Unity, Flax, UE?

Yes, there are quite a few threads regarding engines around and it's just bad timing on my part. I've come to the end of another UE project, have a bit of spare time on my hands and would actually like to continue with my own 3D project which has been stuck in limbo for many years.

My original RPG game was created in Unity of course and it would be a lot of work to port it all over to Unreal, also I use UE day in and day out so it feels too much like "work". Although the positive is I already know it pretty well. So, this isn't a "I've never used an engine before what should I choose thread"..

I could just vet out Unity for the next couple of months by myself but it would be quicker to ask about its current state, it seems static occlusion via umbra is still a thing (unfortunately) although I did write a dynamic occluder (which I doubt works anymore) plus I did hear musings of a DOTS based dynamic occluder but I'm not waiting around for features.

I have no clue about the current state of the rendering pipelines bar what I've read on this forum (and a short test in 2019 that didn't go all that well (buggy). Performance is a necessity and the new terrain system hopefully works well.

But what about AI, Navmesh, audio, UI, input etc? Are we still relying on asset store specials or self made work? Any improvements in the physics realm? Speaking of caveats HDRP doesn't seem to work with grass, which I can fix so not an issue, it's just how many others does it have? I was fine with built-in but apparently it isn't getting too much support, there's of course URP which I'd probably use / start with.

I have noticed another engine called Flax which looks promising, sort of a mid point between Unity / Unreal.. I haven't tried it but development speed seems to be in its favour, since the 5.X Unity / UE4 thread they've managed to create an entire engine plus several features some of us asked for.

From my perspective the issue with Flax is it could just end up like Xenko, there's limited documentation and I'm not the early adopter type. I'd expect it to just work.. Anyway, I can relay specifics of the project if anyone is willing to chime in, it's just this post is already long enough.

 

It's in a transitional phase (that has already been going for years) that it's not going to get out of for at least a couple of years.

Don't count on anything DOTS related being ready any time soon.

They're still not ready. There are games based on URP out, but from my experience they either perform like ass, or they are from big teams that probably customized it a lot. (I'm sure there are exceptions)

I don't know about AI.

I don't think anything has changed for Navmesh in years.
Definitely nothing has changed related to audio in years (it's still a thin wrapper around an outdated FMOD that probably nobody knows (or can be bothered to learn) how to tweak any more).
UI, UGUI still sucks as much as it always did, UI Elements/UIToolkit is not ready.
New Input System is super unintuitive and isn't faster than the input manager. It also has gaps in things it supports. Some people do like it, but those people are wrong.

Most "current" Unity features have been abandoned in favor of developing DOTS equivalents, leaving people that want to use Unity now to keep paying for nothing.

Sure, but it's DOTS related and those are not ready.

 

Hmm, that is a shame.. One of the things I miss the most about Unity is the community, it was super simple to find answers to questions or chat with a bunch of guru's. Most people on here are super friendly and very helpful..

Maybe next time then?

 

You can keep using that unity version, you can keep using internal renderer and just get things done.

Technologies like dots and are a distraction.

Same things is going to happen with Flax engine. Switching engines sounds like a major case of getting sidetracked.

 

If I could have done, I would have done.. Small projects are fine, for a larger game pretty much every native system bar physics, enlighten and animation (to an extent) would have needed some major work, replacement, bypass or overhaul. As a closed system some of it I couldn't do anything about so the project got abandoned.

In 2019 a lot of these issues weren't fixed, if nothing else I know UE can handle it. Yes it has flaws like every other engine but I'm not concerned about finishing the game. Don't get me wrong I'd like to use Unity and push forward but I'd have to A) remember what the issues were and B) find a way to fix them. I opened support cases, asked everywhere I could and got nowhere with my last attempt.

 

you are familiar with unreal and have doubts unity will fit the bill.

what is the decision point? Sounds like a no brainer to me.

 

Maybe if at some point DOTS arrives, and there is enough new and exciting stuff to try (that hopefully work?), then starting a new project in Unity would be a fun idea, even if you are proficient in Unreal, just to learn the new tech.

We are not there.

 

I kinda understand the OP , I also like to diversify the tech I work with not to get burned out. After work I like to use other technologies I will never use at work, just as a runaway from work stress and for fun .

As for Unity I kinda don't know atm which path they want to choose and where the engine is heading atm, I am kinda on a fence too which engine to use for my project. I like to work with Unity, I like its simplicity and easiness of use, also C# . On top of that UE has no Microsplat ) and other assets I like to use in Unity.

On the other hand UE feels more like a true game engine, a full package with all the required tools just in place and its direction is more clear to me and I am kinda calm when it comes to what will happen with the engine in next 2-3 years.

As for Flax, its fresh, new kid in a class. For now I just observe it and want to see where the engine is in 1 year or two before I make any decisions if I want to actually use the engine.

 

It just takes a long time to get things done in UE, Enlighten is a massive draw as well (even though Unity are wanting to deprecate it). Personally I think Unity looks great, you've got C# and quicker iteration testing times.. My work projects are stuck on 4.18 (UE) and hot reload is a bit hit or miss TBH.. Also I'm planning to drop Modo / Maya altogether and UE has a very tetchy import system (even with Modo). So moving to Unity would save me quite a few $$$'s.. Plus of course the community.

Unity certainly does have its allure and I could start scaling a lot of it back, although there would be some reliance on asset store stuff as I don't have much free time at the moment. In short there's benefits to moving from UE (for this project) and I know Unity holistically well enough.. Plus as stated, I use UE for work and it kinda ruins things from a hobby perspective.

Just mulling it over really.

 

You spent multiple years into given point.
Don't know if the point is even suitable for alpha, as minimum viable project.

Porting it to different engine will cost you at least year. You would need rewrite everything, as you can not copy things over just like that. Other than art assets. And maybe some shaders, with tools.

And then spending years to come, just to get move on.

If you haven't touched DOTS don't. I will cost you a year, just to get basics in igven peace. But at best, you could consider jobs. Still, that means need for porting project to at least 2019, preferably 2020.

Do you even need it in RPG games?
If game is for pc, stick to version of unity that you are comfortable with.

You don't want o go int HDRP route. You dont have man power, to utilise it correctly. Focus on specific art style. Otherwise you will hinder years just on HDRP. I can not imagine you need it for anything, in type of game you making.
In few years time graphics will be dated. So better focus on styled art.

Shrink your projeject's goal. Or you may never reach release point, trying add more and more stuff.

 

Agreed for the most part but I'm confused by this statement, if I can wrestle UE with any graphical gubbins it offers why would HDRP be an issue? Unless it requires extensive writing or something then? That would make sense.

 

No offense, but this sounds like feature creep, getting distracted or aiming too high.

In general "pretty much every native system" shouldn't need some major work/replacement/bypass/overhaul, if it is a solo project. You saying that might indicate that something funny is going on.

If at this point you try switching to another engine, you'll most likely just waste time to get back to the point where you are right now. And in case of Flax, it is likely that you might need to ditch it, as this one isn't as battle-tested as Unity/Unreal.

Pillars of Eternity, Wasteland, Shadowrun - were all done in Unity. What does your game require that those games didn't have?

 

I personally haven't had any issues with HDRP. But at the same time, I don't really have much of a choice in using it either as I want to utilize the Ray Tracing features.

But then again, i've been off and on playing around with HDRP over the years so it's not like I just jumped right into it over night and had no issues at all.

HDRP and built-in aren't a 1:1 ratio copy with extra features, there are differences, but the differences can be learned with lots of research, (which as I mentioned off and on over the years played around with it). During the early days HDRP and even URP/LWRP were constantly in a state of flux, things are still changing of course, but it all seems to have at least leveled out enough to where it's usable.

But I really do like HDRP, it doesn't magically make things 'look' better, I still say it's the artist not the engine that makes things look great, if you have a crap material, HDRP or even UE5 isn't going to make it look any better than the crap material lol.

Where HDRP comes in I believe aside from Ray Tracing, is simply the workflow differences, once you get used to them I think they are easier personally. But that's me.

Again, this is just my lonely ol opinion and years of messing with HDRP.

 

Yeah, Pillars of Eternity is tiny compared to what I was trying to pull off. But, it's a pet project, I'm in no rush and I've worked on MMO's, FPS's, RPG's etc. in Unreal without much issue. You might remember me I went under the alias ShadowK at one point? If I can't do my stupidly unreasonable side projects I just wouldn't bother.

You're going to have to forgive me here because it has been so long, so Unity 4.X just crashed because it couldn't handle the size of the game (32-bit). Unity 5.X had issues with navmesh overhead, async tile loading speed, asset bundle loading times, physics stutters (and this wasn't a distance or CPU load thing), extremely long loading times, Speed Tree didn't work right, Frustrum culling etc.

Besides a few bugs like the terrain system not working after an upgrade and some epically random crashes (which were fixed AFAIK) you may have noticed 99% of this was all performance related. In terms of "over reaching" I had a lot of the gameplay / art completed / systems etc. finished, enough for a decent demo anyway. The AI system never worked but I always rolled my own anyway and Umbra was useless for my project.

Frustrum culling was quite annoying, for some reason even when a tile had been unloaded it was still trying to cull it (this happened in Unity's terrain system when I tried 2019 LTS too if I remember correctly)? The overhead from FC was rather high as well.

Also just some rather weird issues like https://forum.unity.com/threads/gfx-waitforpresentongfxthread.731327/ (yes I turned G-sync off and it didn't fix anything) it was there in the build as well. HDRP doesn't do this? There was metric ton of odd issues when it came to rendering so on reflection I think HDRP or URP is a must have.

I managed to sort out quite a few of these issues but it just never really ran all that well, in UE I have to watch physics, lighting, DR translucency issues and particle count. After that it works as expected..

Anyway, the plan was to actually split all of this up into an attack on titan sort of walled world where I'd load each gateway into specific zones and take the heat off most of this. I'd raged quit by then ..

 

I'm doing a title that runs on everything so URP + 2021.2 is just close to the perfect choice for that game. Unreal 5 would be omitting a lot of platforms.

 

That's a good way to never finish the project, but...

You are likely already know that.
(Yes, I remember you)

 

I've had issues but they weren't the fault of the pipeline as much as they were the fault of one of the teams behind the platform implementations. HDRP is basically flawless on standalone platforms.

 

Ah yeah I haven't done anything outside of Windows, so that's true I cannot answer on anything outside of that area lol.
But I hope you got it sorted out

 

I've been reading through a lot of the HDRP forum stuff.. This is the general impression I'm getting which isn't all that different to general Unity dev.

- I want to use SpeedTree, well HDRP doesn't support that.. BUT there's an app for that ---> forward to asset store.

- I want to place grass and have it instanced, well you can't but you're lucky there's an app for that. ---> forward to asset store.

- Okay, Okay, I'll use URP and figure it out.. Where's my decals? Wellll you can't do that but.. Yep, assseeett storrree..

Then you have Unreal which is like this..

- The workflow is too slow I want to try something else.. You say that but have you seen our free megascans library?

- C'mon the import stuff is clunky.. You say that but have you seen our free Blender import toolset?

- I can't understand half of the C++ API because you've just done a bot spam.. You say that buuttt have you seen our practically made from the ground up demo's working at 60+ FPS on mediocre HW using nothing but BP's to create the entire thing?

You look at either sort of puzzled as to what has just happened LOL..

 

Well there's your problem. SpeedTree has never worked well with Unity.

How so?

 

AFAIK you can only really use SpeedTree with LY and UE4.

But that wasn't quite my point, in Unity 3.X / 4.X after some tinkering about I knew exactly what it was. It's not an engine but a cross platform framework which you bolt your own tools upon (let's not get started on semantics), in the days where UE cost $250,000 and XNA or rolling your own engine were the only options (or something like Ogre yeesh) it was simply amazing. I'd have used it just for the simple to use C# integration..

So when it came time to develop the two options were either asset store or build your own systems, AI, Navmesh, rendering additions (like deferred decals or dynamic occlusion), upgraded shaders, physx extensions etc. which was fine, Unity gave you a starting point you added the actual functionality.. Issue is I hit the 32-bit limit, my stuff stagnated, had a go in 5.X, hit a lot of issues then as of 2019 Unity had changed so much that a lot of what I'd built was redundant.

Okay, now if I spin up this project again it's once again asset store or roll my own. The goal posts may have changed but it seems to me the ethos remains the same. Now I've come to that conclusion it would probably be far easier to port the entire project. I mean feel free to correct me here but this thread has lead me to believe not all that much has changed since those days.

I mean it's fine if you're part of the active eco-system but from a now outsider looking in it just seems like project breaking amounts of work, I don't rely on the asset store to fix all of my qualms because in some instances it does the opposite.

P.S project specific of course, just in case anyone reads this out of context.

Usually compilation times for various things (lightmass, shaders, code (if HR doesn't work) and digging around in source or scrambling for useable docs / info. Cooking is always interesting..

 

https://portal.productboard.com/uni...l-effects/c/262-speedtree-shader-improvements
Speedtree improvements have landed but in 2021.2 beta version

 

In fairness, speedtree improvements are always landing or are in the process of landing and it never ends up working properly for more than a moment.

 

Just when I thought an Eagle had swooped in to save the day ..

 

My daughter loves using unreal, but sadly it is way over my head. There's a rather useful c++ course you can look at on udemy. Until they really start to work on Verse, it is always going to be a barrier to entry. What do you feel is best suited to your projects?

 

Speedtree shader is integrated into shader graph as per the productboard so I think basic wind and SSS effect might already be included in the graph and it might be editable (I haven't checked it out yet and there is no documentation available on it)

 

what issues do you guys have with speedtree? I've used it a ton in unity and never had an issue.

 

It's had a rough time on URP and HDRP, but 2021.2 has an official speed tree shader with wind support, instead of a magenta polygon shader or a static badly lit pbr shader.

 

oh i see. I never messed with the new render pipelines because i didn't think they were production ready.

 

Yep I think Unity's striving for maturing those now, along with new features coming to Builtin, which includes lensflares, shadergraph and more.

 

i hope so. i got a prototype game in unity still and would be cool to let it remain in unity once i get to full production with it.

 

I would guess that part of this issue author is having is something we all have to some extent. And that's the 'crazy squirrel, anything that is new and shiny I must have' mentality.

I'm sure we can all get lost in the new render pipelines and toolsets, vertex painting, nanite, lumen, and once you do venture down that rabbit hole it becomes more and more fuzzy the deeper you get.

Keep asking yourself, 'Do I personally have the resources to complete a project of this scale with this level of fidelity, and more importantly is it fun to play?'

If you can't answer that question honestly, go back to the drawing board and start reducing.

One of our mantas, is to relentlessly follow the K.I.S.S principle.

One such video which I highly recommend watching and a book to read is 'Make Time.' If you follow these principles you won't go too far wrong.

At every moment in the day you might have to respond to an 'infinity pool' of emails and distractions. Start your day early, and have one, and only one 'highlight' and be focused on this only. I guess this is what we're trying to do as we're beginning this game creation journey. Start small and stay small, don't get distracted.

We never had such material in my time, maybe it wasn't needed so much.

A link to the video is below:

Take care and have fun

 

We did do a brief test and pretty much everything comes undone once you venture into the newest pipelines. For example, depending on which pipeline and version you are using, you will require a specific update tool/package. HDRP 7.12 or HDRP 7.2.1 or URP 7.12 etc

Terrain painting has issues compared to unreal, you will need a specific water shader and or fire shaders, popular sky plugins, such as enviro sky do not appear to work, vertex painting seems slow and convoluted, does it use polybrush? In short, it is just painful, in the future this may indeed change.

From what I can gather given the credentials of your last game, if you were to go with unreal, you might be interested in following a workflow very similar to the video below, you may have already seen it

Have fun.

 

That's where the game is right now. I have been rebuilding it in unreal.

 

The question to answer is, is my current engine helping me complete my game, or am I still using it just because change is hard?

 

And I suppose if we were to unpack that a little further.

"The question to answer is, is my current engine helping me complete my game, or I am still using it just because change is hard and maybe it is convenient to have something to blame when things don't quite work out."

As with anything in life I tend to follow 6 simple rules. Now that I'm older these have become more meaningful than before. We're all spinning around on this rock zipping through the milky way. Let's enjoy the ride.

1. First of all, don’t try too hard.
2. You show up.
3. Trust what you don’t know.
4. Experiment.
5. Have confidence in yourself.
6. Have fun and be kind to others

 

Nope, the issue with me is I've been doing this a long time and I know too much, sometimes ignorance is bliss. I already understand the challenges I'll face even if I'm not au fait with the latest tech hence me asking. Investing in a platform takes years but at this point I am engine agnostic, from a game work methodology they're all the same to me.

It's not game content, scripting and all that basic jazz that concerns me. I'm fine with UE but for a lot of people it's where projects go to die, due to their extremely rapid iteration they're always breaking / fixing / updating a lot of things and mid way through a project you're probably going to encounter one of these crashy / inconsistent delights.

You might be lucky but unless you have the skill to merge fixes, modify the source in their pretty much undocumented setup then forget about it. An LTS version and better C++ documentation would go a long way towards sorting that out but I doubt by this point they'll do it.

Unity is just barebones in every way, so forget about game length / size / scope the question I ask is how much time will it take me to complete the engine? The "new and shiny" mentality in Unity is what I was using nearly a decade ago in other tool sets. Forget graphics, HDRP is all about getting around their bizarre performance issues.

Also it comes down to expectations, take POE for example there's performance issues, navmesh issues, general bugs etc. and I wouldn't stand for that in my game. The major issue with Unity is can I fix it myself? The answer is usually yes but it will require a different system because of course Unity is closed off. If you find a game breaking engine bug that Unity doesn't want to fix however that IS the end of your project, period. There's no learning to fix it or way's around it, you log a ticket and hope for the best.

There's no such thing as a "perfect" engine, you choose one and go all in.. The ultimate $1MIL questions are how much of a headache will it cause me? And how long will it take me to sort it out? That I'm still looking in to..

 

I'm sorry I don't see the convenience in that, but then again I never understood the shifting blame gymnastics, but you do you.

 

If your project is a labor of love I do suppose not much else matters, I would just sit back and enjoy the learning experience and enjoy the ride. You sound like an intelligent chap, all roads end to the same place right

 

Pretty much all of that list, Unreal does better because it's made to solve every problem you've listed. There's so many problems with building your own tech stack and relying on hobbyists from asset store. You should use Unity if you plan to target mobile, switch, vr etc IMO. Smaller focused titles.

 

I wanted to use Unity for Enlighten essentially and was willing to put up with some pain to do so, C# was just the frosting on the cake.. Guess what? Enlighten is getting the heave ho so, I've got nothing here and I'm not going to wait for Unity's own solution.. There's no practical reason why I'd choose Unity at this point.

Appreciate everyone taking the time, I'll just port it over to UE.

 

FYI: Enlighten just made a (probably temporary, but will see in the long run) come back.

 

Pretty sure it's temporary, I saw an expiry date on one of the threads.

I'm not adding it back to my game because easy come, easy go in Unity land. I've no idea if they'd renew it when it expires again. I'm going to roll my own, which does sound familiar but hey ho.

 

2024 is far away. A lot of things can happen until then. We could release our current work until then among other things.

 

Yeah just once burned twice shy. I already had to remove it from my game. These days I make decisions based on what I know today. What I know today is it'll be removed in under 3 years, in 2024 because that's when the license expires, and that's what Unity did last time.

Unity hasn't said anything else so of the two of us, who is acting on blind faith, honestly?

 

One day.. *Shakes fist at the sky..

 

That's what I said. Until then our current project probably will finish. At least I hope mine will. This is why I am not afraid of this. Obviously if you anticipate that your project won't finish and you will need to update beyond that, you are better off with something else.

 

I anticipate it will finish and I will want to do less work on keeping it alive. LTS is 2 years, and if I wanted to port it to shiny new platform X I would be screwed 100%

For indies, porting an older title is a big win. Heck, it's good cash for any dev.

 

Yep they might get HDRP sorted out by then and I'd just stick to a version w/ Enlighten indefinitely. There's also one thing to consider, for music essentially I moved to Mac but UE doesn't really have good support for it. EGPU's are now a thing so, pro Unity.. I do have a Windows machine laying about, just waiting for it to die essentially..

The upcoming M1X is making it easier to have a potential all in one going by the M1's graphics performance on a 10W chip of all things. If I could have a mobile workstation that would be awesome..

 

In my experience, stuff from the Asset Store is usually great in the short term and not particularly great in the long term. Wrangling a bunch of 3rd party stuff with variable levels of quality and support commitments can be rough. I had great experiences with a handful of Asset Store developers, but knowing which will be which ahead of time is guesswork.*

Seriously, though, based on what I saw back with your ShadowK account I suspect that you don't heed URP or HDRP. Make your own "ZarconisRP", have fun doing it, and make something specific to your game rather than some generic all-things-for-all-needs thing as Unity has to do. Considering that Unity's stuff isn't production stable yet** you've got the option of periodically updating your stuff as it changes, or making your own stuff and controlling the changes in the first place. Since you seem to crave huge projects and highly technical work the latter seems to be right up your alley.

The short term cost will stink, but in the long term you'll reap thee benefits of designing your RP to suit your intended content and workflows, instead of the other way around.


* This isn't because Asset Store developers suck, or anything like that. It's a part of the territory in a high risk business where prices are low and margins slim. If they don't make enough sales to be comfortably profitable they need to do other stuff to pay the bills.

** So I hear. I haven't used it. Our team experimented with HDRP and URP, and then decided to skip straight to making our own.