Unreal Engine 5 = Game Changer

I have been bouncing back and forth betwen unity and unreal for awhile now. Just recently before all of this I decided to build the minimial viable game in both engines. I have used unity more but I also have spent a ton of time learning unreal before this experiment.

My experience for this game is:

Unreal: Long dev times for each feature ending with 30+ browser tabs open trying to find info on their api and best places to put code etc. Horrible compile errors. Positive is built in basic authoritative multiplayer that I need to for simple 1 v 1 multiplayer.

Unity: Very fast dev times per feature. So fast that I my unity game is way ahead of unreal version with way less time spent. Negative is I have to learn third party package for multiplayer.

So even with this news and pretty demo I have to face the fact that I can still dev in unity maybe 10 times faster.

I really hope this just lights a fire under Unity to deliver on a lighting solution and get these preview packages under control.

 

How experienced are you with each game engine?

 

I started using unity right before unreal 4 came out. I switched to unreal 4 shortly after it came out for about two years then back to unity for past few years. Have used unity more obviously but still think unity dev experience is way easier assuming you use c++ in unreal.

 

You make a lot of assumptions about the best way to do things, and I couldn't disagree more with the idea of restricting API to 'save' users from themselves.

We're getting tangled up with personal preferences while running around the fact that a simple solution could be available that meets everybody's needs (which is already available in the interface). And people have been complaining about it for a decade.

Besides that, we're getting lost in the weeds when the real point is that Unity has been transitioning between different features like crazy and focusing on things that are not integral to indie game development, while Epic have aimed directly at the heart of what makes game development difficult.

 

Are we sure anything is restricted? It would seem a lot more likely that the blend tree serves as a substitute for any other animation and by default it just takes 1 animation speed input. I could be wrong here but i doubt much time is spent restricting users for arbitrary reasons, but i am admittedly assuming here.
Regardless this starts entering a grey area. It's a much cleaner solution to adjust the animations themselves to overlap each other, it's a good idea to promote best practices at a certain level if that solution is fairly basic, particularly when you are marrying 2 separate disciplines like animation and code where you're likely to have people not knowing the best practice. It makes sense to push people stronger in code to figure out the proper animation setups so they don't end up going down rabbit holes when the simple solution is to just make all their walk/run animations the same length.

 

not making anything up, the changes made are practically non existent or just minor upkeep, compared to the sweeping changes made to the new api. or to the visual effects with the new node system vs the cpu based particle system

or the build configuration assets


i am strongly pro unity, just pointing out that its not fair to compare ue 5 with the impression most people have of the current state of unity and its seemingly lack of new features

 

How Unity help their users charging for a simple dark theme ui?

 

As far as i know, UE4 is only good for pretty looking first/third person shooters.

Their 2D game support is terrible and did not noticeably improve since Paper2d announcement.
No C# equivalent. C++ is to low-level to write complex game logic efficiently and blueprints are to cumbersome.
Editor extension is awful
They are not planning to improve their UI system with something like UIElements.

With all that Fortnite/Chinese money Epic could've solved this issues years ago and Unity would have been buried but i guess they just don't care about indie dev segment.

 

They won't help me in low light scenarios, so I go to bed early and keep my sleep schedule. Thanks Unity.

 

To be fair, unless you have an eye condition, neither does the dark theme. People just like to think it does.

https://ux.stackexchange.com/questions/53264/dark-or-white-color-theme-is-better-for-the-eyes

 

In case you missed, here's more info about UE5:

PS: Feel free to delete it if already posted in this tread.

 

I know, right? The water looked completely out of place there, I'd say worse than the LWRP tech demo one - and LWRP was never aiming for ultra high fidelity in the first place.

 

For me, the biggest difference between Unity and Unreal is the authoring side of things in Unreal is far ahead of Unity. Designers and Artists love Unreal for a reason. So when things like this are revealed, it just puts a bigger spotlight on how lacking Unity is in that regard

There are way to many tools and ideas stashed all over the place - in-editor, in an API, on Github, on the Asset Store, on the Package Manager - in various states of usability and feature completeness. Regardless if they’re in experimental, preview or released, things often feel half baked and/or are locked in a stasis, never to get standard or interesting features for god knows what reason.

Things like the Animation window, a mainline tool in most productions, is lacking and has been lacking for as long as i can remember. And while usability updates have come here and there, for such an important designer/artist tool, it is so far behind the curve. The Animator graph is ok, but missing a lot of features seen in other anim graphs and has been abandon as far as i can tell. And Timeline tool is relatively new and seems ripe for growth, but feels abandon as well. So now its use is really limited out of the box for a lot of game/gameplay scenarios and almost requiring you to use a pretty under-cooked/complicated api to get most things you’d want in a game up and running. Not very artist friendly for such a typically artist focused tool...

And thats just one aspect of Unity - like lets not even start talking about Unity’s terrain solution... Unreal isnt perfect either, though. Its buggy, it crashes a lot. It has quirks about its pipeline that can be frustrating when building a game. And personally, i think the UI is ugly. But god damn is it so much better when looking at it just from a artist/designer standpoint. Like, its not even close atm

It seems like Unity is trying to get there, though. VFX Graph was Unity’s best attempt at being artist friendly in the entire engine so far, imo. Shader Graph is still getting there. And DOTS apparently is the reset button to all Unity’s problems. It lets them start over and reinvent Unity, to an extent. Hopefully this will give them a good base to build up on and allow them to leave a lot of the cruft/depreciated stuff behind.

Also Unreal has been working on this tech for 3+ years already and it doesn’t come till 2021-2022. Unity most likely has a lot of great tools and features for DOTS that have been under-construction for years as well. Hopefully those tool and features are artist/designer focused. I really want the authoring side of Unity to be treated as a premium.

Anyways, good luck to the Unity team. And amazing job to the Unreal team, they really set the tone for next gen

 

Judging by the comments on this video, lots of people?

And it's called UI Toolkit now for anyone trying keep track.

 

First of all, what's the difference between Nanite and Mesh shading?

Second of all, isn't that GI renderer just the same as RTXGI (which is coming to 2019.3)

Why is everyone praising Unreal for implementing tech that everyone else can implement already and just haven't gotten to?

 

its nice to see that there's competition and game development is progressively becoming more democratised and cheaper vs being more closed in like frostbite engine

but honestly, i will never switch to unreal or even consider unreal as a viable alternative until they reach unity's level of catering to programmers. i can see how amazing unreal must seem if you're an artist and don't at all care about dots or C# in the slightest.

Unreal: You will love this if you're an artist
Unity: You will love this if you're a programmer

 

no, they are not the same.

1) it is called virtual geometry, and technical papers are yet to be released, but Epic confirmed that it is not mesh shaders

2) Lumen does not use raytracing. So no. They are not the same.

 

What everybody seems to have missed is the more important implication to Unity's business model, which is UE's new royalty scheme that you don't pay any royalties until you hit $1 million in revenue per title, it essentially makes it free for 99,999% of users, while before it was around 95%, but now even successful small game studios who release actual products can use it for free. Now that's what really interesting, not some overhyped graphics techdemo.

 

The main reason I absolutely want to try UE5 is because of the painless authoring: I just dump million poly models in the scene and artfully arrange them with infinite bounce lighting.

I don't place a single probe, I don't sigh at failed bakes. I don't cry at bad UVs. It's nirvana for people who have a story inside of them. My pain is gone.

Unity is becoming so hard to do anything high end in, and there are endless Unite streams of developers talking about unity's limitations and the pain they had to go through to solve these problems. Then there are Unity's own engineers - their enterprise teams - going on stage and giving out arcane knowledge about things no sane customer should be caring about but have to.

I've been here a long time, supported Unity and still do. I am not going to stop using Unity, because it's horses for courses but really, the workflow in UE5 is the grail of game design, it goes far beyond the rendered result or even target hardware.

That goes far beyond fancy rendering, and indeed if you are not targeting PS5 epic has made it clear it will scale and be baked right down to performant classical rendering, such as mobile.

That means I can author by throwing 13 billion polys at it, and it's all lit. If Unity add this as an authoring workflow, then all is well and good. I'm not actually all that fussed about the technology, but I want the freedom to author without any care. Gone are the days people should be giving a S*** about polycounts, culling, lighting, probes etc. Make it happen. It's hard, but yeah so? Life is hard. Do the work Unity.

Or should I expect less from Unity? I don't expect less from Unity. Unity doesn't need people who expect less from them.

 

"version 0.1/in development/preview"
Yeah what a deal, can't wait to use it in production. Call me up when they at least settle on the final name.

 

1) Okay, let's say that's true (I'd like your source by the way). Would there be conceivable difference between either technologies? We've already seen demos showing "billions" of triangles being rendered on screen using Mesh shading with great performance. The difference in the level of fidelity seems minuscule at best.

2) I would like the source for this also. I'm unable to find information that suggests that Lumen does not use ray tracing. In fact, I'm finding just the opposite:

Source: https://www.eurogamer.net/articles/...eal-engine-5-playstation-5-tech-demo-analysis

As it stands, both of those claims are unsubstantiated. The article above actually goes in detail about Lumen, and it's extremely similar to existing RTXGI/DDGI implementations. However, V-GIM could very well be new tech, it just seems like it would (or could) have no substantial difference when compared to mesh shading. Brian Karis describes it as being "a more direct extension from VT" (https://twitter.com/BrianKaris/status/1260734555532611584), (which Unity is also developing by the way). Anyways, we'll just have to see. Until then, I'll be spectacle of the outrageous claims being made.

 

I like how everybody just made a bunch assumptions based on limited information and a shiny tech demo that currently only PS5 can run. Let's reserve judgment until we see the UE5 editor, how it actually looks and works in practice.

I'm sure Epic has figured out to break laws of thermodynamics and get the same performance from 5w mobile GPU that's only passively cooled (so not at all) in a tight compartment of a mobile phone as you get from a 400Watt desktop monster with 3 fans running at full speed. Unreal magic, I tell you! All based off of a PS5 demo.

 

I don't even want Unity to add this, I just want them to fix their engine and start moving in a good direction going forward. What good is fragmenting into micro modules if they're all still dependent on the engine version, meaning you can't stay on the LTS release if you want new stuff (the negatives of both a monorepo and microservices without the benefits of either)? What good is calling something production ready (HDRP) if it's not, meaning that we can't trust things to actually be ready? What good is a thousand forever incomplete preview features when it takes months to fix bugs?

 

ACT 1

UNREAL:
Now you can feed teh monster billion of food, you can have the monster for free

DEV:
Wtf, where do I get billion of food to feed the monster

UNREAL:
(smiling and going away for the day)

DEV:
(nervously debate if it's good or bad the monster has a bigger stomach, how much it will cost)

UNREAL:
SO what do you think

DEV:
We can't feed the monster, we have reached the conclusion it's too much

UNREAL:
(smiling) well REMEMBER earlier, I gave you free unlimited monster food ... forever

 

Honestly I wouldn't be at all upset if we had HDRP Terrain Grass.
It's just sad to see where Unreal is going and we are still wanting basic features. I mean I get it they are making a new terrain system, that's fine. But if you're going to get HDRP out of preview, at least make sure the basic features we have now - work.

I've tried the multiple various shader replacements to no avail. The most I get is a gray grass quad.

This is what really gets me, I've known for years that Unreal and Unity were two very different engines for two very different purposes, which is fine - that's healthy. But lack of features that were once working, no longer supported when it's part of the core terrain engine is just sad.. I mean what's the purpose of terrain in HDRP? They should have simply deprecated it and removed it from HDRP. Tree's half work, grass doesn't work so what you get is dirt and grass(texture) and sand.

Other than that I really have no complaints. If HDRP was still in preview then hey okay it's my fault for using a preview package, but no-longer preview? Sounds like EA Games - charge you $60 dollars for a game and $30 dollar DLC for 2 new playable characters.

 

Unreal is good for anything 3d. There are racing games and fighting games made on the engine.

The 2d support is minimal, and I've seen pretty much only one or two games using unreal for 2d visuals.
They also have no mecanim, and follow different philosophy regarding characters. Animation is a part of character and not something interchangeable between all humanoids/all skeletons with simlar topology.

 

if you add grass by using add tree instead for hdrp, how much does it impact framerate, would like to know as I'm doing this so it must be wrong lol

 

more interview with details: https://www.eurogamer.net/articles/...eal-engine-5-playstation-5-tech-demo-analysis

 

Regardless of which game engine you prefer to use, Unreal Engine 5 and the new $1m revenue policy is great news. Hopefully Unity will see the big impact Unreal 5 has on the game development world, and tries to adjust their subscription models and fix many of the various issues game developers have with current unity.

For one, release the dark skin for everyone. I mean come on, this is 2020. Every tool and website has a free dark mode. I also think changing the subscription model so free unity users get most of the features of unity pro, and increasing the revenue limits of free to $250,000-$500,000 and pro to $500,000-$1,000,000 would be nice.

I think Unity should start considering creating the next big iteration of Unity Game Engine (not just a yearly revision like Unity 2020...). This new version would be 100% open source and would contain all the packages installed by default. All the UI would be powered by something as performant as UI Elements/Toolkit. The UI would come in both dark and light skins for free, with a modern UI that feels very fresh and familiar.

Right now Unity feels like a 2007 game engine that's trying its best to be modified into a 2015 game engine with the use of mandatory packages, assets, tech demos, and marketing presentations. I think it would be best to start from 0, try to make a new editor that runs extremely fast and has all the features game developers expect, and integrate all the packages and technologies into one install.

And why do we still have two render pipelines (HDRP and URP)? We can't even use normal shaders anymore, and game developers now have to choose between two different pipelines with different shaders and features? All of these features should be consolidated into one pipeline.

 

Why? What's wrong with having templates of workflows, rather than bloating the default project created with a gazillion things slowing down assembly reloads and hence iteration time?

 

Does anyone know of any academic research or anything related to geometry virtualization? This is a fascinating topic to me as I was working a lot with OpenGL and DirectX in the past year and never heard about it until now. I can't find anything on it either.

I hope it doesn't get patented.

 

If you are intersted in how Nanite works, check out this 10 year old blog post about Virtual Geometry from the Epic Technical Director. Of course, the actual implementation changed probably a lot, but the basic idea should be the same.

 

Good question. I believe all the essential features of Unity, like renders, shader graphs, visual scripting, UI, compilers, jobs and entities, IDE support, art tools, 2D and 3D features, navigation, AI, online features, should be 100% integrated into the game engine. Extra features that aren't essential, like ML research and example projects could still be distributed as packages.

 

i dont get what is so special about rtx, where is the game changing?

 

There's so much talk about dark skin that at this point I wish unity would come in two options: White and Magenta, with Magenta being unlocked by Plus subscription, where going from plus to PRO would also allow you to set text color to lime green.

Wasn't there an UI mod for this anyway?

 

Based on the eurogamer interview, there is nothing unity can't achieve for GI/shadow, it's voxel + distance map + contact shadow, it's basically an evolution of the dynamic AO volume they did for fortnite's destructible (bless them to dog food and solve real problem). It's exactly the hybrid approach I said it would be.
I do like I was right on the money about how they did it overall

The (wild) secret sauce really is nanite!

DANG but 16k shadow map (they do use mesh shader, BUT the technique don't rely on them)

DO notice there is only one light sources at any moment, when she "flashlight" it goes off the moment sunlight enter

 

Neoshaman few pages ago posted a link to a pdf which described it. The pdf described technique of splitting model into patches of geometry, then storing coordinates in textures. Sort of reminded me of NURBS.

I think it was this pdf.
http://hhoppe.com/gim.pdf

The tech used in UE5 demo is probbaly very diffferent, though.

 

Nanites, son!

 

I remember people saying the same thing back in the day when it came to 3D graphics cards (back in the era when they started showing up they were separate from the display adapter cards like VGA). For that matter I remember some of my games having better performance and features in software emulation than with my actual card.

Raytracing has the potential to be a massive game changer from a developer perspective by eliminating the need to use smoke and mirrors for literally everything in your game. Currently everything is faked and it's a tremendous investment to achieve anything good.

 

i mean how real time ray tracing in ut5 better, why it is a game changer?

 

That wasn't ray tracing.

 

I edited my post a bit but we haven't been shown anything concerning raytracing in UE5 yet.

 

Unity CEO was imported from EA what do you expect? Do you think he will give you dark skin for free?

He was also saying Unity is focusing also other sections than games .. well while he was busy with "focusing" using countless different non-working addons .. Unreal dropped a huge bomb... completely changed the ,industry.

He was aggressive - revolutionary!! well - here you go .. deal with it now.

Unfinished many tools, buying and killing assets, trying to aim everyone and every business, making assetstore developers crazy with neverending changes, bugs...

A shader guru told in discord lately - unity is a ship with 12 captains, it is not going anywhere...

There is a big elephant in the room ... Unity's many problems are also clearly management related ...

 

That nice but i hope the games next gen wont be like 500 go to download lmao

 

I wouldn't expect 500 GB but I wouldn't be shocked if the majority of them were at least 100 GB with many being 200 GB.

That both consoles will support the user easily upgrading the storage suggests that they expect it too. By the time the consoles have EOLed I wouldn't be surprised if they were shipping with drives somewhere between 2 and 4 TB SSDs.

 

Raytracing was implemented before, on CPU, all the way in 2003.
http://www.pouet.net/prod.php?which=9461

However, that was classic raytracing with phong lighting and not faux-realtime faux-photon-tracer people are trying to do now. Another problem is that they're still using smoke and mirrors which is the "AI" denoiser.

There's always "potential", but most of the time it isn't realized, and I think the whole RTX thing will face the same fate as AGEIA PPUs. Meaning it will die out, but a generic solution doing the same thing might become available instead. Another problem is that improvements in current games are minor, and because it relies on fakery to deal with final pass, you still can't do things like mirror halls. Kinda defeats the point.

 

No. Consoles simply won't have the performance to do everything in raytracing, and the currently available graphics cards are nowhere near capable of running more than one raytraced system at a time. GeForce 30 series will help but I wouldn't be surprised if it took one or two additional generations before it becomes common in more than one system.

 

They stated in the interview that the statue was 24 8k textures. An uncompressed 8k texture comes in at a whopping 85mb, meaning that statue is a hair under 2gb uncompressed. Even assuming compression busts that down to 50% (though we have to get into texture detail density and other such discussions), that's still 1gb for a single asset in the demo.

 

Currently it's a niche thing that's not very practical, but the new UE5 lightning isn't RTX, it's a more performant alternative.

A reminder that this thing isn't even coming out until later next year, and in any product probably even later, in the meantime a lot of things can happen. Stop being fanboys and believing in some magical solutions there WILL be a trade-off, be it storage or memory, or multiplatform support, there always is. Remember DOTS hype? Things never work out as excepted, there's always diminishing returns, compromise, etc.. Same with choosing tools for the job, UE has its quirks and strengths, so does Unity, Godot, CryEngine etc..

And it's not like this is some dark magic, UE5 will be open source and this will be copied by everyone. It's already known what they did more or less, it's been done before in offline rendering, they just put in realtime first. Of course they got there first, I mean UE4 has been at the forefront of AAA engine tech since forever, Unity only recently seriously began to try to become one.
I mean just look at this list:
https://en.wikipedia.org/wiki/List_of_Unreal_Engine_games

compared to this:
https://en.wikipedia.org/wiki/List_of_Unity_games

And there's nothing wrong with that, Unity has always been a mobile engine, if it wasn't for Unity, making mobile games back in 2010s would be a lot harder. And before you say that UE can do mobile too, CoD Mobile and Pokemon Go (which just so happen to to be one of the biggest games of mobile) were made in Unity, not UE. Recently Unity showed ambitions to get into AAA, but it has a lot to catch up, and it's slowly coming along, in my opinion, it is on the right path. I'm liking the 2020.1beta, it's decently stable, better than 2019.3, we just need a full release, it's almost mid 2020.

I can't believe how many people are fooled by a marketing gimmick, gotta give it to Epic, they always know how to market their engine, even their splash logo animation is cool and devs proudly leave it on splash screen by themselves, this is yet another area Unity should improve on.

 

Not sure if it has been mentioned in this thread or not but Epic did the same type of announcement with UE4 regarding lighting and fancy demos. Anyway they had to completly drop that tech cause they never got it working outside of the demo. I don't think that will happen in this case exactly like that but still its not beyond them to over promise.