Unreal Engine 5 = Game Changer

I had already decided to look at UE4 after putting it off for so long simply because I've become jaded with Unity, so the announcement of UE5 and its features just solidifies it for me.

I've spent more time fighting Unity than I have done actual development of my current project.
From baking, trying URP, trying HDRP and returning back to built-in all of which meant keeping up to date with Unity versions, let just say its been a long ride.
The recent debacle of my build attempting to connect to a server without my permission and no explanation from Unity despite repeated requests was the last straw.

Unity's yearly versions with 3 releases per year (now 2) was a great idea (do you recall the cluster truck of Unity 5?) as was preview packages to test future tech and help find bugs / shape it, but the amount of time it takes to move from preview to verified, even then still have issues is not great.
Unity's acquisitions of late have also been very poor from an indie game dev position. I mean who decides these purchases are a good idea?

As someone trying to move from mobile to PC I think its a smart move to try UE4 now.
The biggest plus when the dust settles is without doubt the revenue share model. You simply can't get any more indie friendly than free, which for me is what it will be as although I'd love to break the Epic threshold of $1mil, I don't see it happening.

Then there's the quality of their acquisitions, megascans free for all unreal users, brilliant.
Unity's only decent buy of late for indie devs (IMO) has been Bolt - currently available to purchase on the asset store for $$. Not so brilliant. Not very "democratizing" is it Unity.

And when you see UE5 bringing GI and improved LOD workflows making life easier it does seem like a good time to get to grips with it.

I am excited by DOTS and I don't quite understand the hate being thrown at it. When you can add an attribute to your code which give you 5-8 times performance whats not to love.
I'm also looking forward to Unity's GI response although it looks a long way off.
I'll continue to use Unity for mobile & contract stuff and I'll be keeping in the loop on DOTS but right now I'm looking forward to learning a new tool.

 

From a Bloomberg article some days ago:

 

And as a result the Largest part of the business is neglected.

 

Well, Unity was targeting Film, Automobile, Architecture and Games lately, right? Now, Unreal (in theory) is better choice in terms of amount of work it needs something to work and it's practically free. Unity probably will dominate Simulation area with DOTS, but damn, it lost Film Industry for sure.

It's going to lose a quite big percent of new users and some of big indie studios now that is free.

I'm not a big time dev but if this thing just works, then sorry Unity but you give nothing back to your users while charging even a DarkTheme of your editor.

If this moment was 3 years ago, I'd choose Unreal without any second thought to learn and invest as an engine.

 

Come on, not in any way. AAA is evolving at a speed that's only possible when you have hundreds, if not thousands, of developers working full-time on the bleeding edge. The demo you're basing this on is made possible due to some (in reality probably a lot of) extremely skilled individuals dedicating years of their lives to getting to this point. Are they just going to stand still and wait for you? All you could do is try to catch up to where they were at that point in time and by that point they have leap frogged your efforts completely, and that is ignoring the sheer amount of manpower required for AAA development to begin with, which effectively makes it an impossible task for a single individual anyway.

AAA development will never be attainable by a single individual or small team. If technology like this and for example new camera/photogrammetry hardware comes out that allows anyone to just go out in the world and scan fully working real life models directly into the game engine then high quality models won't even be a factor worth mentioning when talking about AAA anymore because everyone has access to it. Then AAA will be about other things. They already have access to composers, actors, voice actors, foley artists, sound technicians etc. etc. etc. and all the experience and hardware they bring with them that you don't have access to, but maybe AAA suddenly becomes about artificial intelligence, machine learning or big data? The same could be said for lighting. The day everyone has access to unbiased path traced renderers for lighting their games then AAA development won't be about having impressive lighting features anymore.

And why wouldn't Zbrush assets get outdated? They get outdated just like everything else. Not to mention we've just been shown technology that if it actually arrives in the form it's shown here could disrupt the entire industry. There's no saying how the industry will adapt or where we'll be when this technology actually arrives. Maybe a new tool pops up that is much better suited for the task and Zbrush falls out of favor just like how Substance disrupted the industry?

From my experience you're completely wrong here. If you try importing a "raw" model into Unity, Blender, 3ds Max or any 3d application today they'll become unusable. And that's just with one model. Imagine trying to work on a fully fledged scene composed of such models. None of these applications even come close to supporting billions of polygons, not even the sculpting applications themselves. You need the same polygon streaming tech throughout your pipeline to be able to work with these assets. For that reason for this to be possible in Unity, Unity needs a renderer that works the way the renderer in Unreal Engine 5 does.

And as far as I know the "hardware architecture" of the PS5 is just a very fast SSD. The next-gen consoles are extremely similar to the high-end computers you can buy today. And there's no reason you couldn't just use the same tech for any game. I don't need 30 million polygon source models or 8k textures in my games, but the ability to just import source models and textures directly and not having to worry about texture budgets or poly budget or lods or normal maps sounds like it would be wonderful for anyone. And if I reduce the texture resolution of my textures by 4 times or 16 times and or the polygon counts of my models by 100 times compared to this demo then the SSD requirement will be reduced accordingly allowing the tech to be used on a much wider array of computers.

But for me how any of this will work is pure speculation until we actually have access to any of it. Both these techs sound wonderful. There are so many steps in the game creation pipeline today that are just there due to technological limitations. Removing any of these is a benefit to everyone. The GI is personally what I'm most excited about, but raytraced GI is already a thing that you can actually use today and I can't help, but wonder that by the time this comes out then raytraced GI would be accessible to a lot of people anyways. Raytracing to me is extremely exciting as the only thing holding us back there is hardware which will gradually get better and potentially it can also suddenly get a lot better if someone discovers clever ways to approach it. And there are also software developments making raytracing more approachable with every passing day. Something like Quake 2 RTX or Minecraft RTX looks just as impressive, if not more impressive, than this demo and that's something you can play today, and on a high-end computer probably in the same resolution and framerate as this demo. As far as I know raytraced GI doesn't work well with tons of polygons and this GI obviously does so maybe they end up becoming two different approaches. Choose between low poly and better looking lighting or higher poly and more basic lighting.

As for Unity vs Unreal for me I haven't looked at Unreal since UDK, but back then I remember hating it. The fact that it's artist focused doesn't interest me at all. I'm barely inside the Unity editor and it would be the same in Unreal. And Unreal had so much baggage in its API's and workflows, probably stemming from the fact that it was originally a AAA shooter engine that trying to use it for anything else felt like you were hacking it to do what you wanted. And just from looking at it today I feel like almost everyone is making 3D games with it with all the default post processing enabled making everything look exactly the same. This obviously has more to do with the skill level of most of the people using it than anything else, but it doesn't help that the engine lends itself that way, and that there are so few resources geared towards taking the engine in a different direction. And the fact that anything you actually end up finding involves Blueprint is probably the biggest detriment to Unreal Engine for me. The same goes for C++ vs C# for me. C++ may be the "best" for game development, but in my experience it's extremely hard to find C++ examples on how to approach common things in game development, whereas with C# it's everywhere and you even have access to everything .NET related. With Unity I can just google any topic and find information, tutorials, example projects related to exactly that sort of thing, but with Unreal, at least in my experience, unless it's related to artist workflows or the default 3D games then you won't find much. As a hobbyist this is by far the biggest advantage of using Unity for me. If suddenly everyone switched to Unreal then this would probably change though, but that means everyone else would need to jump ship before me. This could also change if Unity adopted visual scripting and "everyone" suddenly started using it.

 

Unity has cornered indie, as such there is no growth for them there, as such they do not cater to indies at all (and they raise prices without second thought, don't care about basic functionality being broken, have the editor not work, because something broke in their phone-home pipeline and don't think they owe us an explanation)

So Unity is targeting other areas, but stuff like the backgrounds in the Mandalorian was using Unreal, so I don't think Unity's winning that race either.

 

I don't think Unity is winning any race in any field.

Last year, I was waiting for 2019 LTS for the features promised that is production ready. Now it seems we should wait for 2020-2021 versions for the same features that should've been production ready.

 

Such an incredible amount of work ! Woah and directly from zbrush, importation
Damned, its beautiful, loved the last sequence (cringey but hey )
I want to do a 3d full game with full flight hahaha

 

Nope...they are going crazy as expected...

 

Seems people using unreal want a scriptable render pipeline and C#.... go figure

 

I'm a hobbyist but I suspect that my demographic is responsible for a fair bit of the income that comes into Unity.

Hobbyists are often well paid professionals in their 'real' jobs so happy to pay regular money in the Asset Store as they lack the skills to generate their own art assets. They don't usually have tons of time but they often do have disposable incomes to spend on their hobby.

They are also highly likely to 'jump ship' on vanity projects when they hit insurmountable problems thus re-starting the buying cycle of assets, they very rarely finish projects but keep buying stuff to make their hobby enjoyable.

This is all good, the hobbyists have fun, asset developers have a viable income source, the platform grows as more and more high quality content becomes available.

Unity was really easy to get into, that's why I'm here but the recent stuff has been utterly baffling. I can't make head nor tail of all these different pipelines or make a decision about which one to use in future, why all these different acronyms and letters? Nothing is clear, nobody can give a straight answer.

This Unreal stuff is a breath of fresh air, Unity is starting to feel like a tool that's getting in the way of the artistry, it seems like it's becoming more complicated and less effective.

If I'm not going to finish a game, I might as well not finish a beautiful game that's easy to construct....!

 

Scriptable render pipeline... you are joking right? Epic is also working on C# integration now

 

maybe they are a unity employee undercover as an unreal user on the forums to misdirect development
Nah the SRP is fine. I’m using it and i can do some things I couldn’t before as easily. I’ve seen some awesome videos of others getting amazing effects out of it

 

I do not think the situation is so grim. Unreal is the leader in the art field because the push for a cinematic experience and allocate the most resources here. Some days ago i had said that does often not add too much for many cases outside architecture and film because for an untrained eye (like mine) the effect of a technology are only slightly visible. Yesterday changed that when the new technologies reduced the amount of work for an artist.

It is true that Unreal offers more tools than Unity out of the box but these tools were designed for special games and are not well rounded. That means you have to look for your own solutions and they have to dive in C++. I have noted this for some AI solutions which were nice to study but helped not too much.

An advantage for Unity could be ML, Data Management and Simulation. For ML Epic offers nothing and btw in UE4 DataTable can not be updated runtime (of course you can overcome this little hurdle but it shows Epic's approach).. I know already that some firms prefer Unity for simulations because of the use (!) of DOTs. In such cases the small loss in graphics is negligible. In my opinion ML and AI is still in their infant states and could become a huge topic. At the moment nobody knows how to use it correctly - some oversell ist, others underrate ist.

That`s why for technical cases Unity could become more dominant and Unreal may have the edge in visualization (broad speaking).

AR/VR i can not really evaluate.

 

Look at all these people fantasizing and making big plans to switch to Unreal because they showed off some features that only a few top tier AAA studios will ever be able to use, and even they will most likely make a boring soulless uninspired tech demo of a game that will sell well regardless because of marketing. And when you look at what is actually released on Steam by so called indie devs, if anything at all, well...you might as well use OGRE 2.0 or Unity 3.5 for that.

In other news: nothing changed, making good games is still as hard as it always was, none of the people that posted in this thread will even make anything close to this unless it is released as complete pack for free and they sell as an asset flip. So just ignore and carry on, use what you like and try to survive in this insanely competitive market by optimizing your workflow.

 

Wow, you are really arrogant. If you can't add something constructive other than your twisted generalization, then just read. Also, there are many talented, experienced veterans here. Learn to be humble, and maybe you will learn a thing or two.

 

I have decided to avoid the render pipelines until they appear to stabilize. Yes, I could probably go and use them right now and get some benefit, but frankly as a solo developer I have limited bandwidth for learning new features and there seems to be a huge amount of risk in stuff breaking, especially for any asset store products I want to use.

When you are indie you have to ruthlessly optimize your time and investment. I have a vision of how I would like my workflow to be in an ideal world, and the further away something gets from that, the less interested I am. Right now realtime GI without any need for baking is exactly how I envision handling lighting. Automatic LODs is how I envision handling LODs. These new features from Epic could hardly be more relevant to my interests.

 

My game dev time has always been split between actually creating a game and dealing with Unity bullshit. In recent years the time needed for bullshit Unity workarounds has increased, to the point where there are days that it's all I do.

So excuse me for seeing Unreal show stuff and have some hope inside that : "Hey, maybe controllers will work in Unreal and I would been able to update my game now!".

 

Epic has been working on C# for well over a year now and there's still no signs of when it's going to hit. Also, yes, having a scriptable render pipeline IS a big deal, especially if you need to target a wide range of devices.

 

Speak for yourself.

 

Whatever the source of this information, it's incorrect. The actual ongoing work is on a new programming language that intended to help developers in solving the most painful aspects that we usually encounter while making games.

 

Ignore the assets. Modern tech demos are using photogrammetry for most of the assets. What you want to pay attention to is the realtime global illumination and the fact that you can zoom in on an asset and the quality doesn't appear to go down at all.

 

Oh i see. Well friend is using something called 'dxr'

https://forum.unity.com/threads/wip-dxr-test.889612/

is unreal using the same or is it something like radiance probe lighting? thank you

 

Neither. Lumen is a realtime global illumination system like Enlighten but without the need to bake anything at development time. DXR (DirectX Raytracing) has a similar advantage but comes with the disadvantages that it's insanely expensive and depends on hardware support whereas Lumen can run on mobile devices.

 

Apparently Unreal's solution is not using dxr and scales much better.

Raytracing is not that usable currently, frankly. Current generation of cards can barely do a few samples per pixel and that's simply not enough. Maybe in a couple more generations.

 

So what is it?

 

https://en.wikipedia.org/wiki/Global_illumination

 

No I mean what is unreal using to accomplish this?

 

A completely custom implementation they made called Lumen.

 

Ok so there is no research paper.

 

They haven't said. It seems it is some kind of realtime lightmass calculation.

 

Same here, even the simplest stuff is a pain in the you know what these days.. The saying in unreal 'it just works' with Unity 'Nothing works but if your work your but sorting thru the mess you might get it to work' , unfortunately this is the sentiment. Some one high up in Unity not to mention any names now that it is illegal, once said Unity will some how sort it self out, well he ment thru the hard work of Unity users who will forever report bugs, and if you're lucky it will be fixed.

Also, I bet Unity will shut this thread down as it contains no useful information so they will say..

Why is Unity So Silent About This HISTORICAL EVENT IN COMPUTER GRAPHICS?

 

They say this ironically because it's never true of any engine. The only people who say this who think they mean it are people who work for Epic.

Because it was literally just announced yesterday?

 

They're too busy drooling over Ampere.

 

I imagine them with heavy alcohols and trying to figure out why we are so angry at them, it puts a smile on my face.

 

I'm drooling over it as well but this has not much to do with it really.

 

They will release the technical details soon apparently. I am not sure how deep they will go and also if they will ever release a detailed research paper. It is proprietary afterall.

As for more information, the demo was running on ps5, in 1440p 30FPS

Now, with that information, given that ps5 gpu is about 5700/2060S

and assuming that the tech won't scale as well in PC, let's assume that to achieve 1440p 30FPS, we will need 5700XT/2070 level GPU on PC with similar scene.

We do have to assume that nanite takes a considerable chunk of GPU time as well,

So, with all this in mind, I have to say that 1080p 60FPS is very doable with 5700/2060S level of GPU in a vast open scene in traditional rasterization without raytracing.

To me that sounds like a variation of sparse voxel tracing, because there aren't that many techniques that can give that much of GI with that little performance hit...

But that's just my prediction, in a few days, we will get to know it exactly....

 

Both of the new consoles have AMD's RDNA 2 which according to them is 50% faster than their current cards. Clock speed is way higher too with the 5700 XT only running at 1.68 GHz (1.98 GHz boost) while the PS5 is running up to 2.23 GHz.

https://www.digitaltrends.com/compu...ay-announcement-ray-tracing-xbox-playstation/

 

I am not sure if we should take it at face value...the number of CU units though I believe

But assuming the PS5 GPU is about 30% faster than 5700, (50% is a number I just can't believe) I think we can comfortably put PS5 GPU in the region of 5700XT (most likely not as fast...but let's just pretend to believe them)
Then we might actually have to consider the fact that the performance cost of Lumen might be higher than we expected.
Still light years ahead of anything available, but a little heavier than we'd like.

 

I doubt they will say anything about it, their default approach to dealing with riled users appears to be studiously avoiding engagement while focusing on data and analytics. Even when they put up a thread specifically asking for feedback, they appear to avoid replying to comments that are critical or express frustration. I could hardly imagine anyone from Unity jumping into this thread to explain their position.

 

Let's hope it brings more sensible prices to 3-series. Amd need to get their S*** together so we can buy nvidia for good prices

 

It's a very, very impressive tech demo, obviously. Better graphics with better workflow is a win win. But personally I feel the graphics are already good enough. Most PS5 gamers will just run though all those details without even looking twice.

 

VR gamers on the other rhand, they take their time to really take in the scenery

 

Hi! Saw the Unreal 5 tech demo yesterday. looks beautiful! maybe I'm missing something, but isn't the Nanite technology just mesh shaders / meshlets put to use? (https://devblogs.nvidia.com/introduction-turing-mesh-shaders/) (https://devblogs.microsoft.com/dire...on-shaders-reinventing-the-geometry-pipeline/)

Imho, it does not look like they've invented anything. Just took existing stuff, built engine scaffolding around it and gave it a marketable name before anyone else did. Not that it isn't impressive, but it will eventually become the norm, since it's just another step in the evolution of programmable render pipelines just like geometry shaders/tesselation stages were. All engines will have to eventually adopt the same technology.

Issues will shift from runtime performance to "how the hell do we store and move around such a huge amount of data?"

 

To be fair Unreal just bled out Unity on the altar to the 3D gods.

There's blood in the water.

I'm still reluctantly glad to be working in Unity. My project isn't based on fancy ass graphics, but my enthusiasm for Unity has been waning.

Does this recent tech demo directly contrast Unity's inability to create anything tangible and foolproof as of late? Absolutely it does. "Unreal" and "Unity" are just words... a brand. There are solid coders and project leads at both companies. But something is misfiring at the core of the Unity infrastructure.

Do they have code standards? Are they recklessly pushing forward across the board? Why does it feel like the people working at Unity have lost their "spunk"? They need to really look inward and reassess how they utilize and grow their talent. It doesn't appear as though their best minds are influencing others. There is no growth. Their ambitions have not snowballed they have been met with recurring failures stemming from what seems to be poor infrastructure.

On a surface level evaluation you can really tell which of these companies puts out games and which one doesn't. Unity needs more people in senior positions who have actually shipped things. Perhaps this is why we're seeing the buyout of so many third parties as of late "We need people who have finished something, but affordable without too big a head". It's one thing to make a little tech demo, it's another to go through the entire motion of making a product. It's one thing to add a concern in a forum to an endless "to do" list, it's another to fix the issue because it's a pain in your own ass: https://forum.unity.com/threads/eng...e-objects-along-the-least-likely-axis.878014/

Enough of this preview crap. Give us working lighting, working post processes out of the box. Not something that looks and performs like some 3rd party free off the asset store knock off. It's starting to get ridiculous. At this point I wonder if the only reason I'm using Unity aside from the years of investment into my project is because I don't know C++.

Unity needs to ask themselves. What am I? What is my specialty? And they need to ask it 3 years ago and get some passionate developers who ship product to execute on that idea. Then they need another team to look at Unreal and how they handle lighting and post processing and implement usable and effective tools in their engine that can compete with what they had 5 years ago. The fact they have yet to figure this out, after the tech has already been proven and can be picked apart should have heads rolling inside their company.

Whenever they demo stuff they look so smug like they've got some fantastic thing to show off that's going to change everything. And we're over here looking like jackasses for buying into it. When I look at Unreal I see a hyper competitive company that will throw people into the fire for that edge that will make their engine better. When I look at Unity I see a sort of cliquey in crowd that think it's super hip to be in gamedev but have never really shipped anything particularly impressive.

Less blacksmith trailer smoke up our arse videos and more bug-free tools. Stem the bleeding, build your team, create standards of excellence in code and design. "Good enough" isn't good enough. Stop running around with your heads cut off and build a strong foundation to grow on with a vision. Stop burning out your coders with ill-formed ideas. Gamedev isn't a brand, it isn't cool, it isn't smug. It's raw cold tools that do what they are expected to do in an ergonomic manner.

It's frustrating to see the platform you banked your entire future on so recklessly kept.

 

This has come up multiple times, but basically Tom Sweeny has said no.
I thought it was mesh shaders too but apparently something quite new. What exactly, I have no idea.

 

This is the heart of all this reaction...I love Unity...but they keep pushing me away.

 

Same thing they showed, was shown by Nvidia a year and a half ago in the asteroids demo. I'm pretty sure that whatever it is, simple or complex, it's built on top of mesh shaders. Wouldn't make sense not to use them, as they're designed specifically to increase detail and move LODing burden onto the GPU. All we can do right now is speculate though.

Not taking credit away from them, just saying that this is the direction the industry is moving in, they just picked up faster than anyone else. Unity is still rebuilding/fixing basic engine functionality, so undoubtably years behind Unreal. This just pushes Unity further behind.

 

ok i think i finally figured out how they removed the poly limit.

yeah its not dx12 ultimate mesh shaders, more like a hardware driven solution.

they probably have polygon budget for every frame which depends on the resolution. it could be 20m or less. it means there is a budget for every pixel. how many poly per pixel depends on raytrace results. every frame they raycast for every pixel (which makes sense because they can release this now with nvidia raytracing tech and also new raytrace console gpus, it wasnt possible before) and maybe with some octree algorithm & volumes. then show polygons per pixel with similar solution like virtual texturing.

this can be a raytrace results debug


of course there are some other problems on the way, like how to store that much geometry without hitchups, compression techniques etc. maybe they're using geometry images so thats why they call it virtualized geometry. 4k rgba map can hold the data as much as 16 millions poly mesh (or 8k = 64m poly). so they can store meshes as images and when needed gpu converts them into geometry on the fly. it probably has some great advantages, as well as some other problems but it seems they fixed / they're fixing them.

and also everything should work with this new workflow, which means a lot of integrations. but besides the poly limit, other features are not that impressive on that video (especially other than gi, we need some details on gi first).

also if that is a feasible way, all the industry would adopt it as well, hey its open source engine. also its better for them as well, to have improved hardwares for new industry standart.