New video:
Supercharging my procedural mountain forest
Two years of landscape progress in one devlog!
https://www.youtube.com/watch?v=ACjlgxRp69c
Remote
Rune Skovbo Johansen
@runevision@mastodon.gamedev.place
mastodon 4.7.3Indie game developer, procedural generation enthusiast, Dane in Finland. I made Eye of the Temple, now working on The Big Forest.
2524 Followers
115 Following
50 Posts
Joined November 19, 2017
Website & blog:
YouTube:
Discord server:
I wondered how far down the horizon is in our field of view, and did the math.
At two meters above sea level, the horizon is 5km away and 0.05 degrees below horizontal in your field of view.
At 100 meters up it's 36km away and 0.3 degrees down. At 1km up it's 113km away and a full degree down.
For reference, the Moon (diameter) is about half a degree in the sky.
https://www.desmos.com/calculator/0l0mtwy0cb
In video game development, we're used to sound coming from single positions in space, but how do we handle sound coming from an irregular shape?
In this article (with plenty code snippets), I go into details with the Point Cloud Sound technique I'm using for water streams and foliage:
https://blog.runevision.com/2026/06/point-cloud-sound-for-irregular-shaped.html
#GameDev
Oh look, 80.lv has an article about erosion :)
https://80.lv/articles/fast-terrain-erosion-filter-that-emulates-erosion-without-simulation
Replying to
Here's the other Shadertoy, that shows off various animated parameter changes in erosion strength, scale, detail, and ridge and crease rounding.
https://www.shadertoy.com/view/wXcfWn
(You can temporarily control the camera angle while dragging with the mouse.)
#ProcGen #vfx #GameDev
Replying to
I also tweaked colors a lot more since my last post. Especially making the grass notable darker and slightly more yellow, which makes the terrain feel less off. But also tweaks to tree colors (tiny bit less blue), fog color (less saturated), skybox (less bright), sunlight (more intense), color grading (less toe and shoulder), and added vignette. Attached before/after images.
Replying to
Okay, after a lot of experimentation, I got a result I'm fine with, as a result of:
- Darkening grass a bit (and cliffs and paths to match) so it's darker relative to the trees
- Darkening the auto exposure so it doesn't make things (especially grass) quite as washed out
- Brightening the sky considerably
I had thought I needed to make the grass less yellow, but it actually needed to be darker. Before/after images attached.
I tried adding a discard to the terrain shader based on distance to the camera, to achieve a clipping sphere around the camera instead of a clipping plane. The point being to avoid the mountain silhouettes wobbling when merely rotating the camera.
It works fine in the editor in Play Mode, but fails making a build, with an error of
"internal error: argument pulled into unrelated predicate"
No idea what to make of that.
Replying to
I had actually settled on an even paler fog color that more closely matched my references and did look quite realistic. But then when I played the game and vistas came into view, some of the excitement had gone; the mountains didn't feel as alluring.
So I backpedaled half-way to the more saturated fog. (Fog in this context is used as cheap aerial perspective / atmospheric scattering, not to simulate actual fog.)
Attached is comparison with some reference images I took in Japan.
Replying to
Hmm, it turns out incremental garbage collection was introduced in Unity many years ago but it wasn't enabled in my project. This helps a lot reducing the size of the spikes from garbage collection. Not enough to entirely avoid dipping below 60 fps, but it's a lot less severe than the 100+ ms spikes before.
Replying to
But damn, I don't know how to escape a never ending process of trying to tweak different elements more or less blindly. Even with references, it's super hard to figure out what elements (or relation of elements relative to each other), are vital and which are merely circumstantial. I feel like I'm missing some fundamentals in process.
Replying to
I actually don't use the mask based erosion technique from my erosion filter video and blog post here. I don't need crisp mountain peaks and valleys for my usage, so I mostly apply Phacelle noise (with derivatives) directly (4 octaves), with custom modulation based on slope, which works much better here. I think the erosion techniques work better as a toolbox than as a monolithic grand function.
Replying to
@yoh@hachyderm.io The towers are points of interest, and the paths are generated to connect those. The paths are found via path-finding. I have an old article about it here:
https://blog.runevision.com/2016/03/note-on-creating-natural-paths-in.html
Replying to
@aras@mastodon.gamedev.place For one because I'm extensively starting threads from other threads while Unity's can only start threads from the main thread. Having to wait for the main thread queue for each new chunk task would be a significant overhead in both performance and code complexity.
For another because in my experience, Unity's job system is not geared towards long-running background tasks AT ALL. The manual directly says "Avoid using long running jobs".
Replying to
@conchoid@mastodon.gamedev.place That's fair you perceive it that way. To me it's almost the opposite. The extra detail on the cliff surfaces brings them more in line with the detail level of the ground with trees, making it feel more cohesive to me.
Replying to
@benthroop@mastodon.gamedev.place Add to that that everything is generated on the fly, which has save-state complexities whenever a player leaves an area so that it ungenerates. There's challenges related to e.g. bringing an item from one area (which later ungenerates) to another, and so on. All this complexity is compounded with multiplayer to a level where I'm not sure I can even keep it straight in my head.
Replying to
@benthroop@mastodon.gamedev.place I'm genuinely not quite sure what this comment was meant to convey, but I can elaborate a bit.
If this was just a walking simulator with no gameplay, multiplayer might be doable for me. But it's going to be a puzzle game. Not that many puzzle games have multiplayer to begin with since it has non-trivial implications for puzzle design.
Replying to
@crystal717@mastodon.social Glad you like it! This is in Unity.
Replying to
@curiousdynamics@mastodon.gamedev.place Well I still don’t know when it went out of experimental, could be in 2021 for all I know.
Replying to
@curiousdynamics@mastodon.gamedev.place Where do you see this info? I can only find references to it being released as experimental on 2019.1.
Replying to
@toerror@mastodon.gamedev.place Yeah hiding the tiling and small scale grain was a big part of my motivation to getting more detail out of the erosion. I haven’t yet decided what to do about cliff textures at a distance; whether to fade some larger textures in or something. But it’s at least a smaller problem now than before.
Replying to
@vuntra_city@mastodon.gamedev.place Ah, I see! That's great :)
Replying to
@andoni@meetiko.org @andrewblasco@masto.es Right here is where I share progress most consistently. I don’t use a special hashtag but I’d say the majority of my posts is related to it in one way or another. I often use procgen or gamedev hashtags though.
Replying to
@efi@chitter.xyz @kizu@front-end.social Yeah. In the end, I'm not very interested in simulating real world forests; my game is fantasy and doesn't adhere to "the real world". My aim is to make the landscapes evocative. Of course, our experiences from the real world informs what we find evocative, but the details of that are very subjective. What one persons thinks looks super inviting, another person may think looks fake. So I'm mainly guided by my own sensibilities, but welcome comments and input too.
Replying to
@dougbinks@mastodon.gamedev.place @efi@chitter.xyz @kizu@front-end.social Nice! Yeah I’ve been to Switzerland a number of times, it may have helped inform my sensibilities. :)
Replying to
@aeva@mastodon.gamedev.place @efi@chitter.xyz Yeah in the forests in the video there’s only a few scattered clumps of grass here and there (which checks out with some of the IRL forests I’ve been in). The green stuff in the ground texture inside the forests is meant to be moss.
Replying to
@Tama23@cathode.church I’m happy you’re getting such a reaction, that’s what I’m striving for. :)
Replying to
@benthroop@mastodon.gamedev.place Aaahh, but all the complexity! This project has enough challenges as it is. 😅
Replying to
@BuschnicK@mastodon.gamedev.place Thanks! I never go without my speed boots.
Replying to
@efi@chitter.xyz If you mean in the eventual game, then normally you won't be able to fly, but there may be certain circumstances where you can.
Replying to
@nick It is, I mean procedural level design. :)
Replying to
@jonikorpi Thanks a lot, glad you liked it! It’s one of those things where if I had known up front how much work it would be, I wouldn’t have done it. But now it’s finally done and out, and I’m happy it’s so well received.
Replying to
@benthroop@mastodon.gamedev.place Ah, right, thanks for the clarification :)
And yeah it’s not easy to know my game will have puzzle-like elements when I’ve worked on this forest terrain for so long without introducing them yet. But the gameplay graph I’ve worked on in a separate project will eventually be integrated into the forest terrain.
https://runevision.com/multimedia/thebigforest/
In fact I’ve just recently merged the two Unity projects as a step towards this.
Replying to
@jonikorpi@mastodon.gamedev.place Thanks! Yeah I'm familiar with the GMTK analysis. As for the former, do you know if there's a transcript? I can't multitask, and given that, podcasts are usually too time demanding for me.
Replying to
@nick Probably, in some form! The terrain in The Big Forest will be generated primarily based on level design, but I want to try to apply erosion in a subtle way.
Replying to
@portaloffreedom@social.linux.pizza I don't know? Can I? How?
Replying to
@benthroop@mastodon.gamedev.place In what way helpful? I’m not sure I follow :)
Replying to
@roctordush@fe.disroot.org Thanks! Wouldn’t rocks fall down if they’re on the steep cliff faces? Or you mean embedded into the ground? Do you have a reference image that shows what you have in mind?
Replying to
@YinYinFalcon@mastodon.online Yeah, I agree it makes the air feel clearer! Probably especially due to the fog color combined with a bit stronger sunshine.
Replying to
@tinylittleenormous@piaille.fr The towers are points of interest, and the paths are generated to connect those. The paths are found via path-finding that penalize steepness. I have an old article about it here:
https://blog.runevision.com/2016/03/note-on-creating-natural-paths-in.html
Replying to
@johnnesky Yeah plenty of follow up research one could do! It'll have to be someone else than me though :)
Replying to
@efi@chitter.xyz If you have any reference pictures of what you mean, I’d be happy to put them in my inspiration folder :)