7 days to die is a long running alpha stage game starting out with devs who had immense passion and ended up at this point of just being stale. They continue to push updates that don’t really address performance bottlenecks that honestly has finally left most the community in a mindset such that we are wanting a remake or spiritual successor of this game at least mechanically, whilst reaching the performance and AI aspirations the devs had long ago but at this point are just unable to fulfill. It’s worth looking up videos of this game and doing some research to gather a solid understanding of what it’s trying to achieve but i’ll go ahead and express the rough design below.
*Destructible voxel based terrain, using a primitive form of marching cube voxels that are more diamond than cubes honestly. Which includes “structural integrity” (should watch a video on the mechanics it’s long winded) which is a complicated but somewhat realistic system enabling players to get punished for building structurally unsound mines/buildings.
*Monsters destroy anything and everything in their path to reach the player to attack them. Potentially collapsing structures along the way.
*Full building system tied to the voxel backend and what used to be entertaining to blow apart with explosives.
*Basic crafting system akin to rust.
*Random gen worlds.
*Honestly very decent mod support.
*Multiplayer support. Here’s the tricky thing. The game’s built originally as a singleplayer game and the devs have no experience making scalable networking backends. And thus. We have immense dysinc, client side determinism, and runtime/bandwidth overhead issues. Servers literally can’t host more than 38 ish players i believe without immense dips.
It’s weird the game used to run tolerably on an apu quad core with no dedicated GPU. But over time the rickety foundation shows as new things continue to break and slow down.
I’d like to spearhead a community opensource project that’s a spiritual successor to this game and would like to hear opinions on unreal’s scalability and if it would perhaps match the project requirements. I’d like to support at least 100 players in any given server with 50 to 100 individual zeds/monsters per player. Supporting even more players or zeds with proper multithreading and something DOD akin to ECS would be the dream. Unsure yet as to if we want low poly meshes or high poly. I’ve tried shiving a custom ECS implementation in UE4 in the past and while it works it’s slightly a headache and doesn’t get around the Actor pattern that seems to cause fair overhead overall.
Would UE5 be ideal? If not what about unigine since that seems to have a low level flexible API that works wonders with a custom ECS lib, could do this in unity with their DOTS framework that 7d2d doesn’t seem to be using sadly. Only other option i can think of is making a custom engine which while i’m experienced in doing this i’m not experienced with 3D gfx programming yet. Networking i’m expecting to find someone or some material that’ll help get things scalable and that’s one of the first things to get done. In other words there’s a lot of backend work that needs to be done and am wondering the exact tradeoffs with these different approaches if anyone here’s done decently runtime heavy projects as of yet.
I see no reason why it would not work, but personally im jumping ship to godot 4.0 beta ASAP, I believe that might be a good place to start, but UE5 still would work excellently. Its tried and true for high player counts as it is literally just UE4 except with a slightly different file format, lumen/nanite, and a fancy UI, and UE4 has been used in many full size MMOs. Note that lumen will not work for you as custom meshes do not generate distance fields, and nanite will be limited in use. Gonna have to use raytracing for that one or find a workaround.
Both Godot and UE5 seem like good engines, but Godot 4.0 is still pretty primitive in comparison, and 3.3 looks more like a demo of an engine rather than a full engine, a leap im willing to take to be able to not have the engine’s postprocess and lighting system work against me, not with. Its all about what tradeoffs your willing to make, UE5 seems perfect for your uses, but is severely conflicting against mine.
Also, if you make that, im ABSOLUTELY going to buy your game (Preferably on steam). I love the idea of 7d2d but its so haphazardly put together it brings my 3900x up to 40% usage and maxes out my rtx2060, usually struggling to hit 60fps, not to mention the lack of content in 7d2d.
I actually want to make this fully opensource, including any asset data. So essentially an attempt to see if public funding and passion are sufficient means to get something like this going without charging the end user to play. Ideally all funding goes straight toward the project and i wish to have an open book keeping policy such that everyone can witness the expenses and donations.
If you want your game to run around .009 FPs ue5 is a wonderful idea.
For any real project, any real engine would probably be your best bet…
Meaning CryEngine, Godot, maybe even Unity considering that they actually have some decent people who listen to your bug reports, unlike Epic.
Voxel terrain done well will likely require customization of the engine.
In fact, it’s rare to see it done well.
No man’s sky has an interesting example, but it also tends to go away between loading.
To be able to preserve it you’d have to be able to load it somewhere, and in that particular game it could mean an almost infinite stream of data…
For quite some time now I have seen you commenting on the poor performance and lack of interest in Epic Games resolving issues and although I have persevered I am starting to become increasingly frustrated. I recently started a small short film project and after all the hype about the new Path Tracer in UE4.27 and UE5 found that it SEVERELY lacks features - doesn’t even render hair or eyes of Metahumans isn’t this the exact target for Metahumans LOL?
I think that short film project aren’t what you are talking about when you say to jump from Unreal but it does provide an ongoing example of my frustration that is leveling up to yours LOL.
Outside of the short film we are also working on another game (SciFi thriller) and am now very reluctant to bother with unreal (either 4.27 or UE5).
I see you mention CryEngine a lot but my concern is that it is a dying engine, there are almost no positive examples of it on YouTube and most of the examples I find have very dated looking graphics.
I would love to jump ship but I think the graphic fidelity in CryEngine is not going to be where we want it and I have real concerns we will be developing for the next 12 months in an engine that is not going anywhere.
Love to get your opionion on that before we consider any further
Thanks for your stark honesty on these forums LOL, I am starting to see the light.
Unity is your best bet, unless you already have extensive experience in unreal or you like the C++ side of unreal.
Godot is a no go. version 4 included. Not even close.
Godot 4 is taking much longer than it should, by the time it’s near release, unity matured new render pipelines (URP & HDRP) and developed them from the ground up to something impressive in 2021.
Unreal engine 5 is being released.
And we’re still waiting for godot 4. Which doesn’t bring anything really worth mentioning, maybe the one exception being SDFGI.
I would say it’s moving at snail-pace, but it’s slower than that…
And from all the talk you hear from godot users, you’d think it has a respectable number of games released on steam. Or releases per day. Or popular games.
To top it off, godot 3D is still a meme.
Godot 4 launch will be bug prone, as the lead developer tweeted, because of ‘extensive changes’ to the engine.
If you want good graphics, use Unreal or Unity HDRP 2021.2
If you want simplicity, performance, or a foundation to build on, use unity URP.
If you want to post about a game on reddit but never release anything, use Godot.
I’ve tried cryEngine for a while out of curiosity, nothing serious. Lovely graphics.
Seriously, the terrain, vegetation and lighting of ‘Kingdom come deliverance’ is the most impressive visuals and scale I have ever seen. Tops RDR2 and everything else for me.
But the user interface was… tough.
The documentation was also tough.
But the worst part is the fact that there’s no community, little to no videos. Ghost town.
I seriously loved the lighting and the amazing GI, shadows and terrain/vegetation.
But in the end I don’t see myself using it.
If someone likes depending and figuring things out by themselves with little outside help, then cryEngine could be an interesting choice.
One of my biggest gripes with unreal is the scale look it has. Especially when it comes to the terrain.
Single handedly makes me stay away from it. I have no idea what’s the reason behind it, but it’s there, and it’s as noticeable as can be. Especially when a game isn’t made by a big studio.
Sometimes a big enough studio gets rid of that look, somewhat. Indies with UE are the opposite.
I know this is quite different for each person, some people like that look.
CryEngine on the other hand is much better when it comes to that, and visuals in general.
Glad you understand, but it’s just an infinite slur of things that piled up over several projects.
The reason I use an engine is that I suck at the rendering pipeline, instead, I’m almost always stuck having to fix the rendering pipeline for it to do what I need, which like in your example should just be native.
CryEngine is a lot of hands on.
Not dying, but also taxed by covid like anything else. We are all tired of the current situation of things. Even if some of us are prospering (I seriously have more work than I can handle).
IT (Cryengine) can work great.
Crysis is still used as a benchmark after-all.
The problem is that most of those transitions and LOD baking have to happen on your own.
That’s why games like Kingdom Come Deliverance (made by people who DGAF) have LODs that pop like popcorn mid cooking.
That said, I picked CryEngine because of my historical reconstruction requirements fitting in to the base features I found when testing. Not because the engine is great, or anything else. It just looks way better than unreal when properly built, the PBR pipeline does decent metals. Character skin is a little plastic like maybe toonish. But it’s still better overall for my needs.
IF you project is serious, don’t look at the most recent releases, look at stuff pre-covid. UE4 .18 perhaps if you need to leverage landscapes (that should be before they made the change to the LOD distance). Same for Unity or just about any other engine.
Test/Compare to either the latest or the one prior.
Don’t discount Unity either. Yea the rendering pipeline is also a mess, but at least posting in their forums developers may actually answer to your problem. As opposed to here, where developers come on to deliver the cookie cutter response that the eggheads managing stuff told them to deliver / trolling all user feedback - example: https://forums.unrealengine.com/t/epic-looking-for-your-feedback-on-the-direction-to-take-with-answerhub/
I use unity, HDRP specifically and it’s been great so far. Aiming to upgrade to 2021.2 and stick with 2021 LTS.
HDRP is in a much better state now compared to the awful 2019, and to a lesser extent, 2020.
There’s still some particularly problematic issues, which I’ve been vocal about and will hopefully be sorted.
but overall, it’s great and I don’t see myself using anything else.
Developers are also very active on the forums, and sometimes the discord server.
@Trent_4 I had actually jumped from Unity HDRP back to Unreal late last year as a customer wanted a UE project done.
My only problem with Unity now and it seems nitpicking but important for the visualisation stuff we are doing is there is no strand based hair… all our HQ characters now have strand based hair and without it now looks dated. We do our own characters with Alembic curves for hair in Blender and they work fine in UE5 - no such feature in Unity or I would probably have already jumped back
Can’t you try to bring that into unity yourself?
Like using Nvidia hair, for instance…
Ps: it’s not like unreal just works. If you add multiple grooms you are still likely to crash the project/game what have you.
See 4 posts below with the groom not rendering on Mac XD
@MostHost_LA , hmmm I hadn’t even looked into Nvidia hair and Unity.
We aren’t concerned about Mac bugs, what we generally develop are specific to a client and we set the requirements with them for the project generally (i.e. requires Windows 10… Graphics Card of a certain power level… and so on)…
The Nvidia hair is an interesting thought though, thanks for the heads up
Unreal Engine 4.2# or 5.0 would be excellent for 7 days to die Survival game.
I would recommend you be a little familiar with either C++ or fluent in the Blueprinting system in Unreal.
Everything your need is built into the engine, and there are tons of free assets on the epic games store and free characters on Mixamo to help you your project started. I would recommend however before you attempt any release, you build or have someone build your own models.
considering Lumin and Nanite being in the Unreal Engine 5 release, there is no better time to start. and learn how these engines plugins will work best for you. Nanite, guaranteed will cut down on your lag on the client side.
isn’t this the exact target for Metahumans LOL?
