the bad thing about unreal engine is that the tutorials are very limited compared to unity. It looks simple but once you dive deep into the systems it's very complex to do by your own.
@@MicroMintyHourglass It might sound that way but unreal devs are few and far between, and teams often have to go out of their way to employ a dev that knows how to use unreal properly. It's great if you want to be a well paid software engineer, but if unreal was better documented then we would have way more indie studios using it.
@@allenantony6924 That being said game dev is stupidly complicated to begin with and has a enormous amount of failures. You're going to be putting in the hours anyway so why not use an engine that not only can get the job done but has many offers for you now if you can?
Im a lead VFX artist and the volumetric explosions is interesting but very expensive. There are tons of tricks one can use on billboards to make them sit in the environment such as normal maps and dedicated light calculations. Volumetric for fog is super useful and have been used in games for easily 10 years. I think volumetrics for VFX will only be useful in a few cases with current hardware since it will suck up all resources.
I definitely see volumetrics as an aspect of vfx that ai acceleration can help with to overcome some of those resource intensive calculations, especially when light scattering is also a metric to consider.
You could use volumetrics for VFX in important and meaningful moments that the player is sure to pay attention to and not everywhere else where it wouldn't make a difference, it's the best use case i see until the performances are unimpactful in the future
I work with offline rendering, and can only agree. I believe volumetric objects will definitely have their place (for dynamic effects or volumes that need to be spawned when needed, or even things like the fireplace examples where the camera can freely move around it), but billboards or sliced volumetrics with normal map relighting will definitely live on for the foreseeable future due to the memory and performance restraints that we currently can't work around. And to elaborate on what you're saying Andreas: Volumetric fog is a completely different technique compared to sparse volumetrics. (as far as I've understood, I'm not a coder or touch back-end in any way) Volumetric fog can afaik be "faked" with a lot of clever calculations based on light position, depth mattes and vector math.
@@pinolskun8764 I believe the OPPOSITE is true. For important and meaningful moments, the effect can be precalculated and will most likely have fixed lighting setups, meaning that you can use billboards instad as the lighting doesn't have to be dynamic or be viewable from any angle.
Cant wait for the next 500gb game that features UE 5.3. Surely it will be optimized for the average user and not be exclusive to certain gaming rigs and not problematic at launch coz of how demanding the graphics will be. Woww amazing time
Yeah, because even tho we just started with UE5, some games just only give a shit about graphics, but the gameplay is horrendous. Hopefully, that's changes now, I want a balance between em
Its hardly different from LoD, just doing it dynamically, which will be the source of endless amounts of detail fracture. Imagine instead of having a fixed distance at which the detail changes, to instead every object randomly changing their detail as they please. Donno what Asmon is getting his knickers in a twist about, or you KitCat. This option can easily become the most immersion breaking, texture fracturing thing in existence. The only thing to look forward to here, is how bad it will get.
@@soulofartorias9928 If you could comprehend what you are being told, the video literally told you what i did. LoD -> Fixed distance, Nano -> Every object changes complexity individually. Asmon and every each one of his viewers seem to suffer from a case of room temperature iq. All you understood from the entire thing was the pictures. Make sure to clean the drool from your keyboard once you are done watching this.
In LOD: the designers have to make all the versions of the model, so they make a high def model for close range and 2 or 3 other models for when the model is far from the player In Ninite: you just make 1 high def model, and the engine will change the polycount dynamically, even when you're close to an object, it only shows you the part you're seeing in high def, the back of the model will be low def.
Having used LOD generators myself (like DynDOLOD for Skyrim VR), I doubt game devs are manually creating LoDs. Nanite just skips the LoD generation step but seemingly at a performance cost.
Correction: devs use dedicated LOD tools to create lower-LOD models, often automatically and in a batch. Nanite could save space due to only needing one model... but more realistically devs will just push higher LOD models and game size will increase. Nanite could also improve performance as it can have pixel-perfect LOD switching, whereas the classic way leaves gaps in either performance or graphics between switches... but more realistically devs won't optimize their Nanite implementation for performance and games will skyrocket in terms of requirements.
Most retopology is done quite well these days; Nanite has yet to prove it can work at scale. It looks great in confined spaces with limited fulcrum rendering, and in the isolated cases it was used as a test demonstration in Fortnite, but it remains to be seen how well it works (when optimised, like you mentioned) on larger scale projects.@@sujimayne
Someone will use engine in few months, in a year or so... still don't have anything close to UE3 tech demos from more than a decade. CGI and in game/engine cinematic do not count.
@@Wes-Tyler I'm not sure about that, it is much easier and more profitable for indie developers to just keep making vampire survivor clones. or whatever the flavor of the month is.
I absolutely love the tech jumps in Unreal Engine but I also feel sad about the fact that their realistic asset store will most likely extremely lower the amount of large game productions with their own distinct styles
Large studios tend to make all their assets inhouse so this will not really change anything for them, they'll just 3D scan real world objects like normal and import it as 3D nanite mesh into Unreal. The existing megascan asset library within Unreal is catering to Indie first and foremost, to the people who can't just go out and 3D scan 1000s of environments themselves. But AAA studios absolutely will, they been doing it for years -besides also directly modelling what they need for their own styles.
@@ikcikor3670 Key words "able to". If their games start looking like what some 3 man indie studio cooks up in a year then they will start losing money and fast = unable to
18:00 This engine powers games such as Fortnite and Valorant. While these games appear distinct in style, the choice of aesthetic largely depends on the Art Director and the development team. A challenge in the industry is the prevalence of affordable models available for purchase, which often have a similar appearance. While these models expedite prototyping and reduce production costs, they can lead to a homogeneity in game visuals. This trend might result in games looking more alike than the influence of engines like Unreal Engine 5.0.
As a dev working on a game in ue5 5.3, the changes are welcome but things like nanite foliage/nanite landscape and tessellation and sparse volume textures are EXTREMELY expensive. Dont expect games to makeb big use of them soon
with nanite there is actually a bit of a caveat , if your landscape is a city and your surfaces are mostly flat then it's doing a lot of processing for relatively little return at that point you are better off without it in certain circumstances however for forests and more natural type stuff what happens is that there is a slightly higher processing demand but it scales much better and way way more than what they usually show in these videos
The good thing is that not everything has to be nanite, you can toggle each element independently, but I would say for any world where you have even just a hint of distance variance from objects, nanite is the way to go rather than LODs. LODs also take up more package space than a single high-res nanite mesh+texture.
@@Real_MisterSir thats pretty much what you do, use nanite on the objects that benefit most from it, you can also turn it on and off at runtime as well but then you can loose some of the advantages of keeping it running on those objects, realistically every landscape or map is different and should be tuned properly
Devs won't do this; that is extra development time. Nanite games will end up being wholly Nanite with poor implementation and performance will suffer, while game size will increase. A story as old as game engines themselves.
@@sujimayne Nanite will very likely reduce the size of games since multiple lod meshes and textures maps are no longer needed to represent the same object.
Some games using UE 5 with Nanite are already out, and they look pretty good as long as you have a 3080 or higher to run them. Immortals of Aveum and Remnant II are two examples.
@@AG-ld6rvThey don’t looks very good considering how much graphics power they require to run. Immortals has terrible stuttering even with the best of cards. Is actually quite surprising how badly it ran and looked considering all the UE5 hype. Haven’t played Remnant II but Immortals made me very sceptical. It looks good, but not incredible by any mean even on the best cards. Plus it was a very linear game with a short campaign that stuttered so badly. Imagine then how open worlds would work with UE5…
@@pressrepeat2000 Stuttering is a separate issue that good programmers using UE can fix. That's like saying mechanics are bad, because you took your vehicle to one once who broke your vehicle further. That's just a bad mechanic among many mechanics, and a good mechanic will fix your vehicle up. I agree that any game with stutters needs more optimizations like shader precompilation implemented. As for the looks of these two games, they look really good to me on maxed settings and on PS5/series X. Series S looks like dogshit though for Aveum. As for the common "optimization" complaints, this is as old as 3d video games themselves. What happens is the programmers of a game dial up the graphics such that, to run the game on max settings, you need the absolute best hardware out there. So basically, every single new game that has great graphics as a selling point will run like shit on outdated hardware and sometimes even on relatively recent hardware that isn't the best on the market. That or the settings to get the game running totally obliterate the graphical fidelity of the game. The companies have as a goal to create a game that looks amazing and runs all right on consoles and on expensive PCs. Unfortunately, I do not believe it is possible to optimize high fidelity games for the myriad of hardware combinations out there in the PC market. Examine the look of Aveum on series S for an example of how optimizations, even with the hardware fully known, cannot overcome the limitations of a cheaper system for a game using some of the most expensive and cutting edge techniques to create the highest fidelity games possible. The simple fact is that Nanite is a technique that is more expensive to run than the older "level of detail" approach. Any game that uses Nanite will naturally have higher minimum and recommended hardware. They'd have to code the game twice, one with Nanite and one with LoD, to improve performance on what is, when it comes to running Nanite, insufficient hardware. There are base limitations on what can be done, and even with intense optimizations like likely occurred with Immortals of Aveum for series S, the game is simply going to look horrific and still run poorly. It's not like the company is just trying to bone people with cheap PCs. Nanite just takes a certain baseline of performant hardware to achieve. The fact is you need to run some kind of AI upscaler and put up with the massive amount of artifacts if you have a cheap PC but insist on playing games that, in terms of graphics, push the limits of fidelity as much as possible, and UE5 was basically made to make games that can push those limits. An example of that is Nanite as I discussed in my previous paragraph. I'd say that Fortnite is likely the most runnable game out there that uses Nanite, but that's only due to the game creators not having "has the highest fidelity possible" as a core requirement of the game and instead has "we want this competitive shooter to run on as many PCs as possible" as a core requirement. Naturally, the graphics in the game aren't that stunning due to this design priority. Still, Fortnite would run better on even more PCs if the game didn't use Nanite and instead used the level of detail strategy. Like I said, Nanite simply costs more to do. You can't have your cake and eat it too. If a gaming studio wants to maximize the number of PCs that can run its games, they've simply got to avoid using expensive features in UE5 like Nanite. But they likely secure more sales exciting people who own a PS5/Series X/decent PC with Nanite than they'd get not exciting those people just to have an additional, small segment of the market able to run the game decently (while looking horrible).
@@FightClass3 for just graphics yes, but your hardware will be able to run say a gta 5 style game with 200 times the content that is interactive and that opens massive doors to endless madness.
For anyone who doesn't know is that this cuts down on a LOT of production of the same asset that needs to be remade into different levels of detail or LOD. The engine itself does the hard work for you by changing the polygon count on the objects that are being rendered.
Sort of; most really good artists can have the retopology done on a LOD0 asset in a day or two. The question is, is Nanite better optimised for a wide variety of platforms than say a LOD3 asset?
It comes at the expense of having an actual polished and functional game on medium-end hardware. The use of nanite technology will not have any real impact on game development, and if anything, games will take even longer to make and run worse because of the taxing nature of nanite technology. It bloats the games and makes them require insane specs like Immortals of Aveum and Remnant II. Both are great examples as they run at 720p on current gen consoles because of terrible optimization, bloated use of fancy features and the results are mediocre in terms of visual fidelity and artistic direction. Just look at how gorgeus The Witcher 3, Elden Ring, Armored Core 6... etc look on the same hardware and running at higher res with higher frames by the use of smart development and proper optimization. Nanite technology is a great tool to use in certain situations, but as with FSR and DLSS, it should only compliment the development of video games, not make the industry over reliant on said features to make their games barely run.
Another way to look at it is that the easier it gets to make a good looking game with tools like this, the more time can be spent on other aspects, like gameplay mechanics and story - of course, the issue is whether developers will actually use that time, or just push out more crap with a make-over in a shorter time frame.
@@natevoid9955well everything in tech improves then becomes efficient that’s just what I’ve seen in my life. First the photorealistic imagery then the improvement in efficiencies
@@natevoid9955 We are increasing fidelity at lower resource demand, so it's a beneficial scale over time. If graphic demand grew exponentially higher than the fidelity improvement, it would be a problem long term, but in reality we are seeing the opposite. Some things are even running more efficient now than before, so we simply have more headroom than we used to, which incentivizes to make use of that headroom (and maybe push a little beyond).
@@natevoid9955as it stands right now hardware is years further than software. The reason games run bad today is absolutely dogshit optimization. Most gaming pcs can run UE 5.3 given that developers did well on the optimization
I am not impressed until i see this actually being used and implemented into the games we play. It's like on those amazing news you hear, how this is going to revolutionize whatever and then you never hear about it ever.
There's a lot of games that have Nanite + Lumen already, but you can't notice it that well, because most faked lighting looks as good as dynamic GI and when it comes to LODs, you also can't see the difference pretty much. I believe Fortnite has both Lumen and Nanite for example.
i agree with you fully, the only game that ive seen using this technology is, surprisingly, Dead Island 2. the environments and lighting in that game was immersive and beautiful. the hollywood hill houses that you run through were breathtaking.
There are some Unity games that can use lumen with mods. Runs like 1/4th the normal fps, but it is there and made by a regular dude. They could always update the games tech like Cyberpunk and Warframe do.
There's some videos out there showing the evolution of The Unreal Engine...stuff we're seeing today was only a dream 5, 10, 15+ yrs ago...its somewhat sad many take for granted what they see today and don't fully appreciate what's out there now; Unreal or any of the other Engines....
18:00 the fact is its all comes down to art style you can make a game look any way you like its up to the devloper to design the game, if they look the same as other games then its because they chose for it to look that way.
Game developer here. What most do not understand is, even with all this fancy tech, this is only the visuals. If you consider everything else from the game, the performance drops big time. And did I forgot to mention shader compilation? Yeah. This issue still persists. It is all sugar coating at this point, Unreal cater more towards the VFX movie industry than to games at the moment.
and it's why we've gone backwards to a huge degree in world simulation, immersion and interactivity almost constant decline and silent removal of aspects of gameplay, often small things, physics, environment destruction, items that have no purpose that you can play with or effect in some fun way. So many titles from 15-20 years ago made things other than GFX higher priority. Gorgeous but 'dead' world we see today, aside from maybe RDR. This is the aspect of Bethesda games people gloss over all the time. The impact of game wide object tracking for hundreds of hours has on what else they can do. Was the rise of middleware like Havok partly to blame? How many studio's bother to create bespoke simulations today? It's going to the few studio skilled enough to persevere with their own custom engines that stand out amongst the homogenized UE5 decade ahead.
Physics is a different topic but I do understand your point of view. Most modern games tend to focus on gfx instead of a believable world that is filled with interactions and different paths to go. I am not agreeing we go backwards, but it clearly shows how much effort is wasted on photorealism. @@cypherpleb
@@eth7928 how much effort it's taken by the megacorps, how fake the idea of games journalism must be to normalise the market to just accept worlds (and NPC AI) that behaved like Far Cry 2 did on the Xbox 360 in 2008, or MGS2 on Playstation 2 in 2001 basically hardly ever be surpassed by most big flagship releases even today on PC's with supercomputer specs vs the limitations those titles were built within. Say what you want but fixed limited hardware specs show the industry who the real technical and creative geniuses are. Few games even attempt to push what's possible today, perhaps only PC exclusive sims such as Arma, Fight Sims etc... in theory Star Citizen? and they don't feel like games to me anyway. Most other games, especially anything from a public firm simply wouldn't be allowed to leave millions of customers on the table. Lowest Common denominator led game design is what we have. Mid range laptop APU's from 2012 still constrain nearly every multiplatform game you play on you £4k PC in terms of design.
Big studios became lazy and too big. A big studio has too many positions of staff where ideas go through and that complex management results in mediocre games. Look at D4. 9000 Devs and its not coming close to the more than 20y old D2. I absolutely believe that a small team of people can do amazing games if they had the same budget as AAA title. Studios have become so big that risky investments are off the table. This is why you won't see new IPs that much and no revolutionary tech in video games. The tech now is being developed by 3rd party contributers such as Unreal or Unity. Part of that also moves to open source solutions. So what would really help to solve this issue is to actually shrink down the entire industry. I am not talking about some lay offs here and there, the team sizes for major installments have to be not more than 20-30 people. Even small teams of 3-5 devs can make amazing games. The tech and tools have become accessable for everyone these days. It is not so complex as it was back in the 90s anymore. @@cypherpleb
I'm fairly convinced Unreal has gotten to the point where developers aren't choosing between making their own engine or using an established one. They're choosing between making their own worse engine or using an established one that they lack the time and possibly even ability to create. Projekt Red has decided to swap to UE5, for example.
The issue is not their engines sucking. It has to do with the cost and time needed to create and maintain the tools and to train new hires in using them. But them switching to Unreal is quite strange considering how abysmal the documentation is.
It was a pr move, nothing more, nothing less. The game Cyberpunk when it released was a rushed buggy broken mess. So they decided to blame it on the engine, "oh our engine held us back, it was hard to work with. We'll be using UE5 next time! Get hyped because we said UE5!" Rember all those scam nft games that used UE5 as a marketing phrase, for better SEO and coverage. Like that space game. Any Dev team name dropping UE5 as if that guarantees qualify, or anything. Be wary, be sceptical.
Yeah it's definitely PR and people need to be prepared because RedEngine had a TON of optimisations over the years not present in UE5. Cyberpunk sequel will sadly be downgraded because after all the patches, CP2077 actually looks and runs better than every other UE5 game out there. @@TheMADmk
As a gamedev, I can't stress enough how insane this is and how much time and general "performance thinking" you can skip with lumen and nanite. These two features was science fiction 10 years ago!
That lumen reflection is insane, i have been looking for an alternative to planar reflections and filling the map with capture reflections that did not impact performance as much as ray tracing.
Yea I got into UE5 after seeing a superman download able demo in a city landscape and it was awesome. I just started building mario 64 type of lvls and it's fun for me to just mess around with it like a minecraft style game and learn from it like any other game/program. I connected peaches castle womps fortress royal raceway and a underground lvl together for fun in UE 5.2. In 5.3, I'm just gonna build and the idea will come into my head a week later lol.
@@anonymousinfinido2540 haha. I've been live streaming it on Twitch but haven't permanently saved any through UA-cam. Working on a 4th project for 5.3 but waiting for the right time after the WoW HC grind. Also the idea is rebuilding Stormwind in 5.3 Engine using default assets. I will start a bases this afternoon but it'll prob turn out horrid lol and stream it on Twitch.
The photorealistic aspect of video games is simply because they are the default assest in UE, most of them are free or easy to obtain, it is much more difficult to create a game with unique art like Valorant, where textures and artistic elements are created from scratch.
It's not the first time. That game Republic the Revolution had binary tree LOD that mostly has the same effect as nanite, but it requires fiddling with the 3D model.
@@cheeks_of_the_boreal_valleyStill waiting for good new turn-based JRPG's. Sea of Stars is kinda meh, story & characters are especially bad in most modern TB RPG's.
@@thenonexistinghero you are talking about a very specific genre that, to my knowledge, isn't particularly popular ( so you are not gonna have a lot releases, and even less quality ones). I was talking about games in a broader sense. I hope you can find something that pleases you in the future
@@cheeks_of_the_boreal_valley there have been some nice indie games but I just expect more from large AAA releases that charge a good chunk of change for the game
@@Scuba_Bro fromsoftware is AAA and we've got elden ring and armored core 6. Capcom is AAA and we've got street fighter 6 and Resident evil 4 remake. Tears of the kingdom from nintendo, and Baldur's Gate 3 from Larian which is a pretty big studio. I think big budget video games are doing great
2:46 these are the reasons why players somehow manage to slip below the ground and thus discovering bugs. Even way before UE, these kind of technique were used and loads of bugs were discovered/patched because tiny creases from stacked objects.
People say that the next version of Unreal Engine will be a game changer, but that doesn't really fix any of the Unreal Engine games not shipping in a complete mess on PC thanks to Epic not maintaining anything else with the engine that isn't visual features, stuff Fortnite needs, and virtual production stuff. How the hell Epic has not fixed asset streaming stuttering, Vulkan support, controller support (Sorry, only supporting XInput is not good controller support), and ultrawide support (Bad FOV zooming that even affects the editor) in the engine, even many years later is crazy. Any game releasing using UE is a red flag for something that's going to be an incredibly poor user experience outside of stuff from The Coalition and Tango Gameworks, seemingly the only ones that can actually use the engine properly. Pair that with Denuvo for Unreal dropping which aims to make it difficult to make mods for games using the engine, use a command console unlocker to adjust scalability settings the engine has (Thus defeating the purpose on why you'd use Unreal to begin with), amongst other things, and it's really scary seeing how AAA publishers are ditching their own custom engine tech in exchange for something more flawed and more homogenized (at least in the visual department). Epic's talking about how PC SSDs need to catch up to consoles, and yet nearly a year later, they still haven't implemented DirectStorage support in UE on Windows. Fortnite takes forever to boot on PC compared to PS5, and most UE games have some sort of asset streaming related stuttering issues going on (even when not dealing with shader PSO compiling). HDR display support in the engine still sucks ass. Can't even do non-photorealistic stuff in the engine and have that stuff not unlit (meaning, an emissive material that cannot have shadows casted onto it) without hacking together a 200GB compiled source code build of the engine, and that stuff as a developer will make it difficult to distribute official modding tools as per Epic's licensing agreement (Where you can only distribute engine tools to engine licensees). Plenty of games with mod support (like Pavlov or Deep Rock Galactic) use binary builds of the engine straight from Epic's launcher for this reason. If you're wondering why modding tools for UE games are crappy, that's why. Also, Epic opted for their own crappy Verse scripting language when game developers just wanted C# support in the engine, since their C++ documentation is infamously crap and Blueprints has a performance overhead (despite the amount of developers just using that nowadays). Might be an unpopular take, but despite Unity's hang-ups, I'd rather be making (and modding games) in that engine. It's simply way more modular and easier to work around issues. It's crazy to see Epic Games shills desperately attempt to defend the state of UE or the Epic Games Store right now, the ladder which is only usable if you use a third-party launcher called Heroic. Hopefully Godot becomes better and less jank, and Valve comes to their senses and releases Source 2 to the public (at least before Facepunch releases S&Box, which is supposed to be the Roblox of Source 2). We really need more competition in the game engine space.
Studios ditching their own engines is tragic because some of them are tailored to the kinds of games they made. CD Project Red's engine for instance had a really cool scene editor for animating and editing cutscenes and dialogue which they even showed off at GDC. And the issue is not even that the engines suck. Its that the studios waste huge ammounts of time training new hires, which might be gone again in a year seeing how low retention is in the industry.
The more we get these amazing tools to make a realistic, living world...the more we will understand that gameplay, story and craftsmanship are the core of games and not graphics. This is amazing btw, game changer!
if it looks gorgeous but when you poke and prod at it with the tools you are given, hardly anything acts in a realistic manner, it feels like a very expensive backdrop in a play, and one where a team of people agreed on what was the least they could get away with building. This is why games from 20 years ago where things were added because it was fun, tiny details a game like MGS2 is full of, still feel special. They had so much more 'game' in them and respected and rewarded player curiosity in a way we rarely see today. RDR2 is a modern equivalent perhaps, but somehow simulated so much of the mundane aspects of living in a world, but the reward there is different, you end up really being Arthur Morgan. I always felt like I was watching Solid Snake. Physical simulations and interactivity have been neglected in so many big budget mainstream games for so long, many big releases have great visuals and art direction in common. It should in theory make the more important elements of GAME design be the differentiator in appeal though which would be great and maybe encourage progress in neglected aspects of design.
This is it. The only takeaway from most games are the emotional impact of the story and gameplay. I’m excited for ue5 but it also seems like the industry has moved towards making Michael bayesque games with zero substance.
To all the young souls out there, don't rush to bet on it. I've been developing games in Unreal long enough to tell you that. The new Unreal 5 engine is indeed exceptional, the problem with it is that the Unreal community developed for years packages for version 4 mainly and when version 5 arrived it was like something completely different, most of the significant packages do not work with the new engine and this made it ineffective for creating games mainly. So, my recommendation to the young people here, let this engine develop for a few years (I would say) until this engine is able to do more meaningful things for games mainly and then start investing time in development
I just wish there were other engines out there that companies used besides mostly unreal. when every game uses the same brain, it feels the same on a lot of levels :)
There are, and they are just not as good and versatile as UE. Again that issue is not with the engine but with the devs. Final Fantasyt VII remake, Borderlands 3, Sea of Thieves, Valorant, Fortnite and Dragon Ball FighterZ were all made on UE, they all have different feel, style and gameplay. And those are just a couple of examples out of many!
@@TheBanthaPoodoo This is misinformation. Companies have always had the question of whether they should code something up themselves or get a dependency on some other coding team they cannot control. Here, the dependency just happens to be a game engine, but it's still code. When a company asks this question, they have to consider whether using a more generic, one-size-fits-all solution is worth it. The advantage is you ostensibly get to skip writing all the code for an engine (although if you tune your game enough, you end up writing a good deal of custom code anyway). The disadvantages are your game may not play nicely with Unreal Engine (in which case you end up writing a lot of novel code anyway and/or suffer from performance issues) and you don't have control over the evolution of the product (So let's say there is a game-breaking bug in the engine. You either have to have your developers deep dive Unreal Engine code and fix it themselves or beg the owners of Unreal Engine to fix the bug. If you have your own engine, leadership can send an email to the manager of the engine team, and the bug gets fixed. Or let's say you want the engine to be tuned to better support your style of game -- likely, they'll keep their genericity, and you either have to accept it or make a ton of custom code to transform Unreal Engine closer to something you need.). As an example, you can bet the engine behind GTA V was highly tuned to simulate a large, bustling city with missions and simulations of your character and driving and so on, and you can also bet leadership routinely connected the team making the engine to the team using it, so likely, the engine team had a to-do list that basically amounted to helping the team using the engine deliver the game with acceptable performance and fidelity and game features. If you like numbers, the simple fact is Unreal Engine will run slower for any random given game than if a team made an engine precisely to run that one style of game. That's an inarguable fact about code. It's always on the mind of a programmer to create a generic tool that somehow pumps out highly tuned and optimized code, but the sad fact is the fewer things you tune manually, the slower your code will be. With all that said, sometimes a business decides performance isn't a top priority or that the performance of the generic solution is acceptable or believes they can wield the generic code to solve their problem effectively based on what their needs are and the strengths and weaknesses of the generic solution. However, it just isn't possible for a well-written particular solution to lose out to a well-written generic solution.
Interesting, I didn't know that about some of the cartoony games. I loved ff7 remake a lot. I wonder what's with the devs then making so many games the same then :) @@TheBanthaPoodoo
The sad thing is so many devs use Lumen and Nanite as shortcuts to game development when they shouldn't causing games to perform like shit. There is a reason why CS2 still uses a lot of the trick versions of things like global illumination, water simulations, and reflections... because if you are skilled enough you can completely imitate the "real" versions with an extra digit to your framerate. EDIT: It's also the reason why CS2 has a fraction of the file size of most modern AAA games...
The thing I've heard about nanite though is that it has a higher performance floor but then doesn't really go much beyond that. Kind of an issue when you have a fairly older gpu or say the steam deck.
If you have an older GPU, just buy a newer one, because Nanite won't make your new games faster, actually they will be a bit slower in most cases. It is because with all the fidelity of new games, you'd expect your LODs to be pretty much high-quality already and have more polygons anyway as they count people would have at least 3080 RTX+. Nanite will just distribute these polygons more equally, but also will make them more detailed, so in certain situations when developers don't want to support old GPUs.. which in this case they probably should not support them, because they would purposefully make their game more ugly and old looking on the new GPUs. Atleast as of now I don't know of other solution that would be both beneficial to older GPUs and newer ones except of making 2 entire versions of the game, which nobody will do.
UE4 games only started looking like what their dev demos looked like towards the end of the PS4 generation/beginning of PS5 generation. So 5 years from now, development of games that started on UE5 will start to complete and by then hardware will have caught up as well.
The expectation for these to be used by multimillion developer studio is too high that i rather not expect at all from them.. It's just sad to see that all the game lately just sucks.. better making animation movies rather than a game.. i do hope some developer love their work so much that they would put their passion with this engine to good use.. 🙁
Bingo. Today's hardware isn't optimised for a lot of these rendering bells and whistles; most games can't even use this tech with adequate frame-rates on today's machines. Much better suited for movie productions and film/TV.@@BlueTricky
It seems way easier than before, and the performance is crazier as time progress. I remember a few years back, I had to spend days or weeks to optimize the model like crazy if I want it to run on most devices. Lightening is basically a nightmare; light baking and testing takes weeks to test it and optimize it for performance. Environmental lighting is the worst, need to do complex combination to make it right if go for real time lighting. RTX GPU isn't a thing. 50 % of the time is just for optimizing the model. I can imagine they are going to progress this much further with AI, like the software the IT guy in The Bloodshot, the one that Vin Diesel Starred in. The software basically can do all sh't by just typing and moving objects. It's like a dream come true, bcus AI will take care all the useless sh't and u do the creative stuff.
I am a game developer working in Unreal Engine 5 (v5.2). Currently, our team primarily prioritizes graphical quality, with less emphasis on performance optimization. Typically, we leave the task of fine-tuning performance to our graphics engineers, but this tends to occur towards the final stages of production. Unfortunately, this approach often results in inadequate time for the engineers to thoroughly assess all aspects of performance. Consequently, our games sometimes suffer from performance issues upon release.
I would be more worried about the actual game being fun to play. The teams I've worked with have been very small. So if we ever prioritized graphical quality the gameplay itself would suffer dramatically. Most people don't seem to realize these graphical tools don't really matter to a vast majority of indie devs that just don't have the manpower to implement them correctly. These things take time, and unless you want to spend 10 years making a game sacrifices have to be made somewhere or you better have hundreds of devs working on it. And if a game isn't fun, why bother?
@@NeonValleys Making game fun is very relative though. Focusing on graphics can be "fun" for a certain group of people. Why all the boomers watch TV instead of playing games? Because it is more realistic, that's one of the answers I get. And that's also why they would be more inclined to trying VR if that was more realistic and real-life like. Anyway, my point is creating a game by itself is difficult and fun is a very relative term which is hard to achieve for everyone involved. For a kid almost every game is fun, the older you get, the more you appreciate things like graphics since that is still something new and exciting, but gameplay has already been made in almost every form you can imagine, just worse graphics.
@@AldrazFun may be relative, but it still requires effort. Almost no thought or effort goes into the 'fun' factor of most major productions because like 95% of development goes into the graphics. This is why Nintendo games and small devs often do succeed in that regard since they actually take the time to create a good game feel.
you can make the game as graphically impressive as you want, if the gameplay is dogshit, and it runs like ass. No ones going to care. You want inspiration? Look to Valheim, Zelda, etc. Make a fun game, not a tech demo.
@@NeonValleys What are you talking about? Having your artists make a model with more detail doesn't somehow sap the resources of the coding team creating the systems that support the gameplay or "fun" of a video game. All the person said is their company has the people who compose levels and models emphasizes fidelity over performance -- not that the entire company stops doing their assigned role and jumps on the teams that create levels and models.
The good thing is, that UE is a renderer and a modeler as it is a platformer. That has always been their formula. You don’t see this in any other game engine. But what wasn’t in previous UE engines are being implemented into UE 5 and makes games look like a real-time playable movie than a movie or a game. Another cool thing is that UE5 can adapt with even VFX engines in real-time, which makes games look straight out of a movie, but it is all in engine.
I'm honestly not sold on it yet. Maybe this update is better but I recall someone doing a sampling to compare what Cyberpunk 2077 looked like in the red engine and how it would look in Unreal and everything looked flat, dull and just samey. Then they released that clip for the Matrix game and again...flat, dull. I did not like it. It feels like the introduction of 3D in the animation industry, it makes things easier to make so companies can save money at the expense of quality. Ideally I think the best case would be incorporating the technology into other engines, kinda like what we should be doing with AI now. That won't happen because of greed but on it' own I'm skeptical.
That's because half the things working in cp2077 just flat out break in ue5. It would ha e to be built back up. And when it is, it would look far better. You can't just drag and drop games between engines and thrn say obviously the one it's designed in looks the best. Fucking duh, of course it will. But if it were completely rebuilt the same in ue5, it would absolutely look superior due to the vast discrepancy in tech. This is just a bad faith arguement.
@@taylon5200 no it doesn't. It was the same in the Matrix Unreal demo. I wasn't talking about dragging and dropping anything, they just recreated a scene from the game to demonstrate how the cyberpunk aesthetic could look but sure let's use the Matrix demo then since it's a along the same lines. so now that its not a supposed "bad faith" argument what's your point? The issue is the rendering, it's the engine itself. It's about games having a unique look and feel that is lost when everyone just starts using a cookie cutter. That is especially the case with a genre like "Cyberpunk" that is heavily stylized, that aesthetic is at the core of what the genre is. Anyone who has watched the Matrix knows it's super stylized as well and you do not get that from the unreal demo. Every game will have it's own texture, look and feel. That is what I'm saying will get lost if it's just used lazily to save money.
8:10 It's like 3d movies. A lot of movies want to be 3d just to sell the 3D but have no effort to focus on 3d, than the movie isn't good in 3D or not better. Same with VR. Given opportunities don't get used how they could be used.
Unreal always stays true to it's name. I don't use the engine myself, I'm a Unity guy, but Nanite and Lumen alone make Unreal so ahead of the competition that it's actually insane.
@@igorthelight Fortnite doesnt use any of the graphical features its cell shaded lol. And Tekken isnt out yet so we will see if it looks half decent in game
so this technology or rather the base of it has been out for quite a while, just was never utilized like this in such a user friendly way before. For example a similar technique is used in most games for the character models, dx 10 had a demo of a similar technique on a xenomorph, terrains also do the same thing in some more modern engines the whole dynamic tessellation, i think its one of the gpu gems books that was released 10 ish years ago shows how to do similar terrain techniques that are shown here. Its only now that the hardware has reached the level that we can actually do it on a large scale
There is a difference in how it is applied though. Tessellation of terrain (for use as displacement through height map) is not new by any means, but the main issue is that it generates vastly more polygons than the flat surface with a planar texture would - and it's difficult to make it work with LOD when it's terrain, rather than an individual object mesh. The step forward here is that Nanite now supports terrain, so when you use tessellation, that too is affected by Nanite which means it scales linearly depending on the view distance, so it is only really present and complex close to the player, which drastically improves performance compared to old tessellation terrain.
@@Real_MisterSir Come to think of it it would work really well with VR and eye tracking. Hint only a small portion of your field of view has "high resultion". Sony are already doing something similar with PSVR.
@@TheAkashicTraveller Yea I had a similar idea where camera focus would be directly tied to level of detail -and nanite would make it even better. You could even rank it relative to what types of objects are important to keep detailed, and which ones aren't. Dynamic mesh like foliage, trees, etc could have higher ranking to avoid a jarring peripheral experience when transitioning in and out of focus, but other elements like ground and rocks could more easily get away with lower priority of detail. Lots of stuff could be added to optimize the flow of detail, beyond just plain distance.
And the best thing is, UE is free to use for everyone. You can pay for asset-packs, specialized features, but there is enough free stuff on the market, it's rarely going to be an issue.
Rendering hi-res meshes like that is really cool because there's no LODs and stuff, but the models file size are just huuuuuuge, especially for whole terrains. Even if the rendering is uber-optimized, there is a big problem with file sizes for large scope games. Pretty nice for small worlds and arena-type stuff, tho.
I don't care how cool the tree looks. I want to know if I can climb it. Cut it down or uproot it. The interaction your character has with the world makes you more immersed in it. If it's just a picture that looks cool, who cares? That's why Half-Life 2 was so amazing. The gravity gun allowed you to interact with and move shit.
Making great looking games easy is a good thing. Every big dev prioritizes making the graphics look the best it can because it sells, but whats happens once that's one of the easiest things to do? It's like how artists used to train for years to recreate what the eye sees, and then the camera was invented. All of a sudden these new forms of expressionism were able to grow because everyone didn't need to worry about recreating life anymore. This gain in productivity can be used to push forward physics and animation, which is already having huge leaps forward. Every step forward is one that doesn't need to be retread here. Solving this problem only leaves us looking for the next one to solve.
It’s probably the only field in the entertainment industry that tries to make their own tool for every product, like imagine if movies had to create their own video editing software for every different movie. And musicians trying to create new instruments for every new songs. This is why game development is actually still in its infancy, it’s still trying to build its own foundational tools (instruments) that would work efficiently while be as convenient as possible for any type of game they wanna make.
@@Jgvcfguy Just like construction companies will sometimes create their own tools, sure. But you know what I'm talking about don't just disagree for the sake of it lol, there's far more readily accessible and learnable video editing software out there than a game engine, it's the older and more established medium of entertainment which is why it has more instruments.
@@ryanspinoza6586i dont really get you point, no. Foundational tools like unreal, unity, blender, havok among many others? Theres a couple big names in the video production tools, but most is just small software of which you have a lot on the game engine side as well. They are continuously being developed on just like cameras and video editing software is. E.g. unreal engine is used in movie productions as well. And btw, you can download an engine of your choice right now and start building your own game, no problem. Im not sure what your point is, but building a game is quite complex, simple video editing software is basically included in bigger game engines as one small part of many. You know, since they simulate cameras etc. As for purpose built: there are considerations of performance which you dont necessarily have in books or video, also ip of course. You propably wont get rid of that, ever. Because a one size fits all will never be the best for everything.
The costs of building an engine and keeping it up to date with current tech is insane for most studios. It's no surprise so many upcoming games are now built on it.
Arguably most AAA studios will still use their own engine, because the pros significantly outweigh the cons in the long run for bigger studios with enough resources to develop their own. Regardless, this will still improve the quality of those games significantly because the new tech UE5 just showed provides an incentive for these other studios to 'step up their game' (pun intended).
"I think the reason Unreal games run like shit is because the devs concentrate on making the games look good not run good" 1000 percent. There's a bunch of great games that run great using Unreal but the look of a game is usually what makes you think it's Unreal and when you can look at a game and instantly tell it's Unreal it's usually not optimized at all lol
The reason why all unreal games will look similar for a while is that nobody has had time to create new assets for it yet. As studios invest time into new custom assets, games will start to look very different.
There's something about UE5 that Epic isn't bringing up often enough that I feel like they should. All of these improvements to the engine are quite simple to add or enable, which means even someone with very little experience can still make something look really good with the help of procedural generation. Make a ProGen map, go through and add your own little flairs here and there, then boom. What this also means is that devs can be more time on the more important aspects of a game, the gameplay itself.
The funniest thing among it all is that unreal is not just your regular game engine, it is a damn powerhouse, people here must have seen movies, series, and product commercials from perfumes to super cars, which were virtually produced using unreal.
Man, seeing all this UE stuff and who "easy" it seems to make great stuff with it, I can't help but kind of feel bad that I didn't get into game development as a kid... You can do so much with Unreal Engine, it's crazy!
Well honestly since UE4 to UE5 was kinda a big jump in every system, devs have to re-learn a lot of stuff anyway.. so its the right time to jump into it whenever you want. But.. it is far from "easy" to make a game in UE.. any game. You'll soon realize that these new features that Unreal is adding isn't gonna help you much when it comes to the overall time building this. It could even make things more complicated, because all these features also have their own negatives.
Nanite is 100% overblown. In projects where there's not a TON of stuff on screen, it will make the game run worse compared to not using Nanite. For example in my own project, It's all stylized art assets. The game runs around 90fps on my 2080super, but if I convert the entire project to using Nanite, the fps drops to around 60. Nanite is great for projects with a metric ton of stuff on screen when your natural FPS drops super low. Otherwise so far, it's absolutely not worth using.
of cource, instanced static meshes will always be more optimal as the engine only has to deal with one mesh even if you generate 1000's of copies. Nanite is great for things that need to be dynamic or need lod's. They do tech examples of making everything nanite because it's a tech example it's not suposed to be how you do it.
I've noticed that as well when trying UE5 - the overhead of Nanite essentially caps the game at 60fps on high-end GPUs. IMO its a cool art tool, but standard LODs and smart level design are better optimization tools in 99% of cases.
bruh. I've used nanite in real situations. The fact is, unless your game is HYPER realistic, it's not worth using. Every game has or should have LoD's period. It's not a new tech. My original statement still stands exactly as it was with no "bad faith". just facts and experience of using it first hand.@@taylon5200
waiting on unreal 5.3 nvidia caustics branch to come out in a couple months. that will be insane. working in 5.2 caustics for now but these 5.3 changes are massive.
This is perfect for the gaming industry and is going to propel AA and AAA developer teams. I am also sure we will see some indie devs using these new features from Unreal 5.3
It's also going to kill hardware. Right now with UE5 there's games where the minimum system specs are a $5-600 graphics card at today's prices. We're probably looking at 9000/6000 series cards from Nvidia and AMD before this is truly usable. In the old days it was, "But can it play Crisis?" Because nothing could. Well. A $1700 4090 can't do max settings at native resolution and get playable frame rates on some of these without DLSS, which adds input lag. So the question becomes, "But can it play Unreal 5 games.". Consoles? Don't even try for half of this stuff. It's going to get worse. PlayStation and Xbox are right out.
I'd like to mention that VDB's (Volumetric fire/smokes) are not game ready. (Yet) as of 5.3 they lag the hell out of a game with just one or two of these in scene.
There are many ways with billboards and meshes that you can solve the fireplace with current tech. Volumetric effects is of course the future because it is dynamic but it is extremely expensive. Not sure even next gen consoles like PS6 will be able to use them
You hit the nail on the head about the developers. [ last 2 minutes of video ] Right now ALL of the developers are "too busy" playing with the new toys that UE5 releases, and as such it'll eventually settle down and many will go back to focusing on the other aspects of game creation like Lore, mechanics, and other content that's crucial to a good to near perfect game being created.
They say that UE 5 boosts performance because of those polygons and stuff, ive heard that years ago when the matrix city demo came out. Some games are using UE5 now and did we see performance gains? No. Those games are very GPU demanding and the performance sucks.
Is not the engine issue. Is the time it takes the dev team to get used to the new engine. I imagine some stick to the already existing features of unreal 4 for the sake of not falling into the experimentation pitfall
I hope they can soon make it so that animated mesh can be reproduced with nanites, which means reducing 70% of the pipeline for creating a character without the need to create the character from scratch. With this technology, we can focus on storytelling and gameplay instead of the appearance, which consumes a lot of development time to complete the game.
I wonder if if these "nanites" used in Unreal 5+ are the same feature that I saw being developed when I got to take a tour of a game development studio back in college over 10 years ago. It's a similar concept, but seems like a different execution. They showed us a demo of creative tools in development which included infinitely scaling objects, made up of what they described as digital particles, that disregard LoD. It was even more nuts then, and I thought games were going to be revolutionized, but nothing seemed to come of it. I don't recall them having an association with Unreal either, other than using UE3 for some of their games.
Nah, voxels are just boxes as pixels in a 3D environment. The dev tool that I'm thinking of allowed the developers to basically build any object in-engine with digital particles, scale any object infinitely without loss of an object's visual quality, fully dynamic destructibility, and more. They could even make an object a building block of another object. @@digitalrandomart3049
The sad thing about nanite and lumen is that they are heavier than i originally thought, not as much as ray tracing though, and that nanite is much heavier than lumen, also the amount of VRAM needed is ridiculous, you need more than 16 to even be able to use it, i really hope in 5.3 it is better because i only have a mid range card with 12gb
Nanite isn't heavier as such in terms of FPS performance, but the problem is more older games up to like 2020 were made with less polygons, because they wanted it to be playable on GPUS like 2060 etc. Nowdays you would make about the same amounts of polygons that Nanite will give you as default quality. But that means it is like 3080+ GPUs. You can make Nanite use less polygons, to run well on older GPUs, but then you'd only hurt yourself as it would look ugly for other users. Unless they will make it a graphical settings in the future .. and I don't think it is currently, this won't be scalable. Also.. Nanite can make the game by itself like 10x more larger in terms of disk space, which kinda sucks, but it is better than taking away FPS or VRAM.
Meanwhile Zelda TotK has amazing and more complex physics and interactions than any game made with this engine has... on a console weaker than modern mobile phones.
@@Crazy09starkillor it is engine problem. Because epic promotes it as easy to use with a bunch of ready-made solutions and assets. But when optimization is required, it often turns out that it is easier to write your own solution from scratch than to fix everything that the engine generates. Epic sells it like easy solution what is false marketing.
To how nantite works in short think of turning the subdivide simple combined with subdivide Catmull-Clark amount from a static amount to a dynamic amount. Subdivide simple just adds more polygons to the object without reshaping it at all while subdivide Catmull-Clark smooths the edges by using more polygons to do so. Nanite most likely has logic to tell when an object is intended to be rounded or a sharp angle so can go from the insane amount of millions of polygons either made by the system or imported in a 3D model to the varying amounts using unsubdivide and resubdivide. You still will get a more true to what the artist wanted take on it going with hand made but Nanite allows you to have much better lighting with Lumen which forces you to use ray tracing ... kind of in 5.3.x there is a version that you can go back to the reflection spheres and use light maps and shadow maps instead of real time lighting. The performance improvement ranges in a wide range depending on the scene it can be as high as 2.5x improvement moving back to the traditional lighting system while using lumen for the light map.
Knowing Cyberpunk 2 is going to be based on Unreal 5, I think that's a solid one to look out for. But it may take a while (hopefully not 7+ years like the first one lol, but then they were also developing their inhouse RED engine at the same time which severely fucked their work pipeline and schedule).
@@Real_MisterSir As UE5 dev I can tell you that such migration from 1 entirely different game engine to Unreal Engine can be extremely difficult. Maybe they would be better off starting from scratch, but that could take even more than 7 years lol. Because that would include all 3D assets and animations, which most companies just take from their older game titles and customize / change. I think they will try to migrate everything they can, probably will manage to export 3D assets and even animations. And since RED engine uses C++ they could manage to export most of the code, but the C++ in Unreal is not standard, so who knows how difficult that can be. Either way, I think it can be done in a month if they are really smart and less in a year if they are bit dumb. But then it will take like 3-4 years to develop on top of that I would assume, because they need to make a lot of new content.
@@Aldraz Oh they definitely are starting from scratch, I would assume the only thing they will carry over are the assets already made for CP1, and as you say, if animations can be carried over that would be quite a time saver. But they announced pretty early on that they were switching to Unreal for all future installments rather than using their own RED engine - I guess it also helps take a toll off of their devs shoulders not having to both deal with building their own engine while also developing games on said engine. A little less freedom and control, but a lot of time saved and a pipeline that is easier to manage and set expectations for. It'll also be a benefit to hire new talent when Unreal becomes a standard engine to work with, so the training/adjustment process is minimal. Larger studios can more easily afford such expenses tied with doing everything inhouse. That said, it's been a while since they have made any public talk about the 2nd installment in the franchise, so of course time will tell if they deviate from their prior announcements or go through with it. My understanding is that CP2 is going to be entirely separate from CP1, as in they are making things from the ground up rather than iterating on the existing core framework of CP1, and I can understand how it may free up some future potential if they deem the current code and systems to be too much spaghetti to really develop the way they intend it, with potential limitations of the RED engine that can't do everything they want it to without pouring a lot of resources into that part, let alone the game itself.
@@Real_MisterSir Yeah I think you are right, completely agree with you. I believe they also thought it would be too much of a hustle keeping up with the latest trends, technologies when they directly tried to compete with Unreal Engine via their RED engine, which is bit of a madness and all props to them for even trying, but with how quickly everything is growing nowdays, they wouldn't be able to keep up with Unreal Engine that is powered by open source community + multi billion dollar company Epic games. And now that I think about it, maybe nobody will be able to keep competing in the long run. There has been many game engines in the past, now its almost as if it was only Unity and Unreal. If Unreal had a great 2D support, it could even beat Unity.
As someone who's been working on a game on Unreal engine 5.2, nanites truly are amazing and help so much with frame rate. I've got thousands of grass, trees and vines all swaying and affected by wind without a dip in frame rate, I love nanites XD
asmon is totally right, as an indie game dev myself, we switched to unreal engine from unity to increase productivity and save time to make great games, but we are still using 3D lowpoly models cuz of the art style and to reuse assets from prior games ofc
@@robservin7664 so far, we didn't produce many succesful games yet, where we even had to switch platform from mobile to PC so for PC, currently we only have a "Vampire Survivors"-inspired game with some twists here and there in 3D Scifi setting, it is called "Defend Earth: Xenos Survivors", just recently we have pushed a free version, where people can test the game before buying the main game, same name just with a "...- Prelude" at the end the new game on the unreal engine is still in development, so I can't tell anything yet :D ty for being interested btw!
@@spacejunk2186 I am not a programmer, but what I got told from our team, that unity is more friendly for new programmers, since C# is easier to learn while in Unreal, it is C++ and you really have to be very precidse there, but in unity, you have to write the code from the beginning, while in unreal you have blueprints which are basically rdy-to-use-scripts so if you have no coding experience, then unreal is much better, cuz you can build games via blueprints, but even if you are a programmer already, making games via blueprints is way quicker, we basically doubled or even tripled our production speed since we don't have to write our own code anymore, and also can use paid plugins to create procedural levels e.g. we are currently in the 8th production week and after additional 1-2 more months (depending on the polish grade), we would be rdy to launch our game in Early Access
I remember back in the day it was unreal tournament vs quake 3 arena and both were great games but quake and Carmack were always able to beat them out. There was a whole generation of games built on the q3 engine. It was the standard. But my God the tools we had at the time, ie radiant were so cumbersome and hard to use. Took a lot of learning and trial and error. Drawing brushes by hand, which makes what they were able to make with that engine incredible. Alice comes to mind. Makes me wonder what Carmack is up to cause he always has tricks up his sleeve. The last innovation from him I remember hearing about was megatextures in rage. This is the biggest jump I've seen since crytek. The original engine was amazing. I remember seeing the editor demo and being able to just paint terrain that automatically populated with foliage based on the height map was crazy. You didn't even have to compile to test you just hit a button and it dropped you into the map. With old editors you had to render bsp's the load the game then load the map through a console command. Ugh.
Yes by making good graphics more accessible by small devteams big aaa companies will actually have to compete with them on good story telling mechanics and gameplay. Since there biggest strength is gone.
Thanks for the video, love all things UE. I work in UE and love the tech. The one thing I want to comment on is this constant echoed idea that 'all games are going to look the same because it's made in Unreal'. That all UE games look identical. I can't believe how shortsighted that is. It's all about he dev team and the art team. What style are you going for? Who's art directing the team? Can the environment, character and prop artists match the director's vision? It's the same with a film--can you imagine saying, well heck all films just look the same because they are shot on *blank*. No, it's the art team. UE can do anything you want to do. Please everyone understand that. Basically, Az you nailed what I said here the last segment of the video. Cheers love you all!
Cant wait for hundred of games that look exactly the same with the most plain art direction ever. We got to the point where stuff like UE is not necessary anymore all we need is decent art direction and to use this technology to help but most AAA studios wont bother and just abuse those features.
As impressive as this is (and it's really damn impressive), but frequently unreal has performance issues and idk if thats a game dev issue or an unreal dev issue. Does anything in 5.3 address this? Or is it all really pretty new tools?
It's a bit of both. I work with Unreal and most of the time it is a developer skill issue. We have a few people on the team with over 15 years of experience, and they can usually give us great advice. But you also have to keep in mind that this monster of a codebase is changing at breakneck speed. Most projects pick the most recent version when the project is started, and are retiscent to upgrade unless a new feature will be very useful
I’m a Senior 3D Character Artist working in games for over 13 years now and Unreal is the best engine to use, it’s also the easiest to use bc of the tools. I like that it can handle actual has instances instead of haircards, it is pretty expensive though but as hardware will catch up it will be the best way to render and simulate hair both for the artist making it and the person watching or playing a game.
They put out these update fucking fast and just act like it’s fucking nothing. Like the people using this must only just start getting the hang of it then a whole bunch of new shit that blows the older stuff outta the water comes along. It must get a bit annoying at how quick they are
Virtual production with UR 5.3 will replace a lot of Hollywood VFX teams. No need for setting up lights anymore when you can tell where a projected one should come from equally reflecting from the subject at that same angle the lighting was added which was always the most difficult part in CGI. Lighting on skin tones.
I'm happy to see people thinking about how Unreal has the potential to make games look too same'y. It's been a long standing complaint from smaller developers and one of the reasons why people often use Unity instead. My day job is to modify the UE5 renderer so that we can have a stylized game. It requires a LOT of knowledge and it's very difficult to do well. Big studios that have graphics and rendering engineers with experience creating renderers from scratch will be the studios that have the knowledge necessary to create stylized UE5 games. It's going to be very interesting to see how different studios make their games stand out. It's fun to experiment with mixing stylized rendering and UE5's photo realistic features like fog and volumetric lighting.
The impressive thing about the unreal team is that they make it very accessible to indie developers
the bad thing about unreal engine is that the tutorials are very limited compared to unity. It looks simple but once you dive deep into the systems it's very complex to do by your own.
Nice........
@@MicroMintyHourglass🙃
@@MicroMintyHourglass It might sound that way but unreal devs are few and far between, and teams often have to go out of their way to employ a dev that knows how to use unreal properly. It's great if you want to be a well paid software engineer, but if unreal was better documented then we would have way more indie studios using it.
@@allenantony6924 That being said game dev is stupidly complicated to begin with and has a enormous amount of failures. You're going to be putting in the hours anyway so why not use an engine that not only can get the job done but has many offers for you now if you can?
Im a lead VFX artist and the volumetric explosions is interesting but very expensive. There are tons of tricks one can use on billboards to make them sit in the environment such as normal maps and dedicated light calculations. Volumetric for fog is super useful and have been used in games for easily 10 years. I think volumetrics for VFX will only be useful in a few cases with current hardware since it will suck up all resources.
I definitely see volumetrics as an aspect of vfx that ai acceleration can help with to overcome some of those resource intensive calculations, especially when light scattering is also a metric to consider.
You could use volumetrics for VFX in important and meaningful moments that the player is sure to pay attention to and not everywhere else where it wouldn't make a difference, it's the best use case i see until the performances are unimpactful in the future
Sucks for them. Those are the first things I always turn off. Useless post processing fx are the worst.
I work with offline rendering, and can only agree. I believe volumetric objects will definitely have their place (for dynamic effects or volumes that need to be spawned when needed, or even things like the fireplace examples where the camera can freely move around it), but billboards or sliced volumetrics with normal map relighting will definitely live on for the foreseeable future due to the memory and performance restraints that we currently can't work around.
And to elaborate on what you're saying Andreas: Volumetric fog is a completely different technique compared to sparse volumetrics. (as far as I've understood, I'm not a coder or touch back-end in any way) Volumetric fog can afaik be "faked" with a lot of clever calculations based on light position, depth mattes and vector math.
@@pinolskun8764 I believe the OPPOSITE is true. For important and meaningful moments, the effect can be precalculated and will most likely have fixed lighting setups, meaning that you can use billboards instad as the lighting doesn't have to be dynamic or be viewable from any angle.
This actually looks incredible to me. We just need people to build fun games that use these systems.
Cant wait for the next 500gb game that features UE 5.3. Surely it will be optimized for the average user and not be exclusive to certain gaming rigs and not problematic at launch coz of how demanding the graphics will be. Woww amazing time
Yeah, because even tho we just started with UE5, some games just only give a shit about graphics, but the gameplay is horrendous. Hopefully, that's changes now, I want a balance between em
Its hardly different from LoD, just doing it dynamically, which will be the source of endless amounts of detail fracture.
Imagine instead of having a fixed distance at which the detail changes, to instead every object randomly changing their detail as they please.
Donno what Asmon is getting his knickers in a twist about, or you KitCat. This option can easily become the most immersion breaking, texture fracturing thing in existence.
The only thing to look forward to here, is how bad it will get.
@nonusbusinissus5632 you say all this with no evidence of how well it will really work lol. Stop
@@soulofartorias9928 If you could comprehend what you are being told, the video literally told you what i did.
LoD -> Fixed distance, Nano -> Every object changes complexity individually.
Asmon and every each one of his viewers seem to suffer from a case of room temperature iq.
All you understood from the entire thing was the pictures.
Make sure to clean the drool from your keyboard once you are done watching this.
In LOD: the designers have to make all the versions of the model, so they make a high def model for close range and 2 or 3 other models for when the model is far from the player
In Ninite: you just make 1 high def model, and the engine will change the polycount dynamically, even when you're close to an object, it only shows you the part you're seeing in high def, the back of the model will be low def.
Having used LOD generators myself (like DynDOLOD for Skyrim VR), I doubt game devs are manually creating LoDs. Nanite just skips the LoD generation step but seemingly at a performance cost.
Perfect explanation! thanks!
Correction: devs use dedicated LOD tools to create lower-LOD models, often automatically and in a batch.
Nanite could save space due to only needing one model... but more realistically devs will just push higher LOD models and game size will increase.
Nanite could also improve performance as it can have pixel-perfect LOD switching, whereas the classic way leaves gaps in either performance or graphics between switches... but more realistically devs won't optimize their Nanite implementation for performance and games will skyrocket in terms of requirements.
It doesn't support backface culling?
Most retopology is done quite well these days; Nanite has yet to prove it can work at scale. It looks great in confined spaces with limited fulcrum rendering, and in the isolated cases it was used as a test demonstration in Fortnite, but it remains to be seen how well it works (when optimised, like you mentioned) on larger scale projects.@@sujimayne
cant wait to play games using this engine in 10 years
Someone will use engine in few months, in a year or so... still don't have anything close to UE3 tech demos from more than a decade. CGI and in game/engine cinematic do not count.
And upgrade your pc
Indie games will come out much faster than that. 2 years ish
@@Wes-Tyler I'm not sure about that, it is much easier and more profitable for indie developers to just keep making vampire survivor clones. or whatever the flavor of the month is.
2 games with ue5 are already out wtf are you talking about, 2-3 more are on track to release this or next year
I absolutely love the tech jumps in Unreal Engine but I also feel sad about the fact that their realistic asset store will most likely extremely lower the amount of large game productions with their own distinct styles
Large studios tend to make all their assets inhouse so this will not really change anything for them, they'll just 3D scan real world objects like normal and import it as 3D nanite mesh into Unreal. The existing megascan asset library within Unreal is catering to Indie first and foremost, to the people who can't just go out and 3D scan 1000s of environments themselves. But AAA studios absolutely will, they been doing it for years -besides also directly modelling what they need for their own styles.
@@Real_MisterSir AAA will cut as many spendings as they are able to, if they can skip creating assets thanks to Unreal ones they 100% will
@@ikcikor3670 Key words "able to". If their games start looking like what some 3 man indie studio cooks up in a year then they will start losing money and fast = unable to
@@rykehuss3435 indie studios don't have the money to put as much or more into marketing than into the game itself
@@ikcikor3670 They won't skip creating assests because of Unreal. They could, but it would look bad and wouldn't sell.
18:00 This engine powers games such as Fortnite and Valorant. While these games appear distinct in style, the choice of aesthetic largely depends on the Art Director and the development team. A challenge in the industry is the prevalence of affordable models available for purchase, which often have a similar appearance. While these models expedite prototyping and reduce production costs, they can lead to a homogeneity in game visuals. This trend might result in games looking more alike than the influence of engines like Unreal Engine 5.0.
As a dev working on a game in ue5 5.3, the changes are welcome but things like nanite foliage/nanite landscape and tessellation and sparse volume textures are EXTREMELY expensive. Dont expect games to makeb big use of them soon
Expenseive for hardware to render? I'm a fullstack web dev, but I'm planing to learn UE5, seems a very good framework, have a good job
Yeah, there's fantastic plug-ins that solve the issues a little better
Expensive as financially or expensive as in computing power?
@@overlord165 computing power, they're features baked into UE5 lol
with nanite there is actually a bit of a caveat , if your landscape is a city and your surfaces are mostly flat then it's doing a lot of processing for relatively little return at that point you are better off without it in certain circumstances however for forests and more natural type stuff what happens is that there is a slightly higher processing demand but it scales much better and way way more than what they usually show in these videos
The good thing is that not everything has to be nanite, you can toggle each element independently, but I would say for any world where you have even just a hint of distance variance from objects, nanite is the way to go rather than LODs. LODs also take up more package space than a single high-res nanite mesh+texture.
@@Real_MisterSir thats pretty much what you do, use nanite on the objects that benefit most from it, you can also turn it on and off at runtime as well but then you can loose some of the advantages of keeping it running on those objects, realistically every landscape or map is different and should be tuned properly
Devs won't do this; that is extra development time. Nanite games will end up being wholly Nanite with poor implementation and performance will suffer, while game size will increase.
A story as old as game engines themselves.
@@sujimayne partial nanite usage is successfully being done now Satisfactory for example they only use it for rocks atm
@@sujimayne Nanite will very likely reduce the size of games since multiple lod meshes and textures maps are no longer needed to represent the same object.
Every time I see a new UE update I’m impressed by the tech, then I remember there won’t be a single game that looks as good as these tech displays
Some games using UE 5 with Nanite are already out, and they look pretty good as long as you have a 3080 or higher to run them. Immortals of Aveum and Remnant II are two examples.
You mean *my hardware is too garbage to run any cutting edge Unreal Engine graphics.
@@AG-ld6rvThey don’t looks very good considering how much graphics power they require to run. Immortals has terrible stuttering even with the best of cards. Is actually quite surprising how badly it ran and looked considering all the UE5 hype. Haven’t played Remnant II but Immortals made me very sceptical. It looks good, but not incredible by any mean even on the best cards. Plus it was a very linear game with a short campaign that stuttered so badly. Imagine then how open worlds would work with UE5…
@@pressrepeat2000 Stuttering is a separate issue that good programmers using UE can fix. That's like saying mechanics are bad, because you took your vehicle to one once who broke your vehicle further. That's just a bad mechanic among many mechanics, and a good mechanic will fix your vehicle up. I agree that any game with stutters needs more optimizations like shader precompilation implemented.
As for the looks of these two games, they look really good to me on maxed settings and on PS5/series X. Series S looks like dogshit though for Aveum.
As for the common "optimization" complaints, this is as old as 3d video games themselves. What happens is the programmers of a game dial up the graphics such that, to run the game on max settings, you need the absolute best hardware out there. So basically, every single new game that has great graphics as a selling point will run like shit on outdated hardware and sometimes even on relatively recent hardware that isn't the best on the market. That or the settings to get the game running totally obliterate the graphical fidelity of the game.
The companies have as a goal to create a game that looks amazing and runs all right on consoles and on expensive PCs. Unfortunately, I do not believe it is possible to optimize high fidelity games for the myriad of hardware combinations out there in the PC market. Examine the look of Aveum on series S for an example of how optimizations, even with the hardware fully known, cannot overcome the limitations of a cheaper system for a game using some of the most expensive and cutting edge techniques to create the highest fidelity games possible.
The simple fact is that Nanite is a technique that is more expensive to run than the older "level of detail" approach. Any game that uses Nanite will naturally have higher minimum and recommended hardware. They'd have to code the game twice, one with Nanite and one with LoD, to improve performance on what is, when it comes to running Nanite, insufficient hardware. There are base limitations on what can be done, and even with intense optimizations like likely occurred with Immortals of Aveum for series S, the game is simply going to look horrific and still run poorly. It's not like the company is just trying to bone people with cheap PCs. Nanite just takes a certain baseline of performant hardware to achieve.
The fact is you need to run some kind of AI upscaler and put up with the massive amount of artifacts if you have a cheap PC but insist on playing games that, in terms of graphics, push the limits of fidelity as much as possible, and UE5 was basically made to make games that can push those limits. An example of that is Nanite as I discussed in my previous paragraph.
I'd say that Fortnite is likely the most runnable game out there that uses Nanite, but that's only due to the game creators not having "has the highest fidelity possible" as a core requirement of the game and instead has "we want this competitive shooter to run on as many PCs as possible" as a core requirement. Naturally, the graphics in the game aren't that stunning due to this design priority. Still, Fortnite would run better on even more PCs if the game didn't use Nanite and instead used the level of detail strategy. Like I said, Nanite simply costs more to do. You can't have your cake and eat it too. If a gaming studio wants to maximize the number of PCs that can run its games, they've simply got to avoid using expensive features in UE5 like Nanite. But they likely secure more sales exciting people who own a PS5/Series X/decent PC with Nanite than they'd get not exciting those people just to have an additional, small segment of the market able to run the game decently (while looking horrible).
@@FightClass3 for just graphics yes, but your hardware will be able to run say a gta 5 style game with 200 times the content that is interactive and that opens massive doors to endless madness.
For anyone who doesn't know is that this cuts down on a LOT of production of the same asset that needs to be remade into different levels of detail or LOD. The engine itself does the hard work for you by changing the polygon count on the objects that are being rendered.
Sort of; most really good artists can have the retopology done on a LOD0 asset in a day or two. The question is, is Nanite better optimised for a wide variety of platforms than say a LOD3 asset?
It comes at the expense of having an actual polished and functional game on medium-end hardware. The use of nanite technology will not have any real impact on game development, and if anything, games will take even longer to make and run worse because of the taxing nature of nanite technology. It bloats the games and makes them require insane specs like Immortals of Aveum and Remnant II. Both are great examples as they run at 720p on current gen consoles because of terrible optimization, bloated use of fancy features and the results are mediocre in terms of visual fidelity and artistic direction. Just look at how gorgeus The Witcher 3, Elden Ring, Armored Core 6... etc look on the same hardware and running at higher res with higher frames by the use of smart development and proper optimization.
Nanite technology is a great tool to use in certain situations, but as with FSR and DLSS, it should only compliment the development of video games, not make the industry over reliant on said features to make their games barely run.
Dude, Unreal Engine Dev can update their software to better BTW@@praisetheoak
Another way to look at it is that the easier it gets to make a good looking game with tools like this, the more time can be spent on other aspects, like gameplay mechanics and story - of course, the issue is whether developers will actually use that time, or just push out more crap with a make-over in a shorter time frame.
They probably would, if publishers got off their back for once.
@@ablationer Better and easier to make graphics will just make every game look the same...Soulless
They now have more time to design a more addicting gacha mechanics.
This with AI Npcs is going to make gaming absolutely amazing in the future
i cant wait for AI npcs
Yep but imagine the graphics cards and processing we will need
@@natevoid9955well everything in tech improves then becomes efficient that’s just what I’ve seen in my life. First the photorealistic imagery then the improvement in efficiencies
@@natevoid9955 We are increasing fidelity at lower resource demand, so it's a beneficial scale over time. If graphic demand grew exponentially higher than the fidelity improvement, it would be a problem long term, but in reality we are seeing the opposite. Some things are even running more efficient now than before, so we simply have more headroom than we used to, which incentivizes to make use of that headroom (and maybe push a little beyond).
@@natevoid9955as it stands right now hardware is years further than software. The reason games run bad today is absolutely dogshit optimization. Most gaming pcs can run UE 5.3 given that developers did well on the optimization
05:39 thats the reason why alot of unreal engine 5 games feel the same. Atleast indie games.
I am not impressed until i see this actually being used and implemented into the games we play. It's like on those amazing news you hear, how this is going to revolutionize whatever and then you never hear about it ever.
It's not even out yet so we won't see games with if for over 5 years probably. Well big AAA games anyway.
There's a lot of games that have Nanite + Lumen already, but you can't notice it that well, because most faked lighting looks as good as dynamic GI and when it comes to LODs, you also can't see the difference pretty much. I believe Fortnite has both Lumen and Nanite for example.
i agree with you fully, the only game that ive seen using this technology is, surprisingly, Dead Island 2. the environments and lighting in that game was immersive and beautiful. the hollywood hill houses that you run through were breathtaking.
There are some Unity games that can use lumen with mods. Runs like 1/4th the normal fps, but it is there and made by a regular dude.
They could always update the games tech like Cyberpunk and Warframe do.
There's some videos out there showing the evolution of The Unreal Engine...stuff we're seeing today was only a dream 5, 10, 15+ yrs ago...its somewhat sad many take for granted what they see today and don't fully appreciate what's out there now; Unreal or any of the other Engines....
18:00 the fact is its all comes down to art style you can make a game look any way you like its up to the devloper to design the game, if they look the same as other games then its because they chose for it to look that way.
I swear every unreal version gets hyped up like this then all we get is cheap asset flips spamming the steam store.
Game developer here. What most do not understand is, even with all this fancy tech, this is only the visuals. If you consider everything else from the game, the performance drops big time. And did I forgot to mention shader compilation? Yeah. This issue still persists. It is all sugar coating at this point, Unreal cater more towards the VFX movie industry than to games at the moment.
and it's why we've gone backwards to a huge degree in world simulation, immersion and interactivity almost constant decline and silent removal of aspects of gameplay, often small things, physics, environment destruction, items that have no purpose that you can play with or effect in some fun way.
So many titles from 15-20 years ago made things other than GFX higher priority. Gorgeous but 'dead' world we see today, aside from maybe RDR. This is the aspect of Bethesda games people gloss over all the time. The impact of game wide object tracking for hundreds of hours has on what else they can do.
Was the rise of middleware like Havok partly to blame? How many studio's bother to create bespoke simulations today?
It's going to the few studio skilled enough to persevere with their own custom engines that stand out amongst the homogenized UE5 decade ahead.
Physics is a different topic but I do understand your point of view. Most modern games tend to focus on gfx instead of a believable world that is filled with interactions and different paths to go. I am not agreeing we go backwards, but it clearly shows how much effort is wasted on photorealism. @@cypherpleb
@@eth7928 how much effort it's taken by the megacorps, how fake the idea of games journalism must be to normalise the market to just accept worlds (and NPC AI) that behaved like Far Cry 2 did on the Xbox 360 in 2008, or MGS2 on Playstation 2 in 2001 basically hardly ever be surpassed by most big flagship releases even today on PC's with supercomputer specs vs the limitations those titles were built within.
Say what you want but fixed limited hardware specs show the industry who the real technical and creative geniuses are. Few games even attempt to push what's possible today, perhaps only PC exclusive sims such as Arma, Fight Sims etc... in theory Star Citizen? and they don't feel like games to me anyway.
Most other games, especially anything from a public firm simply wouldn't be allowed to leave millions of customers on the table. Lowest Common denominator led game design is what we have. Mid range laptop APU's from 2012 still constrain nearly every multiplatform game you play on you £4k PC in terms of design.
Big studios became lazy and too big. A big studio has too many positions of staff where ideas go through and that complex management results in mediocre games. Look at D4. 9000 Devs and its not coming close to the more than 20y old D2.
I absolutely believe that a small team of people can do amazing games if they had the same budget as AAA title. Studios have become so big that risky investments are off the table. This is why you won't see new IPs that much and no revolutionary tech in video games. The tech now is being developed by 3rd party contributers such as Unreal or Unity. Part of that also moves to open source solutions.
So what would really help to solve this issue is to actually shrink down the entire industry. I am not talking about some lay offs here and there, the team sizes for major installments have to be not more than 20-30 people. Even small teams of 3-5 devs can make amazing games. The tech and tools have become accessable for everyone these days. It is not so complex as it was back in the 90s anymore.
@@cypherpleb
I'm fairly convinced Unreal has gotten to the point where developers aren't choosing between making their own engine or using an established one. They're choosing between making their own worse engine or using an established one that they lack the time and possibly even ability to create. Projekt Red has decided to swap to UE5, for example.
The issue is not their engines sucking. It has to do with the cost and time needed to create and maintain the tools and to train new hires in using them.
But them switching to Unreal is quite strange considering how abysmal the documentation is.
It was a pr move, nothing more, nothing less. The game Cyberpunk when it released was a rushed buggy broken mess. So they decided to blame it on the engine, "oh our engine held us back, it was hard to work with. We'll be using UE5 next time! Get hyped because we said UE5!" Rember all those scam nft games that used UE5 as a marketing phrase, for better SEO and coverage. Like that space game. Any Dev team name dropping UE5 as if that guarantees qualify, or anything. Be wary, be sceptical.
Yeah it's definitely PR and people need to be prepared because RedEngine had a TON of optimisations over the years not present in UE5. Cyberpunk sequel will sadly be downgraded because after all the patches, CP2077 actually looks and runs better than every other UE5 game out there. @@TheMADmk
Btw it's game dev fault not unreal engine fault@@billywashere6965
As a gamedev, I can't stress enough how insane this is and how much time and general "performance thinking" you can skip with lumen and nanite. These two features was science fiction 10 years ago!
That lumen reflection is insane, i have been looking for an alternative to planar reflections and filling the map with capture reflections that did not impact performance as much as ray tracing.
the ability for building of skeletons and weight painting within UE was it for me, that is so awesome!
I mean, ssr, but i guess you dont like that
The optimization of Unreal Engine is literally unreal. The games look great and run great, at least if the dev doesn't spam assets out...
Yea I got into UE5 after seeing a superman download able demo in a city landscape and it was awesome. I just started building mario 64 type of lvls and it's fun for me to just mess around with it like a minecraft style game and learn from it like any other game/program. I connected peaches castle womps fortress royal raceway and a underground lvl together for fun in UE 5.2. In 5.3, I'm just gonna build and the idea will come into my head a week later lol.
Will wait for your video upload?
@@anonymousinfinido2540 haha. I've been live streaming it on Twitch but haven't permanently saved any through UA-cam. Working on a 4th project for 5.3 but waiting for the right time after the WoW HC grind. Also the idea is rebuilding Stormwind in 5.3 Engine using default assets. I will start a bases this afternoon but it'll prob turn out horrid lol and stream it on Twitch.
@@Ragingdevils cool
My therapist: "its a game it cant hurt you"
The guy form the UE 5 video: "this fire is real and it exists in our world"
The photorealistic aspect of video games is simply because they are the default assest in UE, most of them are free or easy to obtain, it is much more difficult to create a game with unique art like Valorant, where textures and artistic elements are created from scratch.
It's not the first time. That game Republic the Revolution had binary tree LOD that mostly has the same effect as nanite, but it requires fiddling with the 3D model.
Now if only game devs could get creative again and make solid games
We've had plenty over the last years, you just need to look outside of trashbin that is Blizzard, EA, Bethesda and company
@@cheeks_of_the_boreal_valleyStill waiting for good new turn-based JRPG's. Sea of Stars is kinda meh, story & characters are especially bad in most modern TB RPG's.
@@thenonexistinghero you are talking about a very specific genre that, to my knowledge, isn't particularly popular ( so you are not gonna have a lot releases, and even less quality ones). I was talking about games in a broader sense.
I hope you can find something that pleases you in the future
@@cheeks_of_the_boreal_valley there have been some nice indie games but I just expect more from large AAA releases that charge a good chunk of change for the game
@@Scuba_Bro fromsoftware is AAA and we've got elden ring and armored core 6. Capcom is AAA and we've got street fighter 6 and Resident evil 4 remake. Tears of the kingdom from nintendo, and Baldur's Gate 3 from Larian which is a pretty big studio.
I think big budget video games are doing great
2:46 these are the reasons why players somehow manage to slip below the ground and thus discovering bugs. Even way before UE, these kind of technique were used and loads of bugs were discovered/patched because tiny creases from stacked objects.
Waiting for terabyte size games. No need to optimize models and textures it will render just fine and fast
People say that the next version of Unreal Engine will be a game changer, but that doesn't really fix any of the Unreal Engine games not shipping in a complete mess on PC thanks to Epic not maintaining anything else with the engine that isn't visual features, stuff Fortnite needs, and virtual production stuff. How the hell Epic has not fixed asset streaming stuttering, Vulkan support, controller support (Sorry, only supporting XInput is not good controller support), and ultrawide support (Bad FOV zooming that even affects the editor) in the engine, even many years later is crazy. Any game releasing using UE is a red flag for something that's going to be an incredibly poor user experience outside of stuff from The Coalition and Tango Gameworks, seemingly the only ones that can actually use the engine properly. Pair that with Denuvo for Unreal dropping which aims to make it difficult to make mods for games using the engine, use a command console unlocker to adjust scalability settings the engine has (Thus defeating the purpose on why you'd use Unreal to begin with), amongst other things, and it's really scary seeing how AAA publishers are ditching their own custom engine tech in exchange for something more flawed and more homogenized (at least in the visual department). Epic's talking about how PC SSDs need to catch up to consoles, and yet nearly a year later, they still haven't implemented DirectStorage support in UE on Windows. Fortnite takes forever to boot on PC compared to PS5, and most UE games have some sort of asset streaming related stuttering issues going on (even when not dealing with shader PSO compiling). HDR display support in the engine still sucks ass.
Can't even do non-photorealistic stuff in the engine and have that stuff not unlit (meaning, an emissive material that cannot have shadows casted onto it) without hacking together a 200GB compiled source code build of the engine, and that stuff as a developer will make it difficult to distribute official modding tools as per Epic's licensing agreement (Where you can only distribute engine tools to engine licensees). Plenty of games with mod support (like Pavlov or Deep Rock Galactic) use binary builds of the engine straight from Epic's launcher for this reason. If you're wondering why modding tools for UE games are crappy, that's why. Also, Epic opted for their own crappy Verse scripting language when game developers just wanted C# support in the engine, since their C++ documentation is infamously crap and Blueprints has a performance overhead (despite the amount of developers just using that nowadays).
Might be an unpopular take, but despite Unity's hang-ups, I'd rather be making (and modding games) in that engine. It's simply way more modular and easier to work around issues. It's crazy to see Epic Games shills desperately attempt to defend the state of UE or the Epic Games Store right now, the ladder which is only usable if you use a third-party launcher called Heroic. Hopefully Godot becomes better and less jank, and Valve comes to their senses and releases Source 2 to the public (at least before Facepunch releases S&Box, which is supposed to be the Roblox of Source 2). We really need more competition in the game engine space.
Studios ditching their own engines is tragic because some of them are tailored to the kinds of games they made. CD Project Red's engine for instance had a really cool scene editor for animating and editing cutscenes and dialogue which they even showed off at GDC.
And the issue is not even that the engines suck. Its that the studios waste huge ammounts of time training new hires, which might be gone again in a year seeing how low retention is in the industry.
The more we get these amazing tools to make a realistic, living world...the more we will understand that gameplay, story and craftsmanship are the core of games and not graphics. This is amazing btw, game changer!
if it looks gorgeous but when you poke and prod at it with the tools you are given, hardly anything acts in a realistic manner, it feels like a very expensive backdrop in a play, and one where a team of people agreed on what was the least they could get away with building.
This is why games from 20 years ago where things were added because it was fun, tiny details a game like MGS2 is full of, still feel special. They had so much more 'game' in them and respected and rewarded player curiosity in a way we rarely see today. RDR2 is a modern equivalent perhaps, but somehow simulated so much of the mundane aspects of living in a world, but the reward there is different, you end up really being Arthur Morgan. I always felt like I was watching Solid Snake.
Physical simulations and interactivity have been neglected in so many big budget mainstream games for so long, many big releases have great visuals and art direction in common. It should in theory make the more important elements of GAME design be the differentiator in appeal though which would be great and maybe encourage progress in neglected aspects of design.
This is it. The only takeaway from most games are the emotional impact of the story and gameplay. I’m excited for ue5 but it also seems like the industry has moved towards making Michael bayesque games with zero substance.
Yep, totally agree
That is just bad dev making bad game Okay. It is not Unreal Engine fault@@cypherpleb
To all the young souls out there, don't rush to bet on it. I've been developing games in Unreal long enough to tell you that.
The new Unreal 5 engine is indeed exceptional, the problem with it is that the Unreal community developed for years packages for version 4 mainly and when version 5 arrived it was like something completely different, most of the significant packages do not work with the new engine and this made it ineffective for creating games mainly. So, my recommendation to the young people here, let this engine develop for a few years (I would say) until this engine is able to do more meaningful things for games mainly and then start investing time in development
The Donut is a running gag in the CG community. Basically it's the first thing blender users create because of andrews amazing tutorials.
Is Andrew blender guru?
@@karchikharval9741 jup, Andrew Price
5:25 Better volumetrics sounds like your typical skyrim mod
I just wish there were other engines out there that companies used besides mostly unreal. when every game uses the same brain, it feels the same on a lot of levels :)
There are, and they are just not as good and versatile as UE. Again that issue is not with the engine but with the devs. Final Fantasyt VII remake, Borderlands 3, Sea of Thieves, Valorant, Fortnite and Dragon Ball FighterZ were all made on UE, they all have different feel, style and gameplay. And those are just a couple of examples out of many!
@@TheBanthaPoodoo This is misinformation. Companies have always had the question of whether they should code something up themselves or get a dependency on some other coding team they cannot control. Here, the dependency just happens to be a game engine, but it's still code. When a company asks this question, they have to consider whether using a more generic, one-size-fits-all solution is worth it. The advantage is you ostensibly get to skip writing all the code for an engine (although if you tune your game enough, you end up writing a good deal of custom code anyway). The disadvantages are your game may not play nicely with Unreal Engine (in which case you end up writing a lot of novel code anyway and/or suffer from performance issues) and you don't have control over the evolution of the product (So let's say there is a game-breaking bug in the engine. You either have to have your developers deep dive Unreal Engine code and fix it themselves or beg the owners of Unreal Engine to fix the bug. If you have your own engine, leadership can send an email to the manager of the engine team, and the bug gets fixed. Or let's say you want the engine to be tuned to better support your style of game -- likely, they'll keep their genericity, and you either have to accept it or make a ton of custom code to transform Unreal Engine closer to something you need.).
As an example, you can bet the engine behind GTA V was highly tuned to simulate a large, bustling city with missions and simulations of your character and driving and so on, and you can also bet leadership routinely connected the team making the engine to the team using it, so likely, the engine team had a to-do list that basically amounted to helping the team using the engine deliver the game with acceptable performance and fidelity and game features.
If you like numbers, the simple fact is Unreal Engine will run slower for any random given game than if a team made an engine precisely to run that one style of game. That's an inarguable fact about code. It's always on the mind of a programmer to create a generic tool that somehow pumps out highly tuned and optimized code, but the sad fact is the fewer things you tune manually, the slower your code will be.
With all that said, sometimes a business decides performance isn't a top priority or that the performance of the generic solution is acceptable or believes they can wield the generic code to solve their problem effectively based on what their needs are and the strengths and weaknesses of the generic solution. However, it just isn't possible for a well-written particular solution to lose out to a well-written generic solution.
Interesting, I didn't know that about some of the cartoony games. I loved ff7 remake a lot. I wonder what's with the devs then making so many games the same then :) @@TheBanthaPoodoo
Diveristy hires.@@Kevfactor
@@TheBanthaPoodooBatman too...
The sad thing is so many devs use Lumen and Nanite as shortcuts to game development when they shouldn't causing games to perform like shit. There is a reason why CS2 still uses a lot of the trick versions of things like global illumination, water simulations, and reflections... because if you are skilled enough you can completely imitate the "real" versions with an extra digit to your framerate.
EDIT: It's also the reason why CS2 has a fraction of the file size of most modern AAA games...
The thing I've heard about nanite though is that it has a higher performance floor but then doesn't really go much beyond that. Kind of an issue when you have a fairly older gpu or say the steam deck.
If you have an older GPU, just buy a newer one, because Nanite won't make your new games faster, actually they will be a bit slower in most cases. It is because with all the fidelity of new games, you'd expect your LODs to be pretty much high-quality already and have more polygons anyway as they count people would have at least 3080 RTX+. Nanite will just distribute these polygons more equally, but also will make them more detailed, so in certain situations when developers don't want to support old GPUs.. which in this case they probably should not support them, because they would purposefully make their game more ugly and old looking on the new GPUs. Atleast as of now I don't know of other solution that would be both beneficial to older GPUs and newer ones except of making 2 entire versions of the game, which nobody will do.
UE4 games only started looking like what their dev demos looked like towards the end of the PS4 generation/beginning of PS5 generation. So 5 years from now, development of games that started on UE5 will start to complete and by then hardware will have caught up as well.
The expectation for these to be used by multimillion developer studio is too high that i rather not expect at all from them.. It's just sad to see that all the game lately just sucks.. better making animation movies rather than a game.. i do hope some developer love their work so much that they would put their passion with this engine to good use.. 🙁
Looking at this stuff it seems far more useful for making movies and TV shows than making games.
Bingo. Today's hardware isn't optimised for a lot of these rendering bells and whistles; most games can't even use this tech with adequate frame-rates on today's machines. Much better suited for movie productions and film/TV.@@BlueTricky
It seems way easier than before, and the performance is crazier as time progress.
I remember a few years back, I had to spend days or weeks to optimize the model like crazy if I want it to run on most devices.
Lightening is basically a nightmare; light baking and testing takes weeks to test it and optimize it for performance.
Environmental lighting is the worst, need to do complex combination to make it right if go for real time lighting. RTX GPU isn't a thing.
50 % of the time is just for optimizing the model.
I can imagine they are going to progress this much further with AI, like the software the IT guy in The Bloodshot, the one that Vin Diesel Starred in. The software basically can do all sh't by just typing and moving objects. It's like a dream come true, bcus AI will take care all the useless sh't and u do the creative stuff.
I am a game developer working in Unreal Engine 5 (v5.2). Currently, our team primarily prioritizes graphical quality, with less emphasis on performance optimization. Typically, we leave the task of fine-tuning performance to our graphics engineers, but this tends to occur towards the final stages of production. Unfortunately, this approach often results in inadequate time for the engineers to thoroughly assess all aspects of performance. Consequently, our games sometimes suffer from performance issues upon release.
I would be more worried about the actual game being fun to play. The teams I've worked with have been very small. So if we ever prioritized graphical quality the gameplay itself would suffer dramatically. Most people don't seem to realize these graphical tools don't really matter to a vast majority of indie devs that just don't have the manpower to implement them correctly. These things take time, and unless you want to spend 10 years making a game sacrifices have to be made somewhere or you better have hundreds of devs working on it. And if a game isn't fun, why bother?
@@NeonValleys Making game fun is very relative though. Focusing on graphics can be "fun" for a certain group of people. Why all the boomers watch TV instead of playing games? Because it is more realistic, that's one of the answers I get. And that's also why they would be more inclined to trying VR if that was more realistic and real-life like. Anyway, my point is creating a game by itself is difficult and fun is a very relative term which is hard to achieve for everyone involved. For a kid almost every game is fun, the older you get, the more you appreciate things like graphics since that is still something new and exciting, but gameplay has already been made in almost every form you can imagine, just worse graphics.
@@AldrazFun may be relative, but it still requires effort. Almost no thought or effort goes into the 'fun' factor of most major productions because like 95% of development goes into the graphics. This is why Nintendo games and small devs often do succeed in that regard since they actually take the time to create a good game feel.
you can make the game as graphically impressive as you want, if the gameplay is dogshit, and it runs like ass. No ones going to care. You want inspiration? Look to Valheim, Zelda, etc. Make a fun game, not a tech demo.
@@NeonValleys What are you talking about? Having your artists make a model with more detail doesn't somehow sap the resources of the coding team creating the systems that support the gameplay or "fun" of a video game. All the person said is their company has the people who compose levels and models emphasizes fidelity over performance -- not that the entire company stops doing their assigned role and jumps on the teams that create levels and models.
The good thing is, that UE is a renderer and a modeler as it is a platformer. That has always been their formula.
You don’t see this in any other game engine.
But what wasn’t in previous UE engines are being implemented into UE 5 and makes games look like a real-time playable movie than a movie or a game.
Another cool thing is that UE5 can adapt with even VFX engines in real-time, which makes games look straight out of a movie, but it is all in engine.
I'm honestly not sold on it yet. Maybe this update is better but I recall someone doing a sampling to compare what Cyberpunk 2077 looked like in the red engine and how it would look in Unreal and everything looked flat, dull and just samey. Then they released that clip for the Matrix game and again...flat, dull. I did not like it. It feels like the introduction of 3D in the animation industry, it makes things easier to make so companies can save money at the expense of quality. Ideally I think the best case would be incorporating the technology into other engines, kinda like what we should be doing with AI now. That won't happen because of greed but on it' own I'm skeptical.
That's because half the things working in cp2077 just flat out break in ue5. It would ha e to be built back up. And when it is, it would look far better. You can't just drag and drop games between engines and thrn say obviously the one it's designed in looks the best. Fucking duh, of course it will. But if it were completely rebuilt the same in ue5, it would absolutely look superior due to the vast discrepancy in tech. This is just a bad faith arguement.
@@taylon5200 no it doesn't. It was the same in the Matrix Unreal demo. I wasn't talking about dragging and dropping anything, they just recreated a scene from the game to demonstrate how the cyberpunk aesthetic could look but sure let's use the Matrix demo then since it's a along the same lines. so now that its not a supposed "bad faith" argument what's your point? The issue is the rendering, it's the engine itself.
It's about games having a unique look and feel that is lost when everyone just starts using a cookie cutter. That is especially the case with a genre like "Cyberpunk" that is heavily stylized, that aesthetic is at the core of what the genre is. Anyone who has watched the Matrix knows it's super stylized as well and you do not get that from the unreal demo. Every game will have it's own texture, look and feel. That is what I'm saying will get lost if it's just used lazily to save money.
8:10 It's like 3d movies. A lot of movies want to be 3d just to sell the 3D but have no effort to focus on 3d, than the movie isn't good in 3D or not better. Same with VR.
Given opportunities don't get used how they could be used.
Unreal always stays true to it's name. I don't use the engine myself, I'm a Unity guy, but Nanite and Lumen alone make Unreal so ahead of the competition that it's actually insane.
you still a unity guy ?
Having watched the progression from early nintendo days .... truly glad to be alive.
This is amazing. And more.
I am blown away! Unreal 5 is taking leaps and bounds in the space and it's fascinating to watch.
Any by the time Unreal engine 6 drops, we will finally have a AAA game made in Unreal Engine 5!
Are Fortnite or Tekken 8 not AAA? ;-)
@@igorthelight Fortnite doesnt use any of the graphical features its cell shaded lol. And Tekken isnt out yet so we will see if it looks half decent in game
@@JRS902 Fortnite is using Lumen and Nanite tho ;-)
Google "Fortnite Digital Foundary".
so this technology or rather the base of it has been out for quite a while, just was never utilized like this in such a user friendly way before. For example a similar technique is used in most games for the character models, dx 10 had a demo of a similar technique on a xenomorph, terrains also do the same thing in some more modern engines the whole dynamic tessellation, i think its one of the gpu gems books that was released 10 ish years ago shows how to do similar terrain techniques that are shown here. Its only now that the hardware has reached the level that we can actually do it on a large scale
There is a difference in how it is applied though. Tessellation of terrain (for use as displacement through height map) is not new by any means, but the main issue is that it generates vastly more polygons than the flat surface with a planar texture would - and it's difficult to make it work with LOD when it's terrain, rather than an individual object mesh. The step forward here is that Nanite now supports terrain, so when you use tessellation, that too is affected by Nanite which means it scales linearly depending on the view distance, so it is only really present and complex close to the player, which drastically improves performance compared to old tessellation terrain.
@@Real_MisterSir Come to think of it it would work really well with VR and eye tracking. Hint only a small portion of your field of view has "high resultion". Sony are already doing something similar with PSVR.
@@TheAkashicTraveller Yea I had a similar idea where camera focus would be directly tied to level of detail -and nanite would make it even better. You could even rank it relative to what types of objects are important to keep detailed, and which ones aren't. Dynamic mesh like foliage, trees, etc could have higher ranking to avoid a jarring peripheral experience when transitioning in and out of focus, but other elements like ground and rocks could more easily get away with lower priority of detail. Lots of stuff could be added to optimize the flow of detail, beyond just plain distance.
And the best thing is, UE is free to use for everyone. You can pay for asset-packs, specialized features, but there is enough free stuff on the market, it's rarely going to be an issue.
The editor(s) adding video footage of whatever Asmon is talking about are absolute chads. Thank you so much.
Rendering hi-res meshes like that is really cool because there's no LODs and stuff, but the models file size are just huuuuuuge, especially for whole terrains. Even if the rendering is uber-optimized, there is a big problem with file sizes for large scope games.
Pretty nice for small worlds and arena-type stuff, tho.
I don't care how cool the tree looks. I want to know if I can climb it. Cut it down or uproot it. The interaction your character has with the world makes you more immersed in it. If it's just a picture that looks cool, who cares? That's why Half-Life 2 was so amazing. The gravity gun allowed you to interact with and move shit.
That is game dev problem, not unreal engine fault
Making great looking games easy is a good thing. Every big dev prioritizes making the graphics look the best it can because it sells, but whats happens once that's one of the easiest things to do?
It's like how artists used to train for years to recreate what the eye sees, and then the camera was invented. All of a sudden these new forms of expressionism were able to grow because everyone didn't need to worry about recreating life anymore.
This gain in productivity can be used to push forward physics and animation, which is already having huge leaps forward. Every step forward is one that doesn't need to be retread here. Solving this problem only leaves us looking for the next one to solve.
Control was pretty good with interaction with objects
Bro the volumetric clouds and fog are already in Red dead redemption 2
“You can tell this isn’t Starfield because there’s a vehicle” LMAO
It’s probably the only field in the entertainment industry that tries to make their own tool for every product, like imagine if movies had to create their own video editing software for every different movie. And musicians trying to create new instruments for every new songs. This is why game development is actually still in its infancy, it’s still trying to build its own foundational tools (instruments) that would work efficiently while be as convenient as possible for any type of game they wanna make.
What are you talking about? VFX companies build their own tools all of the time.
@@Jgvcfguy Just like construction companies will sometimes create their own tools, sure. But you know what I'm talking about don't just disagree for the sake of it lol, there's far more readily accessible and learnable video editing software out there than a game engine, it's the older and more established medium of entertainment which is why it has more instruments.
@@ryanspinoza6586i dont really get you point, no. Foundational tools like unreal, unity, blender, havok among many others? Theres a couple big names in the video production tools, but most is just small software of which you have a lot on the game engine side as well.
They are continuously being developed on just like cameras and video editing software is. E.g. unreal engine is used in movie productions as well.
And btw, you can download an engine of your choice right now and start building your own game, no problem.
Im not sure what your point is, but building a game is quite complex, simple video editing software is basically included in bigger game engines as one small part of many. You know, since they simulate cameras etc.
As for purpose built: there are considerations of performance which you dont necessarily have in books or video, also ip of course. You propably wont get rid of that, ever. Because a one size fits all will never be the best for everything.
The costs of building an engine and keeping it up to date with current tech is insane for most studios. It's no surprise so many upcoming games are now built on it.
Arguably most AAA studios will still use their own engine, because the pros significantly outweigh the cons in the long run for bigger studios with enough resources to develop their own. Regardless, this will still improve the quality of those games significantly because the new tech UE5 just showed provides an incentive for these other studios to 'step up their game' (pun intended).
"I think the reason Unreal games run like shit is because the devs concentrate on making the games look good not run good" 1000 percent. There's a bunch of great games that run great using Unreal but the look of a game is usually what makes you think it's Unreal and when you can look at a game and instantly tell it's Unreal it's usually not optimized at all lol
The reason why all unreal games will look similar for a while is that nobody has had time to create new assets for it yet. As studios invest time into new custom assets, games will start to look very different.
They've had a decade. UE5 is compatible with UE4 assets, and the UE4 asset store has been around for ages.
There's something about UE5 that Epic isn't bringing up often enough that I feel like they should. All of these improvements to the engine are quite simple to add or enable, which means even someone with very little experience can still make something look really good with the help of procedural generation. Make a ProGen map, go through and add your own little flairs here and there, then boom. What this also means is that devs can be more time on the more important aspects of a game, the gameplay itself.
As much as I like good graphics, I will always always take better frames and performance every single time
Play tetris then 😂😂
@@lovrovalentic3056 what sort of shit logic is this lmao
The funniest thing among it all is that unreal is not just your regular game engine, it is a damn powerhouse, people here must have seen movies, series, and product commercials from perfumes to super cars, which were virtually produced using unreal.
Man, seeing all this UE stuff and who "easy" it seems to make great stuff with it, I can't help but kind of feel bad that I didn't get into game development as a kid... You can do so much with Unreal Engine, it's crazy!
You can get in to it now as a hobby. Make a little puzzle game for your family to play at christmas or something.
Well honestly since UE4 to UE5 was kinda a big jump in every system, devs have to re-learn a lot of stuff anyway.. so its the right time to jump into it whenever you want. But.. it is far from "easy" to make a game in UE.. any game. You'll soon realize that these new features that Unreal is adding isn't gonna help you much when it comes to the overall time building this. It could even make things more complicated, because all these features also have their own negatives.
If you did, you would probbably be using Unity more than Unreal.
basically it dinamically uses as many polygons as required, if something is smaller than a pixel in your view, it just uses a single point
Nanite is 100% overblown. In projects where there's not a TON of stuff on screen, it will make the game run worse compared to not using Nanite. For example in my own project, It's all stylized art assets. The game runs around 90fps on my 2080super, but if I convert the entire project to using Nanite, the fps drops to around 60. Nanite is great for projects with a metric ton of stuff on screen when your natural FPS drops super low. Otherwise so far, it's absolutely not worth using.
of cource, instanced static meshes will always be more optimal as the engine only has to deal with one mesh even if you generate 1000's of copies. Nanite is great for things that need to be dynamic or need lod's. They do tech examples of making everything nanite because it's a tech example it's not suposed to be how you do it.
@@DatDirtyDogexactly. Thanks for a realistic counter to this person's bad faith arguement.
I've noticed that as well when trying UE5 - the overhead of Nanite essentially caps the game at 60fps on high-end GPUs. IMO its a cool art tool, but standard LODs and smart level design are better optimization tools in 99% of cases.
bruh. I've used nanite in real situations. The fact is, unless your game is HYPER realistic, it's not worth using. Every game has or should have LoD's period. It's not a new tech. My original statement still stands exactly as it was with no "bad faith". just facts and experience of using it first hand.@@taylon5200
100% this. Anyone that's actually used it would know it. Just like you said. Spot on.@@terminalferret6899
waiting on unreal 5.3 nvidia caustics branch to come out in a couple months. that will be insane. working in 5.2 caustics for now but these 5.3 changes are massive.
This is perfect for the gaming industry and is going to propel AA and AAA developer teams. I am also sure we will see some indie devs using these new features from Unreal 5.3
It's also going to kill hardware. Right now with UE5 there's games where the minimum system specs are a $5-600 graphics card at today's prices. We're probably looking at 9000/6000 series cards from Nvidia and AMD before this is truly usable. In the old days it was, "But can it play Crisis?" Because nothing could. Well. A $1700 4090 can't do max settings at native resolution and get playable frame rates on some of these without DLSS, which adds input lag. So the question becomes, "But can it play Unreal 5 games.". Consoles? Don't even try for half of this stuff. It's going to get worse. PlayStation and Xbox are right out.
Perfect? More like terrible..
I'd like to mention that VDB's (Volumetric fire/smokes) are not game ready. (Yet) as of 5.3 they lag the hell out of a game with just one or two of these in scene.
There are many ways with billboards and meshes that you can solve the fireplace with current tech. Volumetric effects is of course the future because it is dynamic but it is extremely expensive. Not sure even next gen consoles like PS6 will be able to use them
It's still to near to make it useless, because as time goes, it will be used
You hit the nail on the head about the developers. [ last 2 minutes of video ] Right now ALL of the developers are "too busy" playing with the new toys that UE5 releases, and as such it'll eventually settle down and many will go back to focusing on the other aspects of game creation like Lore, mechanics, and other content that's crucial to a good to near perfect game being created.
They say that UE 5 boosts performance because of those polygons and stuff, ive heard that years ago when the matrix city demo came out. Some games are using UE5 now and did we see performance gains? No. Those games are very GPU demanding and the performance sucks.
Is not the engine issue. Is the time it takes the dev team to get used to the new engine. I imagine some stick to the already existing features of unreal 4 for the sake of not falling into the experimentation pitfall
I hope they can soon make it so that animated mesh can be reproduced with nanites, which means reducing 70% of the pipeline for creating a character without the need to create the character from scratch. With this technology, we can focus on storytelling and gameplay instead of the appearance, which consumes a lot of development time to complete the game.
I wonder if if these "nanites" used in Unreal 5+ are the same feature that I saw being developed when I got to take a tour of a game development studio back in college over 10 years ago. It's a similar concept, but seems like a different execution. They showed us a demo of creative tools in development which included infinitely scaling objects, made up of what they described as digital particles, that disregard LoD. It was even more nuts then, and I thought games were going to be revolutionized, but nothing seemed to come of it. I don't recall them having an association with Unreal either, other than using UE3 for some of their games.
that may be voxels you are thinking of
Nah, voxels are just boxes as pixels in a 3D environment.
The dev tool that I'm thinking of allowed the developers to basically build any object in-engine with digital particles, scale any object infinitely without loss of an object's visual quality, fully dynamic destructibility, and more. They could even make an object a building block of another object. @@digitalrandomart3049
The "Infinite objects" is hard to believe. But yes the entire stuff is stunning indeed
Can't wait to not be able to afford a machine that can play a game utilizing UE5
Today it's a 1500$ PC. In a few years that would be a 1000$ PC.
The sad thing about nanite and lumen is that they are heavier than i originally thought, not as much as ray tracing though, and that nanite is much heavier than lumen, also the amount of VRAM needed is ridiculous, you need more than 16 to even be able to use it, i really hope in 5.3 it is better because i only have a mid range card with 12gb
Nanite isn't heavier as such in terms of FPS performance, but the problem is more older games up to like 2020 were made with less polygons, because they wanted it to be playable on GPUS like 2060 etc. Nowdays you would make about the same amounts of polygons that Nanite will give you as default quality. But that means it is like 3080+ GPUs. You can make Nanite use less polygons, to run well on older GPUs, but then you'd only hurt yourself as it would look ugly for other users. Unless they will make it a graphical settings in the future .. and I don't think it is currently, this won't be scalable. Also.. Nanite can make the game by itself like 10x more larger in terms of disk space, which kinda sucks, but it is better than taking away FPS or VRAM.
Meanwhile Zelda TotK has amazing and more complex physics and interactions than any game made with this engine has... on a console weaker than modern mobile phones.
Are they going to fix fps drops in games? Every UE game is a stutter fest.
What games?
Not an engine problem, it's a Dev problem
@@Crazy09starkillor But other engines do not suffer from this issue.
@@Crazy09starkillor it is engine problem. Because epic promotes it as easy to use with a bunch of ready-made solutions and assets. But when optimization is required, it often turns out that it is easier to write your own solution from scratch than to fix everything that the engine generates.
Epic sells it like easy solution what is false marketing.
@@Littlereaper558at the end of the day proprietary engines are better for optimization because its tailored for the type of the game they want
Model quality no longer depend on GPU, but rather on disk space. But it doesn't yet support transparency or skeletal meshes like characters
It’s cool and all but when’s a developer gonna make an actual good game with it?
Remnant II is pretty good ;-)
Just wait for a few patches xD
To how nantite works in short think of turning the subdivide simple combined with subdivide Catmull-Clark amount from a static amount to a dynamic amount. Subdivide simple just adds more polygons to the object without reshaping it at all while subdivide Catmull-Clark smooths the edges by using more polygons to do so. Nanite most likely has logic to tell when an object is intended to be rounded or a sharp angle so can go from the insane amount of millions of polygons either made by the system or imported in a 3D model to the varying amounts using unsubdivide and resubdivide. You still will get a more true to what the artist wanted take on it going with hand made but Nanite allows you to have much better lighting with Lumen which forces you to use ray tracing ... kind of in 5.3.x there is a version that you can go back to the reflection spheres and use light maps and shadow maps instead of real time lighting. The performance improvement ranges in a wide range depending on the scene it can be as high as 2.5x improvement moving back to the traditional lighting system while using lumen for the light map.
Looks pretty good.. now we just need someone to put it to proper use.
Knowing Cyberpunk 2 is going to be based on Unreal 5, I think that's a solid one to look out for. But it may take a while (hopefully not 7+ years like the first one lol, but then they were also developing their inhouse RED engine at the same time which severely fucked their work pipeline and schedule).
@@Real_MisterSir haha.. yea it will probably be a while before we see that one..
@@Real_MisterSir As UE5 dev I can tell you that such migration from 1 entirely different game engine to Unreal Engine can be extremely difficult. Maybe they would be better off starting from scratch, but that could take even more than 7 years lol. Because that would include all 3D assets and animations, which most companies just take from their older game titles and customize / change. I think they will try to migrate everything they can, probably will manage to export 3D assets and even animations. And since RED engine uses C++ they could manage to export most of the code, but the C++ in Unreal is not standard, so who knows how difficult that can be. Either way, I think it can be done in a month if they are really smart and less in a year if they are bit dumb. But then it will take like 3-4 years to develop on top of that I would assume, because they need to make a lot of new content.
@@Aldraz Oh they definitely are starting from scratch, I would assume the only thing they will carry over are the assets already made for CP1, and as you say, if animations can be carried over that would be quite a time saver. But they announced pretty early on that they were switching to Unreal for all future installments rather than using their own RED engine - I guess it also helps take a toll off of their devs shoulders not having to both deal with building their own engine while also developing games on said engine. A little less freedom and control, but a lot of time saved and a pipeline that is easier to manage and set expectations for. It'll also be a benefit to hire new talent when Unreal becomes a standard engine to work with, so the training/adjustment process is minimal. Larger studios can more easily afford such expenses tied with doing everything inhouse. That said, it's been a while since they have made any public talk about the 2nd installment in the franchise, so of course time will tell if they deviate from their prior announcements or go through with it.
My understanding is that CP2 is going to be entirely separate from CP1, as in they are making things from the ground up rather than iterating on the existing core framework of CP1, and I can understand how it may free up some future potential if they deem the current code and systems to be too much spaghetti to really develop the way they intend it, with potential limitations of the RED engine that can't do everything they want it to without pouring a lot of resources into that part, let alone the game itself.
@@Real_MisterSir Yeah I think you are right, completely agree with you. I believe they also thought it would be too much of a hustle keeping up with the latest trends, technologies when they directly tried to compete with Unreal Engine via their RED engine, which is bit of a madness and all props to them for even trying, but with how quickly everything is growing nowdays, they wouldn't be able to keep up with Unreal Engine that is powered by open source community + multi billion dollar company Epic games. And now that I think about it, maybe nobody will be able to keep competing in the long run. There has been many game engines in the past, now its almost as if it was only Unity and Unreal. If Unreal had a great 2D support, it could even beat Unity.
As someone who's been working on a game on Unreal engine 5.2, nanites truly are amazing and help so much with frame rate. I've got thousands of grass, trees and vines all swaying and affected by wind without a dip in frame rate, I love nanites XD
Cool
asmon is totally right, as an indie game dev myself, we switched to unreal engine from unity to increase productivity and save time to make great games, but we are still using 3D lowpoly models cuz of the art style and to reuse assets from prior games ofc
What are these great games you're making?
@@robservin7664 so far, we didn't produce many succesful games yet, where we even had to switch platform from mobile to PC
so for PC, currently we only have a "Vampire Survivors"-inspired game with some twists here and there in 3D Scifi setting, it is called "Defend Earth: Xenos Survivors", just recently we have pushed a free version, where people can test the game before buying the main game, same name just with a "...- Prelude" at the end
the new game on the unreal engine is still in development, so I can't tell anything yet :D
ty for being interested btw!
Is UE easier to use? Last time I used it, the documentation was garbage.
@@spacejunk2186 I am not a programmer, but what I got told from our team, that unity is more friendly for new programmers, since C# is easier to learn while in Unreal, it is C++ and you really have to be very precidse there, but in unity, you have to write the code from the beginning, while in unreal you have blueprints which are basically rdy-to-use-scripts
so if you have no coding experience, then unreal is much better, cuz you can build games via blueprints, but even if you are a programmer already, making games via blueprints is way quicker, we basically doubled or even tripled our production speed since we don't have to write our own code anymore, and also can use paid plugins to create procedural levels
e.g. we are currently in the 8th production week and after additional 1-2 more months (depending on the polish grade), we would be rdy to launch our game in Early Access
I remember back in the day it was unreal tournament vs quake 3 arena and both were great games but quake and Carmack were always able to beat them out. There was a whole generation of games built on the q3 engine. It was the standard. But my God the tools we had at the time, ie radiant were so cumbersome and hard to use. Took a lot of learning and trial and error. Drawing brushes by hand, which makes what they were able to make with that engine incredible. Alice comes to mind. Makes me wonder what Carmack is up to cause he always has tricks up his sleeve. The last innovation from him I remember hearing about was megatextures in rage.
This is the biggest jump I've seen since crytek. The original engine was amazing. I remember seeing the editor demo and being able to just paint terrain that automatically populated with foliage based on the height map was crazy. You didn't even have to compile to test you just hit a button and it dropped you into the map. With old editors you had to render bsp's the load the game then load the map through a console command. Ugh.
Yeah-yeah, everything is awesome... Until AAA game is released and it requires 100 patches.
so you know how not too long ago 6gb was the low end in gpu's? well, with this coming up, you'll need 24gb at minimum :DDDDD
and most importantly, STARFIELD promotes new technologies and requires 4090 owners to update their weak hardware
Yes by making good graphics more accessible by small devteams big aaa companies will actually have to compete with them on good story telling mechanics and gameplay. Since there biggest strength is gone.
Thanks for the video, love all things UE. I work in UE and love the tech. The one thing I want to comment on is this constant echoed idea that 'all games are going to look the same because it's made in Unreal'. That all UE games look identical. I can't believe how shortsighted that is. It's all about he dev team and the art team. What style are you going for? Who's art directing the team? Can the environment, character and prop artists match the director's vision? It's the same with a film--can you imagine saying, well heck all films just look the same because they are shot on *blank*. No, it's the art team. UE can do anything you want to do. Please everyone understand that. Basically, Az you nailed what I said here the last segment of the video. Cheers love you all!
Cant wait for hundred of games that look exactly the same with the most plain art direction ever. We got to the point where stuff like UE is not necessary anymore all we need is decent art direction and to use this technology to help but most AAA studios wont bother and just abuse those features.
As impressive as this is (and it's really damn impressive), but frequently unreal has performance issues and idk if thats a game dev issue or an unreal dev issue. Does anything in 5.3 address this? Or is it all really pretty new tools?
It's a bit of both. I work with Unreal and most of the time it is a developer skill issue. We have a few people on the team with over 15 years of experience, and they can usually give us great advice. But you also have to keep in mind that this monster of a codebase is changing at breakneck speed. Most projects pick the most recent version when the project is started, and are retiscent to upgrade unless a new feature will be very useful
I’m a Senior 3D Character Artist working in games for over 13 years now and Unreal is the best engine to use, it’s also the easiest to use bc of the tools. I like that it can handle actual has instances instead of haircards, it is pretty expensive though but as hardware will catch up it will be the best way to render and simulate hair both for the artist making it and the person watching or playing a game.
UE5 updates never dissapoint, god damn this is so impressive
"In unreal you can now spawn the universe and control the entire thing with voice commands"
So far, I've only played unreal 5 games that run like shit, even when they don't look nearly as good as these demos
Agree. Developers don't really use it correctly. OR maube game engine itself is just too heavy.
Didn't tested that yet myself.
They put out these update fucking fast and just act like it’s fucking nothing. Like the people using this must only just start getting the hang of it then a whole bunch of new shit that blows the older stuff outta the water comes along. It must get a bit annoying at how quick they are
5.3 Frames Per Second
Virtual production with UR 5.3 will replace a lot of Hollywood VFX teams. No need for setting up lights anymore when you can tell where a projected one should come from equally reflecting from the subject at that same angle the lighting was added which was always the most difficult part in CGI. Lighting on skin tones.
Compared with Starfield, everything's a big deal.
I'm happy to see people thinking about how Unreal has the potential to make games look too same'y. It's been a long standing complaint from smaller developers and one of the reasons why people often use Unity instead. My day job is to modify the UE5 renderer so that we can have a stylized game. It requires a LOT of knowledge and it's very difficult to do well. Big studios that have graphics and rendering engineers with experience creating renderers from scratch will be the studios that have the knowledge necessary to create stylized UE5 games. It's going to be very interesting to see how different studios make their games stand out. It's fun to experiment with mixing stylized rendering and UE5's photo realistic features like fog and volumetric lighting.
This was made right before the Unity changes, so I wonder, do you think it will keep happening? Most people will switch to Unreal Engine now