NEW(3/22/2024) UE5.5+ feedback: Please Invest In Actual PERFORMANCE Innovations Beyond Frame Smearing For Actual GAMES.

Your completely missing the point.

This engine is built around an accessibility option(TAA/Temporal blending)

The problem with you is you think this is about my project, this is about the semi-dynamic games being brought to UE in the future. Semi-dynamic, I’m talking about DOORS, windows, basic things that UE’s baked systems can barely handle with good quality. They LACK systems and workflows that promote reasonable performance for players.

They are admitting to wanting to focus on replacing meshes with Nanite, They are admitting to trying to create MORE bullcrap in TSR even though more temporal frame=more chance to smear. There are ways to make this better for everyone and they are investing in poor directions.

I’m asking for basic standard and so are the #104 who voted in agreement.
GameDev needs more priority right now becuase the games that have used it perform terribly and they are mostly static.
Basic systems like volumetric lighting still have stupid amounts of leaking etc.
They are abandoning all reasonable systems and studios are just going with it because of OUT OF CONTEXT information.

1 Like

So, you ignore all of my points about, just, not using it?

Anything and everything you listed as being ‘bullcrap’ can simply not be used.

You owe this developer community an apology for wasting our time with a nonsensical series of assumptions, presumptions and flat-out ‘calls-to-arms’ against the very company who owe you nothing, other than basic human civility.

So, you do not have to be civil, but they do and they must give you everything you want?

It is not about you, and you are ten feet away from trying to rally activists to your cause, because why; a lack of vision?

If you wish to recruit members to your cult, do it elsewhere on your own time.

That being said, if you have legitimate feedback on UE 5.4 before release, I would direct you to github to download, compiles and run under a series of circumstances to add weight to your words.

There is PROCESS HERE which seem to think does NOT apply to you. There are forms to fill in, comments to detail, code to analyse, structures to make determinations upon.

Where is your documented tests and debugs, complete with valid testing environments?

We need proof, not BS claims, based on nothing more than subjective ego wishing well synopsis’s.

I have temporarily allowed you to post here, unblocked by me. Be warned, I have a block hammer in one hand, and my hammer and I will be watching.

7 Likes

So, you ignore all of my points about, just, not using it?

There isn’t any other engine as complete as unreal in several ways I won’t waste my time explaining, still doesn’t mean we shouldn’t criticize it. They are ruining this engine with a whole bunch of other focuses game devs don’t need and are founded on TAA focused tech. Their are thousand of people who hate TAA, that included upscalers like DLSS garbage.

You owe this developer community an apology for wasting our time

It’s not a waste, we have #106 votes and 4 votes away from being #1. They’re systems are too dynamic and focused on content like FN when a lot of studios will be bringing mostly static scene with minor dynamics unreals poor baked systems can’t even handle. Both Lumen and baked lighting leak, their LODs algorithm is still broken and lacks streaming logic. They are catering the engine to virtual production and extremely underperforming systems. And there engine is still super unstable visually, and every new feature VSM, Lumen, Nanite, TSR cost around 3ms for each one, at 1080 for a 3060.

It is not about you, and you are ten feet away from trying to rally activists to your cause, because why; a lack of vision?

If you wish to recruit members to your cult, do it elsewhere on your own time.

I am recruiting MY OWN developers to fix this engine. It’s called a studio, not a cult.

I would direct you to github to download, compiles and run under a series of circumstances to add weight to your words.

Believe it or not I have tried, but it requires 200Gb and tons of addons to even compile which I don’t have storage for.

You are the one wasting time, TAA independence is needed for many gamers and the issues with performant systems before moving to area we know are even worse. Lumen could be higher quality AND faster if they supported baking to a small degree but no, we get tons of noise becuase it’s constantly making sure static content is still, static content.

We should be complaining, Epic should be thanking this thread because we are giving context to their identityless engine and the 106 voters will agree on that.

1 Like

And there it is,

You have not even gone to github, created a test environment, and run a series of tests.

‘I don’t have storage’ are not words in the vocabulary of a games developer!

Games developers have ALWAYS got storage, next to pizza and RAM, it is one of the three things they will never run out of.

TAA IS NOT REQUIRED, like ALL of the elements you mentioned. You do not even need to bake your environments.

If you have a games studio, why not get the members of your team to post here with details of what steps they have taken?

THAT is something I think all of the developers here, and Epic Games, would enjoy reading.

You honestly have worn the wrong pair of shoes to dig your heels in on this topic, and just because you tread lightly, does not disguise your clumsy footwork.

Stop trying to tell Epic Games what they should or should not be focusing on, as they have got a very long term plan of action which I would imagine can only be expanded or contracted, but not abandoned. Development of what they are releasing this year, I imagine would have begun 18-24 months, and you want them to simply stop doing what they are doing on the strengths of nothing more than flicking a coin into a wishing well, and being disappointed that you did not get a satisfying ‘bloop’ when it hit the water?

C’mon, be real - it is easy to tick a box on a forum and therefore ‘claim’ that you have unanimous support for your cause, when we all know that without actually providing testable and repeatable instances where a MAJOR development change for the Unreal Engine (pre subscription model in 5.4, which I know you probably cannot afford and will complain about endlessly) is warranted, is simply not going to happen.

You are trying to tell the engineers how to build the engine, when you are not even steering the ship.

I would like to know, how many of the people you supposedly have on your side, have actually paid for this engine and have testable and repeatable evidence in the context of your point.

So, I am now done with you and would like the people in this thread to post back with details of how long they have had a licence for, what they are using it for, and to detail the evidence in support of your unsubstantiated claims.

You brought this on yourself and, where is my mocha latte?

5 Likes

You’re being extremely hypocritical.

I’m not allowed to criticize Epic according to you because I’m not a engine programmer and “not” steering the ship WHEN YOU are the one “not steering the ship” of my development and personal experience with engine yet you write complete drivel criticizing me for opening this thread.

YOU have not been affected by the saturated and broken effects this engine offers it terms of its graphics pipeline and fundamental attitude toward a DISGUSTING AA method.

You are trying to tell the engineers how to build the engine, when you are not even steering the ship.

Yes, I AM. I am a customer, with the RIGHT to complain, and so do with feedback tag EPIC PROVIDED.

Get off the thread if you can’t handle or stay topic.

I have stated Epic has no obligation to listen to anyone included this thread, just like I have no obligation to answer to you or your irrelevant questions to fixing the clear issues I stated with the engine.

2 Likes

You are not the customer.

You are a member of a community who seems to think they have the right to question the very community others are members of.

If you do not like the community, leave it; because I :heart: this community with all my heart and joy, and to see toxic calls-to-arms because you personally and subjectively feel that something does not work the way you want it, is beyond offensive.

Most of the time, I will ignore such offensive and overtly negative toxicity, because I understand how it can manifest itself in reality.

Someone cuts you up on the road, someone gets your Taxi-cab, you thought you had another can of beer in the fridge on a warm summers day; or the cat scratches you.

We can all have bad days for whatever reason, and sometimes, we can even feel like as if the world owes us something, because we are good people and we pay our taxes on time goddamit…just, please, something go right for me for once.

The world owes you nothing.

The internet was supposed to save us - I would posit the internet has destroyed us, for there are some who have confused other people’s genuine tirades as legitimate ways of conducting oneself in reality for things that are subjective demands.

In other words, people used to use the internet for venting frustration, like t-bagging noobs whilst humming the theme to Hawaii Five-0…not for speaking about proposed changes to a perceived social injustice against a company, for not giving you what they think they should, or even; by operating their company in the way that they do.

One is venting, like a pressure valve, the other is narcissism, wrapped up in a blanket of apathy directed towards oneself - because if someone truly cared about themselves, it would be themselves they would try to fix, not their surroundings/environments/social circles, so that they feel better internally, by trying to affect external factors.

You cannot flip the board of monopoly over because of the roll of dice you got.

Look, being real to you as I can be right now, in spite of your inability to perform tests and submit results to the community, you must understand that what you are saying can ONLY be interpreted as subject and not empirical or objective.

Someone might say they agree with you, but, what are they in fact agreeing to? The thing in your head? Or something which is a standard test, repeatable and understood by anyone else performing the same tests?

You cannot argue subjectivity with objectivity
You cannot argue objectivity with subjectivity.

Run tests, compare and contrast with others, submit reports to Epic Games and maybe one day you will have bought the engine and be using for a project on a game that pays you a bazillion dollars, until then, you are not a paying customer.

You are the member of community that you do NOT control.

So - stop trying to, and have some respect this community…some of which have been developing games since 1998.

Before the dark times…before social media turned everyone into an entitled “Karen”, hell-bent on trying to alter the trajectory of light from the space-time continuum in their favour.

Not that is what you are doing, of course.

I was merely giving an example of how the internet has changed human beings…

4 Likes

Epic has this section of the forums to allow for feedback. Not attacks. I would very much like Nanite and Lumen to improve as well as the engines renderer for performance gains. If comments are made to where people continuously feel disrespected, attacked, etc…, whether your intent or not, you’re missing the mark by lightyears and will lose the very thing you want to fight for.

It’s best to take a step back, breathe, maybe make a new thread where you lay out issues in a more coherent and cordial manner. I would like to create a thread, but I’m pressed for time.

I’ve taken the time to dig through the engine source code and find ways I could potentially optimize my game for the best possible performance. Some of those findings I’ve posted here in this thread. If you take the time to do the same, you’ll be able to better present a more coherent and acceptable argument.

Another solution would be to investigate potential improvements, implement them [or try], then submit pull request via GitHub after following Epic’s code guidelines.

I want improvements just as much as you do. But offending people, especially the epic devs, is the very wrong way to go about seeing important changes come to the engine. You want as many people on the side of positive change in Unreal Engine as possible. So please, rethink your approach.

3 Likes

Everything I have stated on the first message is logical and coherent. It has been after many insults from other people that have invoked my defensive side on the rest of the thread(even tho many of them I retained logic and good arguments).

The direction Epic is putting unreal towards is extremely flawed and I have many post comparing their implementations of basic effects against other engines that have these same effects that perform and look better. I am NOT a engine or graphics programmer, in fact if it was easy to be one for the majority of people, unreal would be worthless, it’s their issue that is causing problems. The state of unreal was dedicated a whole half hour+ to movie production and focused on technology we had already for games but now we have it for something that destroys performance(Nanite vs LODs) for something that will not benefit gameplay of the majority of game releasing within the next 5-10 years years of games, or benefit visuals(Temporal dependency vs mimaps/better LODs and poor framerate=screen issues)

There are THOUSANDS of people who hate unreal for it constant abuse of frame smearing. And the only alternative to frame smearing are broken effects, bland visuals, or poor performance(TSR, which still looks in motion). They are building the engine around an accessibility option. Advil has recently advised against temporal solutions. Epic claims TAA dependency is for optimization, when that is a clear lie when you look at games released 5 years ago, in the same type of rendering pipelines.

Two votes away from being #1.
That’s for a reason. Devs are not trying to target 4070. Most of our consumers buy hardware under $350-300, they should not be punished with broken visuals and this engine is too popular for them not to fix this major, maybe the biggest issue the engine has.

3 Likes

Yea, I expected them to showcase more of the performance upgrades with 5.4 and Render Parallelization. I am hoping they showcase improvements at release or post release.

But the engine is moving in a better direction which I believe is why 5.4 took so long. Refactoring the core RHI system to better support modern graphics api’s, something I mentioned in this thread, is a major step in the right direction.

I mentioned that their use of mesh shading wasn’t as efficient as I hoped and it seemed to be a trend in the industry. But they are making strides towards better use of modern hardware and Api’s.

The two most exciting things for me are these two.

I get the frustration on the virtual production side; It boils me sometimes lol. It does feel like the real-time component of a game engine is being neglected. But I think that is more due to the transition from old rendering API’s to new API’s. Like the jump from DX10 to DX11 wasn’t as crazy as the jump from DX11 & OpenGL to DX12U(I just call it DX13 because why not) and the latest Vulkan. Trying to maintain support for previous api’s(DX11, OpenGL, etc…) and implement modern features on an abstraction layer is difficult.

There are a lot of new things that graphics programmers have to deal with where previously the API maintainers would handle. This is why the shift has been so slow; It’s a learning and resource intense process.


I am a graphics programmer and struggle with a lot of the latest & greatest rendering API’s; So I understand that it’s not fair to expect everyone to dive deep into solving Unreal’s problems through source alterations. However, seeking to better understand concepts can helps us all collectively come with better solutions.

I have hope that when the new pricing changes come in, epic can allocate funds for expanding the VP pipeline production team.

5 Likes

I’m not an engine programmer or skills to do GP, but I know a lot about the concepts becuase It’s been my main study topic for a long time. Temporal accumulation is not the issue, I have no issue with temporal SSR or AO but I know for a fact these can can perform and look better without needing Temporal AA or Upscaling smearing. Epic’s TAA solutions is extremely broken becuase it has too many jitter positions and uses infinite amount frames which causes ghosting.
TSR is too expensive because it tackles the same issue at the cost of more computing instead of using morphologic AA in combination which I stated on the TSR feedback thread.

Lumen could be more stable and fast for the majority of games if they worked on bakeable volumes with some more information like depth functions seen in surfels for GI by EA. But no, that design doesn’t work for FN or virtual production where everything is rendered.

Lumen reflections performance got reduced by a whole ms but still depends on TSR(as commented in the shaders) to become sharp. TSR can’t even make motion look sharp on reasonable hardware and resolutions.

I understand the problem with making LODs, and Nanite is poor solution compared to better workflows that should be implemented instead.

EDIT: This feedback is now #1.

2 Likes

Aside from obvious part (rendering/antialiasing/gpu performance).

Would be also nice to see improvements to blueprint’s vm performance. Even roblox lua vm is faster than unreal’s blueprint vm at the moment.

I am not expert on writing VMs and compilers but there’s probably some way to increase performance by reducing amount of calls and inlining blueprint node calls together into single call or something similar

2 Likes

We may have gotten something like that, I hear the render thread was increased by 50% in 5.4

They claim a 24% increase on the GPU side but that isn’t exactly true on my side, even with hardware similar to 9thgen consoles. I imagine this comment accurately reflects Epics claim on “performance improvements”.

I’m still doing regular test to make sure, just waiting.

Refresher/TLDR to Epic Games if they are keeping an eye after the recent 15+ votes that have catapulted this thread from the original 1# feedback thread:

  • Make effects as performant as other implementations. If I can get ms timing on effects in other games, you can too via api/hardware inspectors. UE’s implementations of the same effects always fallbind other versions with little improved or worse quality in UE. Make them stable indepently of Temporal AA/Upscaling(DLSS included). That includes your terrible dithering effects and making temporal variants of your effects the start up default and not hidden from developers.

  • And make better designs related to shadows, lighting and raymarched reflections that take advantage of the common game environment being: 80% static environments with dynamic lights with precomputed movement and or small dynamic lights. Rather than systems that are made for games like FN were absolutely no information can be baked,tweaked(brightness, area etc), or interpolated by developer artist becuase Lumen fits FN which is 100% Dynamic/Destructible world combined with a constant moving sun where constant iterations on the GPU waste performance. Examples of better designs (in term of caching/performance, not quality) would be The Divisions GI, Quantum Break’s GI , and I have explained some more stuff in the Lumen thread.

  • Do not tie effects to Nanite(like PGC or VSMs) or continue mislead people with it’s very situational benefits. Foliage is one of the few kinds of popular meshes where overdraw should be tackled with something like visibility buffers but WPO explodes both VSMs and Nanite.
    —So for the that reason UE(the many games that will use this engine in the future) desperately needs something like deferred texturing as HFW performs way better than any UE5.3- game with nanite. Another investment related to this area is a better LOD system that optimizes topology for overdraw and creates LODs based on a higher quality algorithm.
    In this link, I state systems that would greatly benefit the optimization workflow for draw calls via precomputation meshlets that developers could create from finished scenes with static(at least) geometry.
    —In the same link I state a system that would help memory as well.

2 Likes

Aside from performance it would be nice to see “new” features being compatible with eachother at all. Like Forward Shading + Lumen + MSAA + Fur that doesn’t glitch like my GPU dived into a swimming pool and died a horrible death until I disable it.

4 Likes

123 votes as of 4/25/2024.

I went ahead and saved the HTML and this thread online.

Please review this comment Epic Games.

I would be happy to give rendering examples references to any engine programmers willing to take this thread more seriously. I’m both a gamer and developer and UE5 and it’s main focuses extremely disappoint me. I have made comments before saying I cringe when I hear developers state they are using this engine when their game designs would benefit from other implementations but your engine remains the only real option for the market.

The difference between my views and Epics is I don’t believe TAA/Temporal upscaling is a viable solution on next gen consoles for quality and performance reasons. This doesn’t make me against TAA, just the current implementation and purposes. My game design is more similar to majority of environment design(90% static meshes, moving directional light and time system, interactable systems we as devs can predict and trigger optimized environmental changes ) rather than FN where nothing is reliable for caching. Epic provides 30fps scalability becuase they don’t believe 30fps is unacceptable on these platforms unlike me and more importantly other developers I communicate with.

As a developer I understand the problems gamedev needs solutions to, but as a consumer I see much better ways and unfortunately Epic seems focused on a path I think they are unaware is broken. The thread isn’t meant to be toxic, it’s meant to be genuine. Fixing Unreal has a better impact on the market than making cryengine or unity the best engines ever made becuase few studios uses those. It makes no difference to me if Epic benefits from a direction change and improvements in Unreal.

EDIT: Let me be more clear. This is not impressive:

A major problem with this is the lack of 60fps, so playing it isn’t going to feel good and it’s also introducing motion judder. Nanite is not going to make or break this game(ofc, if performance is bad enough, the studio can take financial hit) but that’s automatically a goodbye to 3ms worth of performance. In all honesty, the characters are barely ahead of what has been achieved on PS4 which means that shouldn’t be a source of performance problems. The resoltion isn’t anything outstanding compared to 1080p with SMAA and lite temporal methods seen before. Also, how do we even know how clear the game will be in movement? Probably not very clear since TSR goes against requisites for clear temporal motion.

Lumen just got more expensive 5.4 and most of the scene is static, which means computation should be low(like REALLY low), even if lights are destructible from a gameplay stance, that’s not an excuse for them to suck up performance since other proprietary game engines have had systems for these kinds of lights but UE doesn’t(we are talking about baked GI contribution that doesn’t use lightmaps)?

Where any systems developed to take advantage of the world limitation of this game? If the answer is no, then that is my point. It’s using poor rendering designs that over compute for a FN scenario.

1 Like

5.3 Nanite test.

I will be converting the project to 5.4 soon to see if Nanite can match perf on the optimized scene.

Here we are a year later with 5.4 and the same issues…why spending time on so many beta features when 5.0 performance has been the major issue since release? Are they blind? Or do they just not care if anyone uses their engine to make games any more?

RHIT Multithreading was a step in the right direction, but still not much for Lumen performance at all…

1 Like

they just not care if anyone uses their engine to make games any more?

Not games that aren’t 100% destructible like FN or games made by studios that recognize the serious issues with abusing temporal flaws for fake optimization.

Proprietary engines don’t use Lightmaps for big projects becuase of the memory issues, that doesn’t mean they bake everything to 100% staticness. There are no major small scale baking systems that save performance for consumers and allow interpolable senerois, Nanite meshes add overhead unless you use a mesh so dense it needs TAA to reduces insane noise. TSR cost well over 2ms on hardware 85% faster than PS4.

The best performing UE5 game is the finals, which DOESN’T even use Epic’s UE5 or UE5’s Lumen or Nanite. Epic’s has no obligation. But the more vote here I guess show what devs want.

Lumen reflections became 2x fast and GI became 2x slower.
Because there is no performance goal outside a worst case scenario like FN.

most important, the finals aint use chaos, they use proprietary build with physx !!!

1 Like

The Finals doesn’t even USE Epic’s Unreal.
The game is too destrutve to support Nanite(bye 3ms) and virtual shadow maps(bye more performance) and uses NVidia’s custom unreal for faster DDGI.

That’s why I laugh everytime I hear someone defend Epic’s unreal with the finals. It only proves the threads point that Epic’s rendering systems are anti-game performance.

I suppose it’s that day of the year again :slight_smile:

Request for LTS versions of UE and marketplace assets.

1 Like