Coming close to an end on this tallgrass implementation! Yay!
Imho performance is starting to get really really decent considering it’s all blueprint driven. Most of the fps drop comes from overdraw from the unoptimized foggy particles and from recording. I’ll probably nativize some of the code sooner or later to further improve the performance, for now it’s good enough!
The whole foliage interaction system (swapping static foliage instances with interactive BP driven one and running the custom collision simulation on each active BP) cost at most like 0.3/0.4ms on the gamethread which I guess is good enough for a proof of concept. Now I still want to cleanup some of the bleuprints and I’ll actually be able to move on, either doing some art tests on landscapes or work on some AI or other BP related stuff… not sure yet! That being said, I’m super stoked on the result and I am super proud of myself (which is not often!) 🙂
Hey! Oh man I wish I had more time to work on personal stuff and upload this blog more often. I’m driven by an immense passion toward UE4 lately and can’t stop thinking about things to try. I’ve spent an entire WE two weeks ago toying with behavior trees and perception and I had a blast. More on this soon!
Anyway I’m still tackling this grass issue. Here is an update. I’m getting close! I’m despawning grass foliage instances and switching to a custom blueprint interactive foliage with a cyan material to see what’s going on. Still bugs and all but it shows great promises.
I’ve been amazed by this cool tall grass motion in Horizon:ZD
I’m trying to replicate this in UE. So far I made a plastic bender simulator.
On a more serious note, I feel like with the correct settings I could get close to the grass motion but what’s worrying me in UE is the fact that no matter what, the collision will affect the player and its ability to move around whereas the impact on the character physic in Horizon:ZD seems totally inexistant.
EDIT : turns out this last statement is totally false if you play with collision channel 🙂 One object may totally affect another one and not the other way around.
Hey! I’ve been diverging a lot lately about personal projects, tackling many stuff at once. For instance, those last few experiments with Sand Dunes that I posted previously remains unfinished but it’s not forgotten haha. It’s part of a bigger scheme, hopefully. It will make sense later on. Anyway, I was just super excited about my lastest accomplishement with Unreal Engine, one that still wasn’t even working an hour ago. It took me around 40 hours to figure that one out but I’m stoked on the result.
It calculates a wallride trajectory on the fly and offsets it (the offset part was extremely complicated for me, damn!!!)
Hello. Took the day off and wanted to make some progress on those dunes again. I started by taking a step back and thinking again about what I wanted to achieve here and it’s somewhere between those two references.
Then spent most of the afternoon on Substance Designer. I managed to make some improvements that I’m happy with like that “Y branching” of the main dune shape and more convincing secondary waves. I’m quite happy with you it feels in Unreal Engine so I might move on for now and switch to the next step.
I didn’t want to rely on tiling textures purely for those secondary shapes and I’m glad they read well enough with just the landscape mesh. So next step might be trying to texture that sand dune in Unreal Engine?
It’s been a while! I’ve mostly spent my free time here and there on a large-sized terrain project in UE4 that grew on me lately (having bought and played a lot with World Machine!!) I’m heavily inspired by the Wadi Rum landscape but I also want to have massives and tall sand dunes which Wadi Rum doesn’t necessarely feature. So that’s what I’m focusing on for now, nothing serious, doodling and fooling around.
I’ve stumbled upon two references that struck me lately, the first one being those new dunefield they found on Mars which are totally unique and only due to Mars’ own atmosphere. The thing here to notice is that contrary to earth where we only have two patterns in dunefields (one large scaled-pattern for the dunes themselves and one small, centimeter-scaled pattern for the sand), we have one Mars a third meter-scaled pattern that leads to a very interesting and appealing look that I’d like to get. I wouldn’t mind having to rely on a fantasy vibe to excuse the dunefields not looking earth-like hehe so, we’ll see.
Anyway, the second one is a standard close-up sand texture but having that Mars picture in mind I thought it would be great if that pattern was actually a medium to large scale pattern. But there’s a lot going on here to lead to those shapes. Those secondary and tertiary sand waves that are born from the main wave are very appealing but are very hard to recreate.
I tried a lot in World Machine but gave up for now. I switched to working on Substance Designer to create 16bits heightmaps that I could later import into World Machine and integrate in the rest of the terrain… Experiments… experiments…
Here is a try at a 2 kilometers squared dunefield in Substance Designer and imported in Unreal Engine. I like some of those shapes and the scale of it while walking around with the player feels okay but I’ve not yet able to recreate those secondary and tertiary waves branching from the main one.
And here is a try at a close up sand texture. I will probably tweak it to make it match with the final dune/sand pattern that I will be able to achieve. Looks pretty random for now.
One more image. Here I’m focusing on a single Wave. I’m trying to get a bit the feel of that Mars reference. I like how I managed to get smaller and waves in one single side and fading in long strips the other side. But those waves doesn’t read like they are born from the main dune and so it looks a bit random.
Update! Still not much too show really but I’ve been thinking a lot about UVs and tilable textures. I tweaked a lot of things so I could bake as much AO and lighting information as possible without using too many unique textures while still being able to paint details and patterns without too much stretch or repetition.
It’s a long process but in the end the texturing process will be much quicker and easier.
Here is my AO pass so far. Next step is top/down gradient and top lighting pass to further crash test my UV’s.
And, done! Hope you like the textures. I scratched my head a lot to get the right color scheme 😀
Just kidding, I’m planning ahead here before unwrapping by color coding some assets that could use the same texture. The rest will use unique texture with a lot of mirroring. At least that’s how I’m gonna start to texture it. If I have some time left once I’m done and if some props could be largely improved with more unique textures, I’ll go ahead.
Those are the only unique props used in the scene by the way. Everything else is duplicated.