Unreal Engine 5 = Game Changer

Time to sleep

 

Alright.

This is half a billion triangles.


(150 statues 3 mil triangles each)

1 fps. That's without GI and with untextured material.

Not the latest unity version though.
----
Also:

(Why the hell is there nodejs again?)

 

Not that is should make a large difference, but you're also in play mode, in the editor. A published build, connected to the stand-alone profiler app, would be a more accurate test.

It probably won't make a giant difference (maybe you could get up to 2fps!) - just worth thinking about. The editor adds a lot of overhead.

 

If I understand correctly, the UE5 demo wasn't actually rendering billions of triangles. They had a culling and optimization system which brought the total polys down to ~20 million/frame. What's your framerate like with 7 of those statues?

 

What's the specs though? How does it compare to a ps5?

 

Ok but... what's the purpose here? What's being compared?

Anyone here worth their salt knows that just measuring FPS is meaningless, because there's no way any of us would ship what you're showing, and we don't know what your hardware is, and...

Even in the Unreal demo there's not "half a billion triangles" on the screen. They clearly state something to the effect of their billions of source triangles getting crunched down to around around 20 million on-screen triangles. Which, of course, is 100% achievable with Unity today if you're able to put in the work with LODs. And since you're just working with repeats of one model that really shouldn't be difficult.

Of course, "transparently automatic" is still nicer than "not difficult", which is where I do like what we see in that Unreal demo.

 

Yep, you understand correctly.

Six statues give 21.8 fps.

Spoiler

Honestly, I expected more.

Also, this kind of thing makes me hit swap file which requires killing off processes afterwards. So if anyone wants to experiment more with this, the statue can be grabbed from turbosquid. The have tons of free classic sculptures there.

Also, given that this is definitely a scan/poly soup, I can't unwrap it quickly in blender to put on a material on it.

What's the point though? The algorithms aren't there and architecture is different. Consoles have shared RAM, if my memory serves me right, that gives advantages.
Overall I have Ryzen 3600 and 32 GB of RAM

 

One of Unreal's advantages is the willingness to basically start again every few years. They don't need to design their new code so that it works with systems designed back in 2004, with code potentially as old. Even when Unity does make major changes, which they do from time to time, they still have to keep the developer facing interface as similar as possible in the before/afters.

If the complexity of each codebase was reduced then I can easily see that speeding things up rather than slowing them down. I would suggest that Unreal is an example of this, enabled by the fact that they aren't even trying to make projects forward compatible between major versions. They aren't carrying nearly the baggage Unity is.

As mentioned before, though, different design focus. Imagine the disturbance in the Force if Unity announced that their next major version wouldn't be compatible with what people are doing now. I wouldn't care, but plenty of developers would, and reasonably enough.

I think Unity definitely needs to get through this tough patch ASAP.

 

Few posts prior it was asked "How would those statues perform in unity with a single light" so I plugged few statues in. Mostly for the hell of it.

Proper comparison can't be done, because algorithms aren't there.

There's still baggage and remnants in the codebase though.
Then again, unity offers a blackbox that may or may not have eldritch horrors within.

 

I know that you know this, but the proper answer to that question is "it's irrelevant because it's the wrong question".

The algos aren't there but, as I pointed out, a great deal of equivalent functionality is there if you're willing to apply it manually.

 

(-_-)
I posted hardware specs earlier. Ryzen 3600, 32GB Ram, also GTX 1060 3GB (the one that can't slowly do raytracing. Thanks nvidia).

Looking through the specs, my GPU has about 1/3rd of computing of sony's GPU (4 teraflops vs 10), less cores but higher clock, and more ram.

If someone want to try optimizing it then... well, like I said similar models are on turbosquid.

Something slightly related. Ran into this few days ago on reddit.

Spoiler

 

Pointless tests in editors but I don't think you'll get anything better in Unreal, maybe even worse:

i9 9900k, 2080 rtx

 

Typically 4.6 FPS using the HDRP on Unity 2019.3. Ryzen 1600X, 32GB RAM, GTX 1080.

 

I think he had a 1060 from previous post not a 1080, so I imagine on the same GPU HDRP would still run around 1-2 fps

I was actually hoping HDRP would be a bit more performant with high-detail geometry than that (not expecting 20+ million triangles etc but still, not as high as I was hoping for)

 

If this is the same model, what FPS do you get with 7 of them?

 

I don't know what's happening... URP, 2020.2a11




Although I only found STL file on TurboSquid, so I had to convert through 3D Paint, maybe it did some optimization?
Also the FPS is the lowest, it is usually 40-50FPS.
Ryzen9 3900X, RTX 2080, 32GB Ram.

Same in HDRP default project:

 

Some triangles are discarded (zbuffertest/backfaces) so there should be no light calculations for them.

 

At that many triangles the burden is mostly on the GPU anyway, since everything gets uploaded into a vertex buffer and off goes the poor thing.

 

Surprised you got it to 30fps in editor ninja! I wonder if there is a difference between 2019.3 and 2020.2 performance in editor...also, how does all this keep up with a dynamic camera?

Side note, I personally just dream that the new upcoming dots/artist tools (animation, terrain?, timeline, etc) focus on making them useful out the box rather than throwing them out and praying the asset store will make it useful. And it doesn't have to come out feature complete obviously, but at least that there is a drive to keep building it out features wise. This is unlike Unity UI, timeline, the animator, the animation window, etc etc where it's come out with an obvious lack of features, gets minor features and bug fixes, and then are just left alone. But things like the VFX graph give me hope that Unity see that dream too - their seemly razor sharp focus on adding modern features and building a good UX for the tool makes me happy! Hopefully that's emblematic of the mindset for the rest of Unity right now

 

You have less statues. There are 36 statues in your screen.

Mine had 150.

 

The editor started to behave like a slide projector a bit, especially when you're positioning the camera.

Okay, this was fun. Although I don't know what I'm doing, so that's it from me folks. I'm a code-monkey, I have no expertise with meshes and materials, other than drag it and drop it. And coloring red.

 

You could try this out...

https://docs.unity3d.com/ScriptReference/Graphics.DrawMeshInstanced.html

 

Why is there such a high CPU cost involved? Have you looked at the profiler, is the CPU perhaps waiting for the GPU to finish rendering? If the CPU is just waiting, then that stats window is quite misleading

 

Unfortunately I cannot build, it is crashing all the time.

And after I moved the camera during play it dropped to 10 FPS and stayed there indefinitely.
The profiler shows acute GPU bound.

 

The GPU isnt all that stressed out, its a CPU bottleneck. I wonder would what happen if you used DrawMeshInstandced

edit: ah, I see that you talk about that at the end of thread

edit2: If you added some LOD to that it might be playable. Interesting how far hardware have come

 

I think the whole conversation is missing a bit of a point when starting to compare performance with statues loaded into the editor and hitting play and so on...
The UE5 demo clearly stated, what was special about it...namely dynamic streamed geometry and textures. Unity added support for virtual textures to HDRP a while ago, which is a step in that direction, but UE5 took it one further and loads in geo while optimizing/culling/caching it in realtime. That is nothing you do by just loading 5 statues in a blank unity-scene and comparing fps and then stating "well...seemingly Unity sucks".
It's comparing apples with oranges...

The UE5 tech sounds very interesting, but until beginning of 2021, none of us will have had his/her/their fingers on it, so it doesn't make much sense to run comparisons like that (while completely neglecting the hardware plattform at hand)
I'm very interested to learn more about the tech underneath (Nanite, Lumen) because it really seems like a paradigm shift, that i can very much get behind in my line of work. It would help me simplify and speed up current workflows a lot and thus safe money/make new projects possible, that you simply cannot do just now.

If you want to have a somewhat fair number-comparison right now...take HDRP and UE4, align the Post-Effects to match, slap on Virtual texturing or whatever you like and compare those. They are quite similar in tech and you should be able to gather your insights from that, for whatever those insights are worth...

 

They are testing the impact of not having that kind of optimization. Its a bit of fun, nothing serious. Though using simplygon and reducing tris down to a few thousand for LOD2 and maybe 100k for LOD1 would be fun to see

 

Their automatic mesh reduction must be very good since it needs to be used even for "LOD0". Simplygon is good but it's not good enough for LOD0

 

I wonder how UE's indie adoption is going to affect this though. When AAA companies used UE1/2/3, breaking backwards compatibility for UE4 was fine, since there wasn't an asset store to maintain. However, now that they have to cater to indies, and asset store plugin writers, etc, etc... they have to maintain backwards compatibility. UE5 is already announced to be compatible with UE4 projects.

So yeah, I wonder if this is going to negatively affect UE as well since they also have to start maintaining compatibility from here on out. Or risk angering many plugin writers complaining they have to start from scratch.

I would also like to point out that I feel like it was kind of a wrong choice moving to a yearly versioning system, in my opinion. This DOTS/SRP/package manager transition is perfectly suited for a major version change, i.e. Unity 6.

 

I wonder how well a already manually low poly model will be treated in nanite. I mean you basically want it untouched by nanite when within whats traditionally is called LOD0

Take this BMP we have in our game for example, its awesome looking and clocking in at 72k tris



I dont have access to its high poly counterpart. I wonder how nanite will treat assets like these

 

I don't see this being very different from a consumer perspective. Unity Asset Store already has assets with separate payable versions for Built-in / URP / HDRP versions along with some assets having a 2019 compatible version and no doubt a 2020 version on the horizon. From a consumer point of view I'd be more than happy to buy an asset for UE5 if it were going to last the life cycle of UE5 and then buy the same asset again for UE6 bearing in mind UE4 is 6 year old.

 

I completely agree with that....actually, from the point of an assetStore developer/publisher...i think this clear cut is actually even appreciated, because it makes it very clear, which assets work in which version and you don't have to jump through a thousand hoops to maintain compatibility to half a dozen versions...
The forward compatibility should make it relatively easy to transition existing stuff from UE4 to 5 and thus setting a "fresh" starting point for the assets. This also is a nice filter to sieve out all the assets, that are no longer maintained by anyone (for someone here who was cheaply trying to take over this thread to advertise their half-baked products a few posts ago, this should be a concern, btw...but that just as a slight grain of salt...)

3D/Art-Assets on the other hand should be pretty easy to transition by the publishers and i don't really see a challenge. I think it is better to have fewer assets, that actually work with the specific engine than offering a false diversity of thousands of assets, that might have worked 5 years ago, but won't work without loads of work after buying now...

 

I didn't keep the project after I finished playing around with it but I'm fairly certain it's just waiting for the GPU based on how the task manager is reporting 0.4% CPU and 100% GPU.

 

I also tried my luck with the sculptures
Standard Pipeline, Instanced and without shadows on an Rtx2080:

87.7ms on the GPU is not too bad, and it is not bottlenecked by the CPU, Cpu is only waiting for the Gpu.
The performance is approximately the same without instancing (because of the low mesh count).
With soft/hard shadows it only has 1.9FPS (instead of 11.4FPS) respectively

 

They are too busy riding the slide an pigging out at the cafeteria and playing social justice warrior gambits. Read Glassdoor. After enduring blow up after blow up and losing two 40 hour weeks inn 6 days trying to recover to the state I was at..simply finishing the motorcycle sim so i can release Tail Of The Dragon I am back where i was in 5.6. I moved the wheel colliders in the Hierarchy, added to members to a List and hit play and kaboom. everything but the motorcycle disappeared into blackness and the motorcycle shot into the void.

I blame it on a bunch of slack workers whose gravy train needs to end in a hurry. I went from 2019.x series due to crashing to desktop with no bug reporter a few hundred times to 2020 which was even greater hell in a handbasket. Yes the project is big. The actual road is 11 miles..but FFS.. I turned all the terrains into Unity meshes and had 166K vertices on the screen and it frickin' crashed, the screen and editor tearing. I tried updating Azure Sky 7 times in the Package Manager and it would revert every time as though I had downloaded nothing. The Asset Store used to work fine. If it slowed your damned Editor down don't open the panel and dock it. I don't need some MS Nuget styled BS clogging my workflow if it ain't working. In furious frustration I dl-ed Unreal last night and went to bed at 3am after trying from 6:30am to salvage it. They bankrupted my company with a physics bug starting in December and sent me boilerplate responses thru to March for Cyclotronica and I missed a publisher deadline in January. They did not listen to my telling them where it was and eventually in April they wrote me and said where it was..where i ha told them i the first few of the bug reports of bout 100 I sent. The stress knocked my immune system out and I got the covid sickness for 8 weeks. I recovered and just when i was to start a 15K contract to recover financially they locked the bloody world down.

I am now downloading 2019.3.14f1 to see if I can salvage the project. In the meantime i exported all the height maps and the handmade textures of the local terrain and vegetation I photographed and ill abandon Unity to rebuild the vehicle sims and terrains and road in Unreal. I never got involved in these food fights with Unity vs Unreal in all the years prior..but to me this is not about graphics. It is about my bank account. I am too damned overqualified to even get a regular job anywhere nearby. I can build smaller games rapidly and fluently read and write C# in Unity but the stability of the engine at this point comes down to being an investor driven company who hires MS enterprise coders who fancy themselves game engine engineers and they ain't. The verbose BS that has crept into the engine should not be there. Stop with the interfaces, delegates, property driven functions or whatever the hell.I and many others are not impressed with the direction. I didn't sign up to be an MS enterprise coder. I suspect that sh*t is what is blowing the engine to hell and back. Fire these people. First make them accountable for their work product and when they inevitably fail send them back to tending servers for Witless, Shylock, Grumble and Phart, Attorney at Flaw. OK..2019.3.14f1 just installed. This either sinks or swims Unity in my books. Crashing scene with 166k vertices and less than 20 scripts is unforgivable in 2020.

This rant is far from over. If this doesn't work I am gonna hound them to hell on twitter. This is entirely irresponsible. My landlord does not give a hoot about my issues with cranky software. Maybe the CEO can send me a cheque to pay my bills. He is making bank and sending others who rely on the software for a living to the poorhouse..

Edit: So It spent about 15 minutes importing the package I made of salvageable items. I t showed an error for TextMeshPro due to the TMP package being from 202. I deleted the folder and opened it again from the Window menu. I hit import..it stalled for 30-60 seconds and is now reimporting the whole project. FFS Unity. It usually takes 3 seconds. I might as well make a cheeseburger. It's gonna be a long evening.

 

Sorry to hear about all of the headaches. Do you have a work in progress thread about all of this? I'm always interested to know about any issues with large projects, so I can try to plan around them in the future.

I have run into various "large project issues" in the past with Unity, and it can be very frustrating. Small things tested separately in little prototype projects work great. Then as I merge things together and try to polish for release, that is when things can feel risky.

 

I know how things can be rough, take care @ippdev

 

I am close to salvaging it I hope. I should start a thread about it. I am going to release it soon and the gentleman who got me the trademark to use for this game for this road knows the fellow who owns the store and trademark. It is considered one of the top ten motorcycle runs in the world according to Road and Track magazine.. 318 turns in 11 miles. I got on him just now about contacting his buddy and using his twitter, facebook and website resources to promote pre-release so we have a good rollout.. He listened to me rant for good hour+.. Mind you I get some good ones off when on a heated rant. My father was a salty tongue drill sargeant who didn't swear so he had to be highly creative when ticked off so its in the blood..LOL. I will oblige you with a WIP after i get back on a roll. Thanks for your interest.

 

Thanks. I remember your words of advice on the terrain thread I made a while back. If I get sad it kills me inside. I prefer getting furious and getting a good rant off. Make my warrior blood feel better like i can't be defeated. Damned fools floating from paycheque to paycheque. They need to get off their duffs and stabilize this crap and bring in the best minds from the community and pay them well. The guy who write MegaSplat should be their terrain engineer is one example. I used a POLM or similar named plugin a while back that spit out lovely GI maps with awesome ambient occlusion and dial in color bleed. He already had the GI stuff licked. Why wasn't his plugin incorporated. It was already working. Bad management. Give the CEO his golden parachute.

 

OK.. This is it. I am moving Tail Of The Dragon over to Unreal. I went to sleep at 2am, woke at 4:50 and tried to salvage the project. In a totally stripped scene Unity crashed over and over and over and over. I went around with bug QA a dozen emails.. Sends me this, send me that. Why?? It crashed when I open Package Manger. It crashed when I changed the UI camera. It crashed when I reset the TextMeshPro justification on the UI. It crashed when I reopened it. I rebuilt the Library, updated every goddamned driver, changed the settings from DX11 to OpenGLCore. As I sit here the Game view window is experiencing video tearing. I am done. I have all the height maps, hand made textures and UI graphic and vehicles I modeled and spent a hell of alot of time one.. They use PhysX in Unreal so I can redo the sim over there without crashes. I will resubscribe to SpeedTree for one month and re-export the handmade trees for Unreal. I will still use Unity as I see fit for small game contract work but it is obvious that no game with a financial stakeholder involved should have to endure me dealing with this crap instead of working. If Unity wants to survive they need to fire the MS enterprise coders and start hiring the best developers in the community or they are gonna be out of the game. I want to thank the voice that showed up in my email that was the final decision maker for me. They described the migration experience in a similar type of situation and it was the final weight on the scale.

 

There's conversion project in my signature that may or may not to be of use to you (there are glitches, and it can crash, but static meshes and terrain transfer should work).

Also, judging by the things that surfaced in this thread, I think it would be a good idea to focus on improving my Unreal-fu in the future. Despite Unreal having messier API.

 

What do you mean by ' In a totally stripped scene ' ? I sort of aggree with your frustration but things are not as simple as that and also It is still very unclear what it is you're trying to do. My suggestion is to create a brand new project and start incorporating your assets, code, scenes one at a time. It will not take as long as you think, were you using 'ANY' other third party products that could cause issues?

So you had a very old project done in a 5.6 and youre trying to import into non released Unity 2020. My suggestion is ' First ' Get your ' Core ' assets working and then start adding other third party assets such as Azure sky etc and what ever else you have in your project.

 

Its much better to migrate in baby steps. Thats why we always have a separate migration branch. We started on 5.2 and are now on 2019.3.14 and yeah stuff breaks and unity are doing breaking changes (like this last change that OnTriggerEnter/Exit trigger when toggle iskinimatic, totaly do not agree with Unity that that was a sane change) but it just to suck it up and fix the problems.

 

What happened that it needs "salvaging"?

I understand that the Editor is crashing, which would frustrate the heck out of me, too, but I'd just roll back to a version which wasn't crashing.

 

I think this thread is really sending the wrong message to Unity team. By all means there are problems with Unity, but to put it all in this thread implies that the problem is that Unity doesn't have the kind of cutting-edge rendering tech that Epic has been working on for over a decade, which is a given. And if you're considering switching to Unreal, you need to realize that it's a much different engine. It's great for large studios, and artists especially, but for a one man or small team with a focus on programming, dealing with blueprints and the obtrusive and confusing API is a huge hassle. The gameplay workflow in Unity really is miles better.

And over on Unreal forums there are countless complaints about bugs and documentation and lacking features as well. It is just the reality of developing such a complex piece of software. I think if Unity is able to fix the stability problems and get DOTS production ready (which is supposedly what's now being focused on) relatively soon, it will remain the best option for the vast majority of indie devs.

 

I think this thread is not about sending a message, but about people discussing things, sharing experiences and voicing their opinions.

 

This is the myth that UE4 is meant for large studios or not friendly for programmer. While I wouldn't generally recommend UE4 for 2D game or mobile game where more happens in the UI than in scene (UI can be even placed in the scene, great workflow for UI-focused games), I'd like to deal with some myths here supported with "countless complaints about everything".

UE4 was designed with indie teams in mind as Epic already filled "large studio niche" as it was possible at a time. That's why there are blueprints in UE4, visual scripting and dozens of tools working with each other out of the box. Especially if somebody working on a game when gameplay uses characters or heavily-directed content (implementing and scripting small things in every place of the game instead of relying on programmer every day). Vanilla Unity provides nothing in this regard. It took them over a decade to provide a prefab system that doesn't break easily

Actually, you could ignore my entire post and browse unreal's blog to check examples of solo devs and small studios utilizing the engine. To gain some perspective and drop this "large studios myth".

https://www.unrealengine.com/en-US/...eloped-puzzle-game-kine-using-only-blueprints
https://www.unrealengine.com/en-US/...er-koral-with-the-help-of-the-ue4-marketplace
https://www.unrealengine.com/en-US/...n-to-craft-a-hauntingly-memorable-puzzle-game
https://www.unrealengine.com/en-US/...yle-and-story-make-for-a-jrpg-dream-come-true
https://www.unrealengine.com/en-US/...how-tetris-effect-became-a-modern-work-of-art
https://www.unrealengine.com/en-US/...ackbone-s-dystopian-noir-2-5d-pixel-art-style
https://www.unrealengine.com/en-US/...tifully-inventive-art-of-creature-in-the-well
https://www.unrealengine.com/en-US/...ldlike-whimsy-to-create-a-fantastical-journey
https://www.unrealengine.com/en-US/...tension-in-the-sci-fi-horror-moons-of-madness
https://www.unrealengine.com/en-US/...e-editor-utility-widgets-to-stylize-your-game
https://www.unrealengine.com/en-US/...ing-a-powerful-story-through-ue4-environments

And the list of UE4-based used in next Xbox marketing. The first one developed by a single Chinese guy. Only the last Vampire games look like it requires a large team. Other games might look like large productions, but I worked in 10-and-less people team that achieved a similar effect, so I can't be sure
https://www.unrealengine.com/en-US/...veals-wave-of-new-unreal-engine-powered-games

-------------------
I rarely find UE4 API confusing. Most of the time is easy to understand what happens in the engine. Although the gameplay framework was designed with multiplayer support with the mind, so it's initially confusing why there are these few extra basic gameplay classes. It's clear why thinking about multiplayer.

If somebody prefers workflow-based purely on traditional programming language and "make his own engine" (by getting basic systems from the asset store and establishing his own gameplay framework), Unity is awesome. It doesn't mean though this is the only great engine for small teams.

--------------------
The reality is there many small studios ditching Unity and switching to Unreal for the next project. Got enough of fighting with an engine or waiting years for fundamental features like "nested prefabs" or upgrades to the renderer that destroy entire features (i.e. grass vanished from the landscape). Seriously, reading this thread is sometimes like reading horror stories...
I didn't yet notice any actual studio (not counting hobbyists trying many engines) going from UE4 to Unity. None.
Although being on the Unity forum, you might throw some examples, please? I might be biased.

Surprisingly (or not), teams are loyal to Unreal and no Unity update or announcement makes them consider changing the engine, open up huge threads. There are must be reasons for it?
Like engine upgrades introducing just maybe a few minor annoying bugs (even if you need touch engine's source to fix, you can do it easily), not wreaking havoc like here. And causing constant stress to developer "will update crash my features/tools/game?" or "which engine version is stable enough?". Well, UE4 doesn't provide even LTS since is generally stable in every version. And new huge systems are worked as preview/experimental for a year or two. Mostly working well even before marking it "production-ready". Artists often used beta features. Environment Query System is marked as "experimental" for 5 years although dozens of projects use it, no issues. Author of the system just wasn't satisfied enough it, left "experimental" mark on it. This a difference in the software quality provided to the developer we're talking about.

Epic doesn't need to rewrite the entire engine and throw away dozens of tools to let you create performant games. Unity's MegaCity demo could be easily recreated with any UE4 version (OOP tools would handle that just fine, can't see anything that could prevent that).

The gap is so huge I usually ignore job offers from studios working with Unity. Even messy in-house engines are less stressful to work than Unity. I'm not interested in 2D or mobile games but also not huge AAA productions too

--------------------
PS I do hope I'm not seen here as someone "selling UE4" or something. Just entered discussion after reading some incorrect technical information, i.e. "there's one Resharper for C++ and C++ is a nightmare" - yes, pure C++ without tooling is difficult, but we have tools and UE4 doesn't make you use pure C++.
Or again and again "UE4 is for large studios". While UE4 is commonly used one-man army or few people team, achieving artistic and commercial success. Even if programmers need to accept anything else than C#

 

If I can add my little experience, im with MothDoctor, I test this week ago some of UE 4.25 and look for tutorials about "Unreal C++" because i want max control BUT in a fast way, then I see some features that was a WTF:

Because I face many issues with Unity I check the worst cases, in my case, animation (multiLayer, change animation masks, animation transitions, Shaders (Water, Grass, hello HDRP...), Multiplayer so I try to check if Unreal have that problems... and what I do is:

- I create a C++ Unreal proyect, then generate the Visual Studio files with right click, after some seconds I open the Unreal project, and the sln file with Visual, and after some youtube tutorials and able to create a class, make functions in C++ (sound weird, but comming from C# and PHP sound.. familiar...) and i can expose these params and functions to the blueprint, I press compile, take 4 seconds, and the variables and function are available to blueprint, is that really easy...
Another thing that I see is the C++ function by default, have MANY functions to help you to work, is like someone tell the Unreal people the functions you need to work with transforms, quaternions, collisions, etc, and you have exposed functions that already make this for you and you dont need to search in Stackoverflow how to solve it.

- Go to unreal market, look for the first water shader, grass shader and PBR FREE textures I found, import into Unreal simply go to the asset, purchase (free), and press the button "add to project" select it from the list, and when ask because is not for my actual version to use "nearest version", then import. This make appear a folder in my Unreal project in 10 seconds, now enter in this folder, pick some grass BP, water BP or texture and drop in the scene... you know what? WORKS simply WORKS, i get in 1 minute free grass with wind AND collision detection working in max setting project, and the water? reflective, refraction, displacement working dropping in the scene, no f***s, no issues, same for the material, i can add it to any object dropping on it without care about the size of the texture.

- Also the animator work with blueprint to, you can simply drag the animations in something similar to the Unity 2D matrix to get movement mixed animations, or add conditionals easy to every animation, and import it is extreme easy, no issues like i have in Unity like keyframe issues (maybe this is my fault, but I have many issues with Mixamo animations in Unity)

- Then I jump to Multiplayer examples, Gamestate, Player, Pawn.. is stupid easy sync object with many conditionals to sync this or this or this only in specific cases, and is Server Imperative, you know what pain is this in Unity? The last time I do (or try to do) a multiplayer game in unity, i test Proton, UNET, and some other libraries, and after many problems (and waste like 1 week making my own lag compensation component for unity, to make a decent way to get a real time multiplayer with Extrapolation and Interpolation based on player-server lag) i leave for impossible to do and test all myself, not do the class, make a class and have time to be sure have no issues/bugs.

- Then I try to test the terrain due all issues I have with Unity Terrain and.. what say.. you know who the Artist love it, the tools are year ahead of Unity terrain, and you can place grass in massive way without kill the game performance.

I check more tutorials about make AI (im think how to change my actual 1500 lines AI to a mix of blueprints to the logic and C++ function for the things like raycast and the pathfinding using multiple threads, and have little orgasms)

But i find some things that I dont like:

- Unreal Editor is really heavy, also compile the shaders can take some minutes.
- The integration of the Unreal C++ and Visual Studio 2019 is.. bad? I dont know but is EXTREME slow the intelisense, and some times make weird things, like I forgot add the #import of some class, and the error is not in the class declaration, is in the properties.. wtf? In this case Unity C# and Visual work really fast, you can change C# code, and in the time you need to change the window the code is compiled and Unity refreshed.
- The Blueprints are more intuitive than the lateral panel, but in some cases use to many screen, i mean, in Unity you can select a object in the scene, and with 25% of the screen, the lateral column show all the info you need, with Blueprint you need other tab to work fast (I have only one screen )
- The "generate package for XXX platform", what I think is the Unity similar to "Build & Run", take AGES, i mean, you can have your CPU al 100% like 10 minutes in a simple project, in Unity i have a project with like 500 assets and take 3 minutes build USING IL2CPP (Ryzen 7 3700x, 32Gb Ram, RTX 2060)
- The base resource consuption is also big, in Unity my project with this quantity of assets and a map of 4000x4000 consume at 1080p with many post effects arround 600mb of Ram, the test Scene Im using, with grass, water, particles and so in Unreal consume 1Gb of Ram. The disk usage is another, 100Mb Unity, 300Mb Unreal, so I think Unreal create many versions of the shaders, textures, etc.
- Other thing I see, but this is maybe my initial experience is with Unity, i feel like Unreal use more "base block" for the things, i mean, with Unity you have the minimun empty object, and you add components to them, Unreal work at inverse, you have "all already" pieces, and if you dont use some part, you need to fight to remove them.

This is what i feel in one week of test things by myself, obviously is a cropped vision, I need to test ALL (and make more complex things in C++) to have a real vision, but initially feels like I say in some pages ago, feels like Unity is a Developer program done for Developers, and Unreal is a program done for developers to people that WANT to make games, and dont care about nothing more.

Ofcourse, for UI and 2D games, for me, Unity win, the options that Unity have right now for UI and 2D is ahead of what Unreal have, and when they release the CSS compatibility for the UI... grrr...

 

I love Visual Studio. Visual Studio is fantastic for creating small to medium projects. That said once you step into large project territory you quickly run into the exact problem you're having. IntelliSense isn't slow because of a fault in UE. It's slow because it was never designed to handle a project the size of a game engine plus the game itself.

You only have one choice if you want to use Visual Studio with Unreal. You need to have ReSharper.

https://www.jetbrains.com/resharper/