How is this going to impact Environment Art? I started 3 years ago, got a job last year and I just got to the point where I consider myself good to really good. I can't draw and don't know anatomy so most of the stuff I make is real world stuff from references and concept art, especially guns. Also, I really like doing these kinds of objects, really understand how they work and adapt them into a game asset, even if it's just a humble TV. I don't like the idea of scans replacing what I do. What use is there for me if you can just scan all the parts of a gun or prop and make a perfect game asset of it?
I realize that photogrammetry can't be used for stuff that doesn't exist but won't it basically cut a lot of the jobs and only the most experienced will get what's left? I think I'm pretty good but I don't have years of experience under my belt and I'm afraid I won't be able to catch up.
You put quite some emphasis on the fact that you know stuff. Why do you worry, then? Photogrammetry already exists within the gaming industry, and you still have a job.
I just realised that, if everything runs so much faster in UE5, does that mean we are free to use many more drawcalls? Because I would rather spend my budget there, than in high poly meshes
Nanite gets 1 draw call per material, no matter how many instances or variations in a scene. That'll save you a lot of draw calls to use else where.
How is this going to impact Environment Art? I started 3 years ago, got a job last year and I just got to the point where I consider myself good to really good. I can't draw and don't know anatomy so most of the stuff I make is real world stuff from references and concept art, especially guns. Also, I really like doing these kinds of objects, really understand how they work and adapt them into a game asset, even if it's just a humble TV. I don't like the idea of scans replacing what I do. What use is there for me if you can just scan all the parts of a gun or prop and make a perfect game asset of it?
I realize that photogrammetry can't be used for stuff that doesn't exist but won't it basically cut a lot of the jobs and only the most experienced will get what's left? I think I'm pretty good but I don't have years of experience under my belt and I'm afraid I won't be able to catch up.
what exactly are you afraid of? photogrammetry is already heavy in use for realistic productions.
you can't scan everything, there will still be plenty of work.
If a production is using nanite, you could skip some baking and working directly on your highpolies. baking times will possibly be replaced by converting to nanite times.
We will see if simlar features will catch on with other engines. But even if it does not all productions will use it, just look at the size of that demo by epic games, a hundred gigabytes for very little content, the fastest ssd wont help your streaming that if you dont have the storage for more than one actual game.
Scans are more work than making stuff by hand unless you buy them - in which case you're limited to whatever megascans are flogging. You process as an environment artist will remain basically the same.
Larry - you still have to worry about the overall drawcall limit on your target platform (for xbone that's about 14 ) however... Nanite instances very well so you can basically use as many copies of a given nanite mesh as you want without needing to worry about drawcalls. RVT goes a long way to helping with drawcalls. The smart batching stuff they put in fairly recently (4.24?) seems to be somewhat effective although it still lags behind a couple of the proprietary engines I've worked with
How is this going to impact Environment Art? I started 3 years ago, got a job last year and I just got to the point where I consider myself good to really good. I can't draw and don't know anatomy so most of the stuff I make is real world stuff from references and concept art, especially guns. Also, I really like doing these kinds of objects, really understand how they work and adapt them into a game asset, even if it's just a humble TV. I don't like the idea of scans replacing what I do. What use is there for me if you can just scan all the parts of a gun or prop and make a perfect game asset of it?
I realize that photogrammetry can't be used for stuff that doesn't exist but won't it basically cut a lot of the jobs and only the most experienced will get what's left? I think I'm pretty good but I don't have years of experience under my belt and I'm afraid I won't be able to catch up.
Photogrammetry is just a tool and quite a welcome one too to be honest. Your skills aren't going to become useless overnight and you will have plenty of time to learn more so I wouldn't worry about it so much. Painters/sculptors never went away just because photography was invented, if anything it just gave them more room to play.
How is this going to impact Environment Art? I started 3 years ago, got a job last year and I just got to the point where I consider myself good to really good. I can't draw and don't know anatomy so most of the stuff I make is real world stuff from references and concept art, especially guns. Also, I really like doing these kinds of objects, really understand how they work and adapt them into a game asset, even if it's just a humble TV. I don't like the idea of scans replacing what I do. What use is there for me if you can just scan all the parts of a gun or prop and make a perfect game asset of it?
I realize that photogrammetry can't be used for stuff that doesn't exist but won't it basically cut a lot of the jobs and only the most experienced will get what's left? I think I'm pretty good but I don't have years of experience under my belt and I'm afraid I won't be able to catch up.
Unreal 5 isn't about scans. Nanite means you don't have "lowpoly" geometry anymore. Scans are a quick way to get hipoly geometry for this type of environment but there's plenty of assets in the Unreal demo level that aren't scans.
You can scan a gun but it generally will look a bit worse than a good subdivision surface model.
I really wish Epic would have done a demo with a detailed scifi scene or some clean subdivision models. The type of assets done by Tor Frick and Paul Pepera would work extremely well. People are getting the wrong impression Nanite is only good for sculpts or photogrammetry, which is far from how it actually works.
"I just realised that, if everything runs so much faster in UE5"
Just because an engine has a way to process and load assets with xxxx billions of polies doesn't mean that it is faster at everything. An optimized game will always run faster than a messily thrown together one, regardless of new engine features. A poorly optimized game running badly in UE4 will still run badly in UE5 (unless the assets from the poorly running game happen to be good candidates for the new optimization techniques, which is unlikely to be the case). As a matter of fact, the UE5EA in-editor performance seems similar or even a little bit lower than current versions of UE4 at least on my machine.
In short : don't panic, do clean work, it'll be alright.
Just wanted to say thank you all for chiming in on the state of things. It was nice to read through everyone's posts from many devs that I know have been in this game for an exceptionally long haul. Optimization is king, and while I look forward to what UE5 will bring, I'm happy to wait a year or two until an actual game can be fully developed in this engine with as many of its new features made accessible to be fully utilized across multiple platforms. Being able to use these features for film and TV is one thing. Being able to use these features on every platform available to game developers is entirely different. There's so much more behind the scenes driving a game than just the visuals on screen. AI, physics, networking, and more that truly impact performance and that isn't as strong of a development balancing challenge when working in film. In UE5's EA state there are much harsher cut off points that dictate what can be used as part of a game's global design than there are now in uE4, and if I have to make harder decisions on the design of the game because it spits out crap for crap FPS on a lower end platform, then I'm not able to bring my title to its full realization and I'm limiting my reach. In the end I'm saying that one customer gets all the juicy goodness and the other customer, well, they should just buy a better rig or just not buy the game. Maybe that's okay for AAA studios. They can take the hit and they have the resources to squeeze on specific target platforms and get the best return. As an Indie, you can't survive like that and you need to ship your title to as many platforms as possible with as many features as you can that run in parity to every other platform you released your title on. They call it early access for a reason and it was refreshing to know many developers are not fooled by the current hype that many inexperienced devs are simply not realizing. It is 100% hype worth promoting because it does show an amazing path forward years down the road, but it doesn't change that a game should be heavily optimized in every way to run as smooth as possible, fully featured, on multiple platforms. Until then, nap time is strongly encouraged.
TLDR; looks insanely impressive but isn't game dev ready and limits indies in multiple ways.
Triple-A is the one with hard limitations - if you want to make your money back you have to be on console and you have to make do with the hardware capabilities and the file size limits for distribution media, store fronts and maximum patch download sizes on these machines.
I'm not up to date on current TRC's but I imagine filling up the entire PS5 SSD with some rock sculpts isn't going to fly.
in short nanite is allowing me to cut out a couple tedious steps for certain art assets, thus simplifying workflow.
Very empowering for indie developer.
I imagine for large teams full of specialist that sort of improvement only helps increase fidelity but doesn't make a major impact on any persons day-to-day workload. In the end tech can't save you. Only united voice of the people can accomplish what anybody really wants. Until then, whenever technology saves you work it just means your disconnected boss will now feel justified in asking you to do more.
But for us basement dwelling game makers nanite is only a good thing. It's a bigger stick so we can get more termites from the mound more quickly. That's all.
The new TAA is so much better, just out of the box, sharper, less ghosting, finally feels more aligned with the best TAA currently in games.
TSR (Temporal super resolution) is amazing as well, at 50% resolution it looks basically full resolution except some pixelization around the edges of objects when there's movement. I don't have a graphics card with DLSS, but I imagine it's pretty close to that magic.
Virtual shadow maps, if they can fix the aliasing, these shadows are fantastic, completely eliminate the need for having contact shadows, multiple shadow cascades, and far shadows. Feel really close to ray traced shadows in features and quality. The performance impact is halved when using DX12 in editor and game. Also part of what makes Nanite detail look better than normal maps.
The new TAA is so much better, just out of the box, sharper, less ghosting, finally feels more aligned with the best TAA currently in games.
TSR (Temporal super resolution) is amazing as well, at 50% resolution it looks basically full resolution except some pixelization around the edges of objects when there's movement. I don't have a graphics card with DLSS, but I imagine it's pretty close to that magic.
Virtual shadow maps, if they can fix the aliasing, these shadows are fantastic, completely eliminate the need for having contact shadows, multiple shadow cascades, and far shadows. Feel really close to ray traced shadows in features and quality. The performance impact is halved when using DX12 in editor and game. Also part of what makes Nanite detail look better than normal maps.
This! Also, better polygon handling means, you can populate your scene with (optimised) rubble without worrying too much about their numbers
TLDR; looks insanely impressive but isn't game dev ready and limits indies in multiple ways.
It seems to me that this update is more aimed at the virtual production and visualization industries. The techniques ILM used on The Mandalorian has gotten the film industry very interested in game engines - and Epic wants to be the company associated with that.
How well does this scale? In true Epic fashion, the demo runs horribly on everything except the latest high-end hardware, but they have always done that. Remember the Kite demo? The question I have is, how does the demo run if you turn off the fancy TAA, turn off the new shadowing(if possible) and go back to CSM shadows, downscale those 4k/8k textures, and turn down the Lumen samples(or just turn it off)?
How well does this scale? In true Epic fashion, the demo runs horribly on everything except the latest high-end hardware, but they have always done that. Remember the Kite demo? The question I have is, how does the demo run if you turn off the fancy TAA, turn off the new shadowing(if possible) and go back to CSM shadows, downscale those 4k/8k textures, and turn down the Lumen samples(or just turn it off)?
But that is simple product advertising, it just shows how far you can go with just commercial gear. Previously, or currently with all other software, an average PC can handle around 20mil raw vertices. Huge step forward, and that is what I love about CGI stuff. You can keep what you need and discard the rest.
I really wish Epic would have done a demo with a detailed scifi scene or some clean subdivision models. The type of assets done by Tor Frick and Paul Pepera would work extremely well. People are getting the wrong impression Nanite is only good for sculpts or photogrammetry, which is far from how it actually works.
Hey I noticed this today - in this interview Brian Karis (epic graphics programmer) specifically calls out how nanite also works for hard surface mechanical models as well as rougher models and scans. (I've set the video to play specifically where he talks about it)
If you can afford the base cost of the fancy new systems they allow you to go a lot further than you could without them .
Eg. I suspect you'll have trouble trying to shove an RVT, nanite and lumen based game up a switch but if you're working to a target platform that can support them they allow you to shove a shit load more content on screen than you'd be able to without them
If you can afford the base cost of the fancy new systems they allow you to go a lot further than you could without them .
Eg. I suspect you'll have trouble trying to shove an RVT, nanite and lumen based game up a switch but if you're working to a target platform that can support them they allow you to shove a shit load more content on screen than you'd be able to without them
You can't even get a XBOX one/ps4 game on a switch without extensive optimization. So that's not surprising.
If you can afford the base cost of the fancy new systems they allow you to go a lot further than you could without them .
Eg. I suspect you'll have trouble trying to shove an RVT, nanite and lumen based game up a switch but if you're working to a target platform that can support them they allow you to shove a shit load more content on screen than you'd be able to without them
You can't even get a XBOX one/ps4 game on a switch without extensive optimization. So that's not surprising.
and good programmers that know how to utilize an engine extensively and allocate intelligently. This is just one of those hardware behind the software situations, just with a massive gap during a very strange time when hardware development hit a production wall like a ton of bricks affecting multiple sectors and software devs went full throttle making some incredible breakthroughs. Still, let's hope that lot check is prepared for this once their hardware is up to snuff. I can see the UltraGB packages getting massive kickback already, hence the nanite compression to the rescue.
Loving ue5 so far. From the docs here: https://docs.unrealengine.com/5.0/en-US/RenderingFeatures/Nanite/ The nanite rock example uses Base Color and Normal map only. What about roughness? Then I am thinking of using vertex color for base color of the nanite mesh which will reduce textures as well as the alpha channel of the vector for roughness.
How is this going to impact Environment Art? I started 3 years ago, got a job last year and I just got to the point where I consider myself good to really good. I can't draw and don't know anatomy so most of the stuff I make is real world stuff from references and concept art, especially guns. Also, I really like doing these kinds of objects, really understand how they work and adapt them into a game asset, even if it's just a humble TV. I don't like the idea of scans replacing what I do. What use is there for me if you can just scan all the parts of a gun or prop and make a perfect game asset of it?
I realize that photogrammetry can't be used for stuff that doesn't exist but won't it basically cut a lot of the jobs and only the most experienced will get what's left? I think I'm pretty good but I don't have years of experience under my belt and I'm afraid I won't be able to catch up.
Photogrammetry is just a tool and quite a welcome one too to be honest. Your skills aren't going to become useless overnight and you will have plenty of time to learn more so I wouldn't worry about it so much. Painters/sculptors never went away just because photography was invented, if anything it just gave them more room to play.
Yes, I understand and I do think the technology is cool and the results are amazing BUT I don't like what it represents. What is the future for 3D artists then? Are we going to spin around a model 100 times and take pictures and then do a lot of grunt work to make it game-ready? One of the reasons I started working in this industry is because I can make something from scratch and at the end of the day everything looks good because I made the right decisions and not because I spinned around a model 100 times. And I love doing even really simple stuff like barrels and TVs. I want to make my own medieval wall pieces and bring them to life myself.... And yes, I understand that not everything can be scanned but I want to create realistic stuff too. For example, I haven't had the opportunity to work on medieval stuff yet and I really want to. I really like creating a model from a simple cube, adding wear in zbrush and then bringing it to life layer by layer in Painter....It will never be as good as a scan but it was made by a human. A computer will never be able to appreciate all the work that goes into an asset but for me part of good art is appreciating the skill of the artist...
I suppose Hard Surface sci fi stuff is "safe" but what if they develop some GAN software that takes a look at a concept art and then automatically makes modular 3D models with perfect textures and with model variations on top of that....
The human caused sixth mass extinction is depressing.
A few humans in a niche market maybe needing to diversify skillset a bit some time in the future because technologies... that's just normal life. Soldier, be strong!
How is this going to impact Environment Art? I started 3 years ago, got a job last year and I just got to the point where I consider myself good to really good. I can't draw and don't know anatomy so most of the stuff I make is real world stuff from references and concept art, especially guns. Also, I really like doing these kinds of objects, really understand how they work and adapt them into a game asset, even if it's just a humble TV. I don't like the idea of scans replacing what I do. What use is there for me if you can just scan all the parts of a gun or prop and make a perfect game asset of it?
I realize that photogrammetry can't be used for stuff that doesn't exist but won't it basically cut a lot of the jobs and only the most experienced will get what's left? I think I'm pretty good but I don't have years of experience under my belt and I'm afraid I won't be able to catch up.
Photogrammetry is just a tool and quite a welcome one too to be honest. Your skills aren't going to become useless overnight and you will have plenty of time to learn more so I wouldn't worry about it so much. Painters/sculptors never went away just because photography was invented, if anything it just gave them more room to play.
Yes, I understand and I do think the technology is cool and the results are amazing BUT I don't like what it represents. What is the future for 3D artists then? Are we going to spin around a model 100 times and take pictures and then do a lot of grunt work to make it game-ready? One of the reasons I started working in this industry is because I can make something from scratch and at the end of the day everything looks good because I made the right decisions and not because I spinned around a model 100 times. And I love doing even really simple stuff like barrels and TVs. I want to make my own medieval wall pieces and bring them to life myself.... And yes, I understand that not everything can be scanned but I want to create realistic stuff too. For example, I haven't had the opportunity to work on medieval stuff yet and I really want to. I really like creating a model from a simple cube, adding wear in zbrush and then bringing it to life layer by layer in Painter....It will never be as good as a scan but it was made by a human. A computer will never be able to appreciate all the work that goes into an asset but for me part of good art is appreciating the skill of the artist...
I suppose Hard Surface sci fi stuff is "safe" but what if they develop some GAN software that takes a look at a concept art and then automatically makes modular 3D models with perfect textures and with model variations on top of that....
All of this is just really depressing to me...
Don't worry, machines will be spinning the models and taking the pictures too. In fact, it's easier for them to do that. Joking aside, Why are you depressing about a problem that does not exist yet and might not even exist? Even if you think you are not good enough for this industry, it seems like a cheap excuse to get out of this and try something else. You will always have to work hard no matter what you do, and nobody said this industry is easy. My conclusion is that, "grunt work" is done by machines. Humans add the "human touch". Eventually it is A LOT cheaper to make something in 3d than in real life and then scan it. Otherwise everybody would be prototyping stuff, patends, ideas etc. The technology you are talking about is not going to come in our generation. It won't appear over night (mostly for monetizing purposes),so you can rest assured that you would be able to gradually adapt with it, and if it "suddenly appears" your years of experience in the field will give you an advantage over fresh people. Plus, realism is just one end of the spectrum, and only a few companies go for AAA realistic stuff. Everyone else plays in the styles between, because it is more creative
This thread gives a pretty good taste of what it must have been like around the time photography appeared (or any technological advances really) - with some people panicking because jerbs, some predicting wild things without any practical knowledge of the topic at hand, and some rejoicing at the new possibilities and opportunities
I suppose another way to gain a little bit of perspective would be to replace "3d enthusiast" in any of the posts above by "scale model enthusiast", since the technological barrier to entry to CG is lower than ever and information on how to use the tools is extremely easy to access, so in many ways the two fields are very comparable now, with level of detail not being a limiting factor anymore.
To anyone hoping that a tech allowing you to load subdiv models will make everything faster at the click of a magic button : just take the time to try and texture a highpoly model, and you'll understand than not everything is a simple as "baking and retopo = bad".
Also, a higher possible level of detail doesn't mean that indie games will all suddenly look like treeple-ay games ; it just means that any production that doesn't have solid art and tech direction keeping the levels of details firmly in check will likely end up wasting a lot of time with always increasing datasets. More details never means less time, and even with stylized/simplified it is already barely sustainable, requiring more and more outsourcing.
Thank you for keeping this thread up with the latest insights, Oglu. I would love to get on that nanite action at some point (all of the dirt clumps/paper clips/dog hair I could splash everywhere is enticing) and I'm really looking forward to this version of the engine coming out of EA. I truly hope Nintendo ups their hardware game beyond sleeker displays in the future.... fingers crossed?
I read it up to page 84 so far and, i get the feeling like there is a giant push for realism and for me i haven't seen the hype from would be buyers or fans of artwork in general (for even more realism). Probably comes from the head space of we are Living it, people say life is boring and with whats happening by the worst of society it is even more so currently.
Also my concern is for what we already know, like hand paints works, stylized works and the like that do use the "longer methods".
The technical talk though is great since i am learning new things so thanks for sharing the Siggraph Paper. (still haven't installed or downloaded the ue5) for two reasons i can't use the new features and i really, really dislike signing up for anything that claims to be free. (just rubs me wrong in all sorts of ways, but who cares what i think right.)#To be clearer, i just dabble in this "community" doesn't seem to exist anymore around the art world as expected anything "they" touch is ruined.
Looking forward though to look at all the things others do with it, i only saw people flooding youtube with copies upon copies of david statues, ha the coincidence is amusing to me.
Since the beginning in 2020, people said this was marketing, this was not possible without certain problems or "1tb" game sizes. I responded with logic and what was seen, with the information available and then the demo was released, it was all confirmed and all the people saying "i'm mr engineer with experience and this is fake, certified" were completely wrong, which i was sure they were, only 20 years of gaming, reading about hardware and graphic software and some very few of 3D art were enough to see that the Demo was exactly what Epic implied and what was seen, no conspiracy theory from Epic games.
Another reality strike, CDPR next Witcher, Tomb Rider, Gears, MMORPGs with already ongoing projects and many more games are going to be made in Unreal 5, some of these MAJOR titles and companies ditching their own engines that they worked very hard to make. This is the reality, this is the real industry talking to you, you like it or not "professional". As with other "situations" in the world, credentials sometimes mean nothing, actual facts and logic, knowledge, is what matters., no argument is better than other based on who makes it and his credentials. The guys with the credentials were mostly wrong, and someone showing his super duper important credentials told me "don't fall for social media hype" (i don't even use social media) and social media was in fact in the same tone "this is marketing again, they did it with unreal 3 and 4".
The credentialists and technocrats were wrong. Very few actually looking at what was presented and using logic, going around reading and reading other people with experience, the tech existing and all kind of data, we were right. But i knew this was a huge change when the first video appeared in April 2020.
Also, some people who post in this forums is working in an UE5 game, people with the highest lvl for making things happen and look great, and the game is looking amazing.
tbh wish there was a way to update easier, 'being' NOT having to download 20gb's again every update, but the updated files themselves, i get it, its compiled, still sucks to dl 20GB+, infinitely while i am here might as well add more of "my" dislikes, i get you want to "protect" your product, with all this signup, b.s. when in the past there was no sign up bs before and we all got along swimmingly with UDK before all these "new" things, is kind of dumb to me and just shines a sun on what the real reasoning's probably are which is just to stick everyone online imho-(these investors are everywhere). So anyone got an updater somewhere on the net where one doesn't have to download 20gb+s? Alt methods pm here or on station, everyone knows unreal its not like they will get away with anything...Also Thanks for sharing "epic", appreciate the learning utilities.
FWIW : the launcher is only needed for the initial download and install. Once the engine is installed there is no need to login to anything ever (outisde of things requiring a download of course, like downloading Quixel assets or marketplace products). The launcher is not even needed to ... launch projects.
Also and just like any OS, there no need to always be on the latest anyway. On top of that 4.7 4.27 is still much more fluid to work with, as it has a much more sensical Content Browser behavior as opposed to the UX mess they introduced in 5.
Unreal 4.27 is also good if you're trying to do something non-photoreal and stylised. There's no point having the latest and greatest lighting system if you're just going to ignore it.
I have been using ue5 in a lot of different projects since its release. There were a few issues regarding professional work. Can't say for 5.1 yet as I just started using it. The upside is that you always get a wow effect from the clients when they even see your grayboxing with lumen. The same projects look way more professional and much better quality even with bad materials, without spending time to get that 'detail lighting vs smoothing' bakes. BUT!
You can't easily render professional work stuff due to some 'firefly' bugs in lumen. It breaks light continuity and it's never the same if you render it twice. It's good during production but not acceptable for final delivery
Nanite had (at least in 5.0) no control in decimating, resulting in problems with material IDs going all over the place. For complex product work (ads) even close ups might be having some sort of decimation that breaks mat IDs
"virtually limitless' triangles is still not the case, they still need to be manageable, no surpassing 2mil for a hero prop, and that still ends up with small-scale projects being around 40gb (uncooked).
At least one of the things we definitely need is better storage technology to be able to keep up with the graphic quality rising.
5 is still a bit shaky and neither lumen or nanite are anywhere close to finished but 4.27 is no better in terms of stability/bugs (worse probably) and doesn't feature any of the big improvements that actually help with development (extensions to subsystems, OFPA, world partition + more) .
Having used both in production I'd suggest it makes sense to go with 5 unless you plan to use static lighting - in which case you won't be able to use world partition or (afaik) OFPA til "after 5.2" which is not a date you'd want to schedule against
Replies
Larry - you still have to worry about the overall drawcall limit on your target platform (for xbone that's about 14
however...
Nanite instances very well so you can basically use as many copies of a given nanite mesh as you want without needing to worry about drawcalls.
RVT goes a long way to helping with drawcalls.
The smart batching stuff they put in fairly recently (4.24?) seems to be somewhat effective although it still lags behind a couple of the proprietary engines I've worked with
You can scan a gun but it generally will look a bit worse than a good subdivision surface model.
Just because an engine has a way to process and load assets with xxxx billions of polies doesn't mean that it is faster at everything. An optimized game will always run faster than a messily thrown together one, regardless of new engine features. A poorly optimized game running badly in UE4 will still run badly in UE5 (unless the assets from the poorly running game happen to be good candidates for the new optimization techniques, which is unlikely to be the case). As a matter of fact, the UE5EA in-editor performance seems similar or even a little bit lower than current versions of UE4 at least on my machine.
In short : don't panic, do clean work, it'll be alright.
Very empowering for indie developer.
I imagine for large teams full of specialist that sort of improvement only helps increase fidelity but doesn't make a major impact on any persons day-to-day workload. In the end tech can't save you. Only united voice of the people can accomplish what anybody really wants. Until then, whenever technology saves you work it just means your disconnected boss will now feel justified in asking you to do more.
But for us basement dwelling game makers nanite is only a good thing. It's a bigger stick so we can get more termites from the mound more quickly. That's all.
The new TAA is so much better, just out of the box, sharper, less ghosting, finally feels more aligned with the best TAA currently in games.
TSR (Temporal super resolution) is amazing as well, at 50% resolution it looks basically full resolution except some pixelization around the edges of objects when there's movement. I don't have a graphics card with DLSS, but I imagine it's pretty close to that magic.
Virtual shadow maps, if they can fix the aliasing, these shadows are fantastic, completely eliminate the need for having contact shadows, multiple shadow cascades, and far shadows. Feel really close to ray traced shadows in features and quality. The performance impact is halved when using DX12 in editor and game. Also part of what makes Nanite detail look better than normal maps.
Also, better polygon handling means, you can populate your scene with (optimised) rubble without worrying too much about their numbers
Eg. I suspect you'll have trouble trying to shove an RVT, nanite and lumen based game up a switch but if you're working to a target platform that can support them they allow you to shove a shit load more content on screen than you'd be able to without them
The nanite rock example uses Base Color and Normal map only. What about roughness?
Then I am thinking of using vertex color for base color of the nanite mesh which will reduce textures as well as the alpha channel of the vector for roughness.
Yes, I understand and I do think the technology is cool and the results are amazing BUT I don't like what it represents. What is the future for 3D artists then? Are we going to spin around a model 100 times and take pictures and then do a lot of grunt work to make it game-ready? One of the reasons I started working in this industry is because I can make something from scratch and at the end of the day everything looks good because I made the right decisions and not because I spinned around a model 100 times. And I love doing even really simple stuff like barrels and TVs. I want to make my own medieval wall pieces and bring them to life myself.... And yes, I understand that not everything can be scanned but I want to create realistic stuff too. For example, I haven't had the opportunity to work on medieval stuff yet and I really want to. I really like creating a model from a simple cube, adding wear in zbrush and then bringing it to life layer by layer in Painter....It will never be as good as a scan but it was made by a human. A computer will never be able to appreciate all the work that goes into an asset but for me part of good art is appreciating the skill of the artist...
I suppose Hard Surface sci fi stuff is "safe" but what if they develop some GAN software that takes a look at a concept art and then automatically makes modular 3D models with perfect textures and with model variations on top of that....
All of this is just really depressing to me...
A few humans in a niche market maybe needing to diversify skillset a bit some time in the future because technologies... that's just normal life. Soldier, be strong!
Joking aside, Why are you depressing about a problem that does not exist yet and might not even exist? Even if you think you are not good enough for this industry, it seems like a cheap excuse to get out of this and try something else. You will always have to work hard no matter what you do, and nobody said this industry is easy.
My conclusion is that, "grunt work" is done by machines. Humans add the "human touch". Eventually it is A LOT cheaper to make something in 3d than in real life and then scan it. Otherwise everybody would be prototyping stuff, patends, ideas etc. The technology you are talking about is not going to come in our generation. It won't appear over night (mostly for monetizing purposes),so you can rest assured that you would be able to gradually adapt with it, and if it "suddenly appears" your years of experience in the field will give you an advantage over fresh people.
Plus, realism is just one end of the spectrum, and only a few companies go for AAA realistic stuff. Everyone else plays in the styles between, because it is more creative
I suppose another way to gain a little bit of perspective would be to replace "3d enthusiast" in any of the posts above by "scale model enthusiast", since the technological barrier to entry to CG is lower than ever and information on how to use the tools is extremely easy to access, so in many ways the two fields are very comparable now, with level of detail not being a limiting factor anymore.
To anyone hoping that a tech allowing you to load subdiv models will make everything faster at the click of a magic button : just take the time to try and texture a highpoly model, and you'll understand than not everything is a simple as "baking and retopo = bad".
Also, a higher possible level of detail doesn't mean that indie games will all suddenly look like treeple-ay games ; it just means that any production that doesn't have solid art and tech direction keeping the levels of details firmly in check will likely end up wasting a lot of time with always increasing datasets. More details never means less time, and even with stylized/simplified it is already barely sustainable, requiring more and more outsourcing.
http://advances.realtimerendering.com/s2021/Karis_Nanite_SIGGRAPH_Advances_2021_final.pdf
Also my concern is for what we already know, like hand paints works, stylized works and the like that do use the "longer methods".
The technical talk though is great since i am learning new things so thanks for sharing the Siggraph Paper. (still haven't installed or downloaded the ue5) for two reasons i can't use the new features and i really, really dislike signing up for anything that claims to be free. (just rubs me wrong in all sorts of ways, but who cares what i think right.)#To be clearer, i just dabble in this "community" doesn't seem to exist anymore around the art world as expected anything "they" touch is ruined.
Looking forward though to look at all the things others do with it, i only saw people flooding youtube with copies upon copies of david statues, ha the coincidence is amusing to me.
Since the beginning in 2020, people said this was marketing, this was not possible without certain problems or "1tb" game sizes. I responded with logic and what was seen, with the information available and then the demo was released, it was all confirmed and all the people saying "i'm mr engineer with experience and this is fake, certified" were completely wrong, which i was sure they were, only 20 years of gaming, reading about hardware and graphic software and some very few of 3D art were enough to see that the Demo was exactly what Epic implied and what was seen, no conspiracy theory from Epic games.
Another reality strike, CDPR next Witcher, Tomb Rider, Gears, MMORPGs with already ongoing projects and many more games are going to be made in Unreal 5, some of these MAJOR titles and companies ditching their own engines that they worked very hard to make. This is the reality, this is the real industry talking to you, you like it or not "professional". As with other "situations" in the world, credentials sometimes mean nothing, actual facts and logic, knowledge, is what matters., no argument is better than other based on who makes it and his credentials. The guys with the credentials were mostly wrong, and someone showing his super duper important credentials told me "don't fall for social media hype" (i don't even use social media) and social media was in fact in the same tone "this is marketing again, they did it with unreal 3 and 4".
The credentialists and technocrats were wrong. Very few actually looking at what was presented and using logic, going around reading and reading other people with experience, the tech existing and all kind of data, we were right. But i knew this was a huge change when the first video appeared in April 2020.
Also, some people who post in this forums is working in an UE5 game, people with the highest lvl for making things happen and look great, and the game is looking amazing.
and your point is?
5.1 is out https://docs.unrealengine.com/5.1/en-US/unreal-engine-5.1-release-notes/
tbh wish there was a way to update easier, 'being' NOT having to download 20gb's again every update, but the updated files themselves, i get it, its compiled, still sucks to dl 20GB+, infinitely while i am here might as well add more of "my" dislikes, i get you want to "protect" your product, with all this signup, b.s. when in the past there was no sign up bs before and we all got along swimmingly with UDK before all these "new" things, is kind of dumb to me and just shines a sun on what the real reasoning's probably are which is just to stick everyone online imho-(these investors are everywhere). So anyone got an updater somewhere on the net where one doesn't have to download 20gb+s? Alt methods pm here or on station, everyone knows unreal its not like they will get away with anything...Also Thanks for sharing "epic", appreciate the learning utilities.
Could download the 500mb source from Github and compile it yourself.
FWIW : the launcher is only needed for the initial download and install. Once the engine is installed there is no need to login to anything ever (outisde of things requiring a download of course, like downloading Quixel assets or marketplace products). The launcher is not even needed to ... launch projects.
Also and just like any OS, there no need to always be on the latest anyway. On top of that
4.74.27 is still much more fluid to work with, as it has a much more sensical Content Browser behavior as opposed to the UX mess they introduced in 5.@pior
i've seen a few people say 4.7. That's just shorthand for 4.27, right? Or do you literally mean 4.7?
Hi there @Alex_J - sorry, that was a typo. I did mean 4.27.
I have been using ue5 in a lot of different projects since its release. There were a few issues regarding professional work. Can't say for 5.1 yet as I just started using it. The upside is that you always get a wow effect from the clients when they even see your grayboxing with lumen. The same projects look way more professional and much better quality even with bad materials, without spending time to get that 'detail lighting vs smoothing' bakes. BUT!
At least one of the things we definitely need is better storage technology to be able to keep up with the graphic quality rising.
5 is still a bit shaky and neither lumen or nanite are anywhere close to finished
but
4.27 is no better in terms of stability/bugs (worse probably) and doesn't feature any of the big improvements that actually help with development (extensions to subsystems, OFPA, world partition + more) .
Having used both in production I'd suggest it makes sense to go with 5 unless you plan to use static lighting - in which case you won't be able to use world partition or (afaik) OFPA til "after 5.2" which is not a date you'd want to schedule against