=7d0fa197d5dfafd2840cbf25b78cd32146032ff1
My plan was to implement hallway connections between 2 rooms that are really close, but i cannot do that without making the entire map a hallway haha.
I don't know how I can explain this. But how can i go about this? The map gen is made in JavaScript. Since the "game" is a web based game.
https://redd.it/1glow8v
@proceduralgeneration
