Ive been playing around with VR in UE for the past couple years also running on a mid-potato chasing the latest nanite/lumen implementation on UE Github while balancing the Quest 2 and now 3 processors. Slowly waiting to make my move to strike and make something. Fantastic video. The voice, the music, and inspiration is spot on. Keep making more videos please. Also doing the parallax shader technique with building windows like in the Matrix demos really help a story . Best.
excellent video Alistair!! thanks for continuing to evangelize VR with your awesome projects and I appreciate the visual shout-out to a few of my vids at the beginning. one thing to be aware of : when you bake lighting with nanite meshes, the baked lighting goes onto the proxy meshes (the lowest LOD ones). Won't notice for the most part with this city view, but getting close to, say, nanite furniture you'll see some funky shadows.
Thanks a lot :) Same to you! Your videos are always such a great guide for what's possible in UE5 - and its always months before anyone else! Looking forward to seeing what comes next :) And yes, that's one of the more annoying limitations on static lighting for Nanite at the moment. There is a workaround that is fine for mid-poly workflows though. If you adjust the relative error for the fallback mesh down to 0 (or a smaller figure for more complex organic meshes), then the fallback mesh that is used for the bake, is the same is the Nanite mesh. Its not ideal, as the entire scene is now a lot heavier for non-nanite platforms. Ideally, there would be an option to 'Use Nanite Mesh for Static Lighting' that gets included in future versions. I haven't requested a feature like that yet, but I also haven't checked if anyone else has. Other workarounds we're looking into include: a) Changing the fallback mesh at build time back to defaults. b) Having a more explicit workflow that builds simplified levels for non-nanite platforms (mostly mobile). Both come with drawbacks, but in the short term we're fine with the current setup, as we can still keep building our projects :)
Nice Video! We tried using Nanite and DX12 with our latest VR project, but in the end the framerate was just too slow, so we had to revert back to DX11 which is a real shame, but it is still a heck of a lot faster!
Ah, sorry to hear that! I've had the same experience with UE5 as well. Its been really frustrating and a little scary to see how slow UE5 can run with VR when you even try to 'look' at the new features. But 5.3, to my testing, has been incredibly fast across the board. It takes a few minutes to balance and optimize settings, but once there, it rips. I believe that a small team that loves VR inside Epic Games put a lot of work to get it this far, and I'm incredibly grateful to them :)
I've had similar results with the low frame rate and had to revert to DX11. It makes me wonder where PCVR will be in a few years, considering the current trend of increased resolution in VR headsets. It only makes it more challenging to create high-fidelity games that can run smoothly. Will software eventually solve the optimization problem, or will we need to wait for midrange GPUs to reach the performance level of the 4090, making it more accessible to most people?
@@kidzorro We completely agree. and think that the resolution of the Valve Index Headset is probably the sweetspot. Any resolutions higher than that is really not going to be helping VR developers atm!
This looks fascinating and, as someone with a 2080 Super/Quest3, looks like it's running very well. I can't wait to see what you're developing. Hopefully something story-focused.
As a fellow Deckard believer, I must say, that kit is the only hope of PCVR surviving in the next 5 years. Mobile XR2 Quest is the vast majority market (80%+ marketshare that is growing every second), and no respectable studio with a sense of business would dare to make a PCVR game. It's a dead platform, unfortunately. Meanwhile, Valve has added steamlink that works with Quest, which is the first big PCVR support software we have seen in 7 years other than steamvr. Here's hoping that more devs can see the light, a huge market of PCVR owners that have given up becasue Half Life Alyx was literally the first and last AAA PCVR game.
A worrying but accurate assessment, I think. Especially since my studio is making a PCVR game that absolutely could not run on Quest - even the 3. Its a small team, but a very cool project. It uses Nanite, millions of triangles and tries to push for next gen visuals at every turn, without requiring a 4090 to play. We're targeting PS5/Xbox Series X performance as our benchmark, with FSR to boost perf on lower end machines. These days, its best to program it to run on flatscreen as well as VR, to make up the numbers and access a wider player base, which is our approach to justify targeting high-end VR. PSVR 2 also helps, though its still quite niche. We think we can pull off a successful game without the Deckard, but I want to see the high-end industry thrive. That's where the excitement is, as well as the true frontier of immersive gaming! I'd rather do a series of high end games than pivot to mobile, as most of my peers have done. I guess we'll see :) Wish us luck lol! Worst case, we'll port it to the quest 5 in 2028
It's fantastic how UE is advancing. Of course, even an AAA VR experience doesn't need huge volumes of assets. I'd be more interested in focusing on the game, then, the asset package/balance since these can be fine tuned later. We really do need more AAA class games/experiences to really sell VR, and focussing too much on the tech can, and frequently does, detract from the end product. Still, and as always, yes, it would be nice to 'have it all' - and you've done well in demonstrating what might be possible.
Really been enjoying your exploration of this topic for over a year now. Curious what features and different practices would you employ if you knew your target audience was strictly 3080 - 4080. An updated video testing the absolute limits of the different lighting techs, dynamic effects and Nanite in 5.3 with VR would be great. Doesn't have to be a complex, crafted world but something quick that aims to reveal the most effective techniques and technologies for the highest end VR.
That's a great question! To be honest, I'd probably just do a few things: Design slightly larger levels Use a few more dynamic lights, that cast shadows Feel safer using a few more 4k textures Have optional settings for users with more powerful GPUs to enable screen space effects like Lumen Reflections (which work without needing the GI component as of 5.3), Ambient Occlusion, etc etc. Add a lot more props and tiny details to the level Use more atmospherics and particles If I knew that a 4090 upwards was my benchmark for a future game, I'd begin to completely re-scope the way I design VR games, and look into environments with a lot more complexity, day/night cycles, etc. You can totally pull that sort of stuff off this generation. A good example is Assassins Creed VR, which is coming out soon. But to do it at the level I'd like to see, I would want to crank the detail upwards a lot. The biggest problem is that more complex assets take longer to make (usually), so there's always a sweet spot. When I finally get a 4090 or similar, I'll do some tests going absolutely nuts. Probably won't be something like The Matrix Demo, as that runs rough even on flatscreen. But something large, dynamic and detailed will be the aim :)
@AlistairHume Thank you for the detailed response. Using Unreal to teach virtual production and VR for immersive education and a virtual location scouting so always looking to push the realism to make it believable. Thanks for the perspective. Get that 4090. Your work deserves it.
Indeed! The assets are optimized for UE4 specced games :) The main uplift here is much larger cities and more detail on screen. I remember back in the day, I couldn't really run this at all in UE4, but it runs nicely here. UE5.3 is finally really fast for VR!
You should look into using level instances and packed level actors so that you can get better performance with instanced static meshes. Noticed all of your "cities" are one big group of normal meshes. And changing some of those lights to spotlights 😄
I watch your video again as I continue build my first few levels. I was thinking it would be sweet to see hover crafts flying throughout the level. The night scene would be sweet. I wonder what kind of load it would put on the gpu. Just admire your detail.
I just started learning how to develop in vr with UE, and the education department is extremely poor, I simply can’t find the information I need it order to begin. More new features more confusion
These tools are awesome, but my 1660 already doesn't like running ue5 and playing in VR, let alone with Nanite or Lumen. Excited to see VR tools improve though.
Now if you got this working in standalone I’d be impressed. I haven’t seen much UE5 projects on standalone Quest with Nanite running. I am not sure if it’s possible?
I've see some people get Nanite working on the Quest 2 back in the day. But back then, it was barely working at all. Also, I'm not sure whether they'd even made it run properly or just claim they did, as it was early into UE5, and no one really knew the tech yet. We're getting closer to Nanite on Standalone with the Quest 3, but to my knowledge, it currently isn't supported on Android at all. They've put a lot of work getting it to run on Apple's M2 chips, so I would guess its a software architecture issue, as well as a performance issue. I wouldn't be surprised if we see some Nanite stuff on Mobile in the next few years. It is fast, but there's still a base cost to using the system, that's approximately 4ms on modern PC hardware. Until Mobile VR is ripping fast, I think a really optimized UE4 workflow, that doesn't use Nanite at all, will still get some amazing results. I'm personally curious to see what Valve is cooking up, as the rumours suggest that it brings the best of PC and Mobile VR together. I have a lot more interest in building for powerful devices, as you can pull a lot of amazing visuals off, but the Quest 3 is a really nice headset, and its power is really impressive. I guess my dream would be a solution that lets you scale Nanite assets between Mobile and PCVR, for example by reducing the mesh complexity for simpler platforms. No extra building or level swapping, just sliding a scale in the editor. If we can hit that, I will do a backflip :)
With some luck. If you use Virtual Textures, which is a way of storing higher resolution textures without blowing your GPU's memory, then its possible to have more 4k and 8k textures so everything looks really high detail. But you still have to model, light and texture things well from an artistic perspective. For VR, I still use them a little sparingly, because performance has to be top priority. But there's definitely an uplift, and things are starting to look sharper and crisper than I've ever seen before in VR :)
@@AlistairHume sounds good, I love VR no question about that , but i feel like its time that it stops to look like gaming from 2005 , whenever i tried to get friends into VR they got turned off by the graphics/model details and stuff.
Need to see actual framerate stats before I get too excited. I have yet to get nanite + VR to a shippable framerate. Motion smoothing will help hide the low frame rate, but you can't rely on that to ship. Still seems like a non-starter to me.
That map!! Jet packs and big team battles, you have a top selling game on steam guaranteed! No maps look anything close to as fun as what you created here! Any plans to release on sidequest so we can take a look?
Do you know how i can disable physical movement for my character while im playing? I mean, when i move in the physical world, the player in game also moves. How i can disable that? (You are my last hope)
I dont think its wise to use ue5 for vr, since the only vr game which uses this engine, that i tried on my 3080, ran at about 40 fps.. at lowest settings. And i dont know if this changed in quest3, but for quest2, you need to put the resolution slider all the way to the right for it to look native, because of the lense curvature or something, which is ofc much harder to run, but the 1,0 resolution looks like total crap.
I typically don't use the template, but instead build VR projects from my own codebase (or use the UE4VR codebase, which is excellent). Without seeing the error, its hard to know for sure what's happening. If its a common one that I'm not aware, of, I would hope that it will be resolved by the team at some point soon. If its a strange one, there might be someone else that's experienced something similar in the forum. As a wildcard, ChatGPT is actually surprisingly good at helping to diagnose errors. Could be worth pasting the key parts of the error message, and asking :)
So in the headset, the frames held at a solid 90FPS on the Quest 3 using PCVR. That was at full resolution. But its possible to reduce the pixel density a little, and use upscaling (TAAU) to still get a pretty crisp image if your frames drop a little. You can also do strong foveated rendering to get an extra few frames. I reckon I could boost it to 120 FPS with some tweaks to settings. But as a general point, it runs super nicely. The screencap didn't quite pick up the smoothness, but it felt really nice :)
@@joeysipos you can have fixed foveated rendering on almost any headset, it's just not following the eye movements but rather having full res only at the center, which is clearly not something I'd use personnally, but you can do it
@@joeysipos Eye-tracking enables DFR (dynamic fovated rendering); whereas, any headset is capable of doing FFR (fixed fovated rendering). DFR is incomparably better than FFR.
@@AlistairHume Is that 90fps after ASW? because it looks less fluid in the video (quite stuttery in fact), so it's hard to tell. I use an Index, and run at 120 when I record, to try and get a smooth 60fps video. I'd love to get a hands on demo, to see how well it runs locally. And do keep us informed about the "commercial project" as and when you can.
Nice to not have to worry about making imposters or LODs. I'm sure that will still be applicable in some circumstances, but I think it'll eventually be one of those things in the pipeline that goes the way of manual UV unwrapping or entering skinning weight values using a table instead of a brush. With each game engine advancement, more busy work is removed which liberates the artist more. My only concern is that the first couple of commercial games that use Nanite tend to make environments way too noisy and busy just because they can, however, it doesn't always equate to good design.
To memory, a few of the textures can't be shared. But if I have time in the future, I will look into re-authoring the scene and getting it up somewhere :) No promises in the short run, but it would be cool to see it out there
It's neither - it is deferred rendering (more flexible than Forward if you know what you're doing), but doesn't use Lumen. It uses baked lighting. Some Lumen experiments to come soon, using 5.4 and a very beefy GPU :)
Big time capture issue - it was running nicely at 90fps with the equivalent of Epic settings. I'll play around with a better capture solution next time:)
@vrgoskert One is in the works in my Studio :) It is open on my PC right now, but we're not ready to announce yet. Also a few others I've seen behind the scenes. They'll come, but because it is smaller studios taking them on, they'll take a little time
Really great sci-fi city scape looks fabulous and the interior design is very pleasing 🌟good to see you testing the nanite with an RTX 2080 to proove that developers with older GPU's can have a go at working with these types of environments 🌟I have a 3080 coupled with a Ryzen 3950x and was very much in need of the nanite development to even consider a personal project that I would like to develop. 🌟Subbed and look forward to your videos expanding on this particular creative project >> great work and channel here perfect 👌💯👀🎯😎🌠
I was expecting another "Look how amazing this looks using my $1500 4090 card" video. Nice to have a more realistic representation of what most people are running.
Every VR related video that you have uploaded so far is f**king amazing, I consider that is my bench mark of vr tech, but this time Meta Quest 3 + RTX 4080 + UE5.3, all these power house together I was expecting something more from you. Apart from Nanite capability there is nothing much i can see. You have throw some way too old City wid extremely limited details as u said. Real clarity, next gen graphic and stuff, i did not see any of those. None the less, thanks for your effort and surely waiting for something more in next video!
@JD-od6tw this is true, and will sadly probably remain true. Hopefully deferred will be a good option for certain types of games for this and future generations, but it requires a bit of a strong focus on AA to make sure the image finishes well. I usually suggest Forward Rendering unless devs are wanting to push into new visual techniques, and want to add a lot of model detail into their worlds. IMO both methods are viable. Forward is awesome for well scoped projects, and deferred (despite its drawback) gets us a little closer to new levels of detail. Fun times :)
Bold move using a 2080. As a dev I'd recommend targeting more powerful hardware so you're ahead of the curve, not behind it. That way when you ship you aren't already dated based on low cost development investment. Get a 4090 brother. Plus it's a writeoff. Cheers.
The progress in price/performance of GPUs is slowing down, so I wouldn't make games for 4090-level hardware right now, unless you plan to ship your game in 2030.
@@rhadiem It's not even just that, but new production processes provide less gain. If people don't have as much reason to upgrade, for whatever reason, they are going to buy a new GPU less often. You're not suddenly go see the big mass of $300 buyers, go for $1000 cards.
No offence I couldn't really see any changes. The issue for me I think is whatever you can see in the VR headset doesn't translate across to video or seeing it on a video screen. Alot of the data in VR is unique to the lenses and the headset, so all that depth and upgrade done is not translatable on screen unless you have the headset on. Nice video 📹 though I just wish I could of seen what you can see.
Idk i don't really care much for nanite at all. For me, post processing, shading, and rendering methods, such as SSAO, SSR, GI, upscaling, etc, are more important to me. God, especially SSAO - I'm not sure i know of a single VR game rendering using multiview, that has some form of SSAO. I see no reason to use UE5 over just man-handling something more lightweight like unity or godot, and writing my own rendering/shading methods. Especially with the file sizes... I'm not midas.
every single ue5 game released so far has been a shit show of performance. Problem is devs think nanite is a lazy way to drop massive unoptimized models into levels, it just does not scale well across different hardware. The dream is not there yet.
@@Shadow__133mostly due to 95% of game devs using megascans (real life 3D scans just look ugly in games/VR). from scratch models with a large amount of stylization looks way better. more vibrant and better saturated coloring etc
@@Hadkek The cutscenes looked good, the game ok but as I said, unimpressive. Just normal. And the performance was not the best, with some huge lags on certain areas (sniper scene on the mine/dump/construction map).
All we see are tech demos for the last 10 or so years. So far when it comes to games the best graphics are probably The Last of Us 2. Spider man 2. So if they don’t translate into games what’s the point.
Always a foggy, smoggy environments, never a clear day. I wonder has any game developer/cgi artist ever seen or visited a place without smog/fog? Or maybe it's the grimy glasses they wear that makes them see everything as foggy?
Looking forward to seeing more VR support. Using praydog’s mod in everything UE5 till now it really does take a small glimpse at the future!
Oh wow that looks like a cool mod! I might have to pick up a 4090 and spend a few hours checking out awesome looking worlds :D
Agreed it's pretty sharp 👌
dude pls tell me how to use that mod injector 😭
@@abunknown234t7 will be hopefully released this year for free. Pretty easy, just pick a running UE 4/5 game and click inject :)
@@AlistairHume it's incredible!
Looks great would love to see you do more VR combined with Nanite videos.
Thanks very much :) Nanite's finally here, so there will be more fun to be had with it!
As an artist and a software engineer, I am enjoying deeply your thoughts and effort and passion.
Ive been playing around with VR in UE for the past couple years also running on a mid-potato chasing the latest nanite/lumen implementation on UE Github while balancing the Quest 2 and now 3 processors. Slowly waiting to make my move to strike and make something.
Fantastic video. The voice, the music, and inspiration is spot on. Keep making more videos please.
Also doing the parallax shader technique with building windows like in the Matrix demos really help a story . Best.
excellent video Alistair!! thanks for continuing to evangelize VR with your awesome projects and I appreciate the visual shout-out to a few of my vids at the beginning.
one thing to be aware of : when you bake lighting with nanite meshes, the baked lighting goes onto the proxy meshes (the lowest LOD ones). Won't notice for the most part with this city view, but getting close to, say, nanite furniture you'll see some funky shadows.
Thanks a lot :) Same to you! Your videos are always such a great guide for what's possible in UE5 - and its always months before anyone else! Looking forward to seeing what comes next :)
And yes, that's one of the more annoying limitations on static lighting for Nanite at the moment. There is a workaround that is fine for mid-poly workflows though. If you adjust the relative error for the fallback mesh down to 0 (or a smaller figure for more complex organic meshes), then the fallback mesh that is used for the bake, is the same is the Nanite mesh. Its not ideal, as the entire scene is now a lot heavier for non-nanite platforms. Ideally, there would be an option to 'Use Nanite Mesh for Static Lighting' that gets included in future versions. I haven't requested a feature like that yet, but I also haven't checked if anyone else has. Other workarounds we're looking into include:
a) Changing the fallback mesh at build time back to defaults.
b) Having a more explicit workflow that builds simplified levels for non-nanite platforms (mostly mobile).
Both come with drawbacks, but in the short term we're fine with the current setup, as we can still keep building our projects :)
@@AlistairHume great point!!
Nice Video! We tried using Nanite and DX12 with our latest VR project, but in the end the framerate was just too slow, so we had to revert back to DX11 which is a real shame, but it is still a heck of a lot faster!
Ah, sorry to hear that! I've had the same experience with UE5 as well. Its been really frustrating and a little scary to see how slow UE5 can run with VR when you even try to 'look' at the new features. But 5.3, to my testing, has been incredibly fast across the board. It takes a few minutes to balance and optimize settings, but once there, it rips. I believe that a small team that loves VR inside Epic Games put a lot of work to get it this far, and I'm incredibly grateful to them :)
Very cool, maybe change from Unity to UE5 now
I've had similar results with the low frame rate and had to revert to DX11. It makes me wonder where PCVR will be in a few years, considering the current trend of increased resolution in VR headsets. It only makes it more challenging to create high-fidelity games that can run smoothly. Will software eventually solve the optimization problem, or will we need to wait for midrange GPUs to reach the performance level of the 4090, making it more accessible to most people?
@@kidzorro We completely agree. and think that the resolution of the Valve Index Headset is probably the sweetspot. Any resolutions higher than that is really not going to be helping VR developers atm!
Thanks for this Alistair will be having a go at this to test our VR project
Quality content. Future is promising. We need this to get polished and micorOLEDs in our headsets.
If every VR Devs could be like you....
Bro! Love your videos! Long time. Hope all is awesome 💜🤓
This looks fascinating and, as someone with a 2080 Super/Quest3, looks like it's running very well. I can't wait to see what you're developing. Hopefully something story-focused.
Story, environmental exploration and some scifi mystery themes across the game :)
Another great video! Excited to see more
As a fellow Deckard believer, I must say, that kit is the only hope of PCVR surviving in the next 5 years. Mobile XR2 Quest is the vast majority market (80%+ marketshare that is growing every second), and no respectable studio with a sense of business would dare to make a PCVR game. It's a dead platform, unfortunately.
Meanwhile, Valve has added steamlink that works with Quest, which is the first big PCVR support software we have seen in 7 years other than steamvr.
Here's hoping that more devs can see the light, a huge market of PCVR owners that have given up becasue Half Life Alyx was literally the first and last AAA PCVR game.
A worrying but accurate assessment, I think. Especially since my studio is making a PCVR game that absolutely could not run on Quest - even the 3. Its a small team, but a very cool project. It uses Nanite, millions of triangles and tries to push for next gen visuals at every turn, without requiring a 4090 to play. We're targeting PS5/Xbox Series X performance as our benchmark, with FSR to boost perf on lower end machines.
These days, its best to program it to run on flatscreen as well as VR, to make up the numbers and access a wider player base, which is our approach to justify targeting high-end VR. PSVR 2 also helps, though its still quite niche.
We think we can pull off a successful game without the Deckard, but I want to see the high-end industry thrive. That's where the excitement is, as well as the true frontier of immersive gaming! I'd rather do a series of high end games than pivot to mobile, as most of my peers have done.
I guess we'll see :) Wish us luck lol! Worst case, we'll port it to the quest 5 in 2028
It's fantastic how UE is advancing. Of course, even an AAA VR experience doesn't need huge volumes of assets. I'd be more interested in focusing on the game, then, the asset package/balance since these can be fine tuned later. We really do need more AAA class games/experiences to really sell VR, and focussing too much on the tech can, and frequently does, detract from the end product.
Still, and as always, yes, it would be nice to 'have it all' - and you've done well in demonstrating what might be possible.
good stuff bro!
Thanks a lot :) Great to see some the awesome stuff you've built!
Nice one, you are THE bob ross of ue 🎉
Amen haha
5.4 currently seems like a step back for VR. I get weird tearing in the Quest 3. It goes away in forward rendering.
Really been enjoying your exploration of this topic for over a year now. Curious what features and different practices would you employ if you knew your target audience was strictly 3080 - 4080. An updated video testing the absolute limits of the different lighting techs, dynamic effects and Nanite in 5.3 with VR would be great. Doesn't have to be a complex, crafted world but something quick that aims to reveal the most effective techniques and technologies for the highest end VR.
That's a great question! To be honest, I'd probably just do a few things:
Design slightly larger levels
Use a few more dynamic lights, that cast shadows
Feel safer using a few more 4k textures
Have optional settings for users with more powerful GPUs to enable screen space effects like Lumen Reflections (which work without needing the GI component as of 5.3), Ambient Occlusion, etc etc.
Add a lot more props and tiny details to the level
Use more atmospherics and particles
If I knew that a 4090 upwards was my benchmark for a future game, I'd begin to completely re-scope the way I design VR games, and look into environments with a lot more complexity, day/night cycles, etc. You can totally pull that sort of stuff off this generation. A good example is Assassins Creed VR, which is coming out soon. But to do it at the level I'd like to see, I would want to crank the detail upwards a lot. The biggest problem is that more complex assets take longer to make (usually), so there's always a sweet spot.
When I finally get a 4090 or similar, I'll do some tests going absolutely nuts. Probably won't be something like The Matrix Demo, as that runs rough even on flatscreen. But something large, dynamic and detailed will be the aim :)
@AlistairHume Thank you for the detailed response. Using Unreal to teach virtual production and VR for immersive education and a virtual location scouting so always looking to push the realism to make it believable. Thanks for the perspective. Get that 4090. Your work deserves it.
Very cool! but this looks rather low-poly already, why is nanite needed here?
Indeed! The assets are optimized for UE4 specced games :) The main uplift here is much larger cities and more detail on screen. I remember back in the day, I couldn't really run this at all in UE4, but it runs nicely here. UE5.3 is finally really fast for VR!
You should look into using level instances and packed level actors so that you can get better performance with instanced static meshes. Noticed all of your "cities" are one big group of normal meshes.
And changing some of those lights to spotlights 😄
As a vr enthusiast, I'm excited to see the potential for proper nanite usage in vr games.
Same!
Honestly all I could think of while looking at the result was how passable a simple prerendered skybox would have been
I watch your video again as I continue build my first few levels. I was thinking it would be sweet to see hover crafts flying throughout the level. The night scene would be sweet. I wonder what kind of load it would put on the gpu. Just admire your detail.
what happened to the game you were working on?
I really wanna makea VR Game but I'm still stuck with my NON VR game and having a full time job is a huge slow down for me :\
I just started learning how to develop in vr with UE, and the education department is extremely poor, I simply can’t find the information I need it order to begin. More new features more confusion
Wow. I would looove to be able to walk through a city like that at night in VR. Jeez i better get a move on and learn UE5.
These tools are awesome, but my 1660 already doesn't like running ue5 and playing in VR, let alone with Nanite or Lumen. Excited to see VR tools improve though.
would be interesting to see fpsvr in scenes like this. to give an idea how performacne light or heavy things are.
Hiya. Fun to meet you here. It looks awsome
Now if you got this working in standalone I’d be impressed. I haven’t seen much UE5 projects on standalone Quest with Nanite running. I am not sure if it’s possible?
I've see some people get Nanite working on the Quest 2 back in the day. But back then, it was barely working at all. Also, I'm not sure whether they'd even made it run properly or just claim they did, as it was early into UE5, and no one really knew the tech yet.
We're getting closer to Nanite on Standalone with the Quest 3, but to my knowledge, it currently isn't supported on Android at all. They've put a lot of work getting it to run on Apple's M2 chips, so I would guess its a software architecture issue, as well as a performance issue.
I wouldn't be surprised if we see some Nanite stuff on Mobile in the next few years. It is fast, but there's still a base cost to using the system, that's approximately 4ms on modern PC hardware. Until Mobile VR is ripping fast, I think a really optimized UE4 workflow, that doesn't use Nanite at all, will still get some amazing results.
I'm personally curious to see what Valve is cooking up, as the rumours suggest that it brings the best of PC and Mobile VR together. I have a lot more interest in building for powerful devices, as you can pull a lot of amazing visuals off, but the Quest 3 is a really nice headset, and its power is really impressive. I guess my dream would be a solution that lets you scale Nanite assets between Mobile and PCVR, for example by reducing the mesh complexity for simpler platforms. No extra building or level swapping, just sliding a scale in the editor. If we can hit that, I will do a backflip :)
3:22 what is the title of this music ?
So , do you think it might be possible to see games popping up with good textures soon?
With some luck. If you use Virtual Textures, which is a way of storing higher resolution textures without blowing your GPU's memory, then its possible to have more 4k and 8k textures so everything looks really high detail. But you still have to model, light and texture things well from an artistic perspective. For VR, I still use them a little sparingly, because performance has to be top priority. But there's definitely an uplift, and things are starting to look sharper and crisper than I've ever seen before in VR :)
@@AlistairHume sounds good, I love VR no question about that , but i feel like its time that it stops to look like gaming from 2005 , whenever i tried to get friends into VR they got turned off by the graphics/model details and stuff.
Can we pay some dollars just to get the exe file of this level? I’d love to see the results first-hand
High fidelity VR MMOs just feel like being a lot closer now
Yep high resolution screens and 4090 tech that will trickle down to the masses is getting really good.
It's all in our hands now
Can't wait to see them bringing Nanite to mobile platforms, I'm working on a Quest 2/3 game atm. It's on their roadmap but no idea when to expect it.
That's a 2080? Oh man. That's awesome!
Need to see actual framerate stats before I get too excited. I have yet to get nanite + VR to a shippable framerate. Motion smoothing will help hide the low frame rate, but you can't rely on that to ship. Still seems like a non-starter to me.
Will you be providing a sample of this VR project on marketplace? How did you generate the city?
ima check it out is it on steam
Can’t wait to see what naught dog will do with something like this
i`m imagining what vtol vr could look like with this level of detail...
Sweet, love the work you have been doing. Was this city used in a VR program called "Air Car".
Woo,Awesome!👍💪👏
While im watching this, i made a game in my mind
That map!! Jet packs and big team battles, you have a top selling game on steam guaranteed! No maps look anything close to as fun as what you created here! Any plans to release on sidequest so we can take a look?
Do you know how i can disable physical movement for my character while im playing?
I mean, when i move in the physical world, the player in game also moves.
How i can disable that?
(You are my last hope)
I dont think its wise to use ue5 for vr, since the only vr game which uses this engine, that i tried on my 3080, ran at about 40 fps.. at lowest settings.
And i dont know if this changed in quest3, but for quest2, you need to put the resolution slider all the way to the right for it to look native, because of the lense curvature or something, which is ofc much harder to run, but the 1,0 resolution looks like total crap.
Do you have any solutions how to Fix the Open XR error? I try to package the VR Template in 5.3 and it cant even package! :( Thanks for the video.
I typically don't use the template, but instead build VR projects from my own codebase (or use the UE4VR codebase, which is excellent). Without seeing the error, its hard to know for sure what's happening. If its a common one that I'm not aware, of, I would hope that it will be resolved by the team at some point soon. If its a strange one, there might be someone else that's experienced something similar in the forum. As a wildcard, ChatGPT is actually surprisingly good at helping to diagnose errors. Could be worth pasting the key parts of the error message, and asking :)
Visual Stuido needed to be updated @@AlistairHume
Very good. Is lumen now working in VR too?
time to buy out Praydog UE injector? maybe?
What is the reason that this requires PCVR?
Foveated rendering in effect?
Hi
Are you plugged in or streaming over wifi?
Time for Bathesda to launch Skyrim for the 20th time but now in UE5 😂
I think I would actually rebuy it at that point..but everything in the game would have to be re done.
This makes me want to replay Aircar a bunch.
I did play it again recently, and it is still brilliant :)
One day you might be able to recreate AirCar (2017) which was my favorite experience in VR back in the day. By then Unreal 6 might be out. 😅
Nice vid, but curious, what was the framerate ? For VR it is so important and I am wondering if this runs high enough for a good time ?
So in the headset, the frames held at a solid 90FPS on the Quest 3 using PCVR. That was at full resolution. But its possible to reduce the pixel density a little, and use upscaling (TAAU) to still get a pretty crisp image if your frames drop a little. You can also do strong foveated rendering to get an extra few frames. I reckon I could boost it to 120 FPS with some tweaks to settings. But as a general point, it runs super nicely. The screencap didn't quite pick up the smoothness, but it felt really nice :)
@@AlistairHumeQuest 3 doesn’t support foveated rendering. Only headsets that have eye tracking.
@@joeysipos you can have fixed foveated rendering on almost any headset, it's just not following the eye movements but rather having full res only at the center, which is clearly not something I'd use personnally, but you can do it
@@joeysipos Eye-tracking enables DFR (dynamic fovated rendering); whereas, any headset is capable of doing FFR (fixed fovated rendering). DFR is incomparably better than FFR.
@@AlistairHume Is that 90fps after ASW? because it looks less fluid in the video (quite stuttery in fact), so it's hard to tell.
I use an Index, and run at 120 when I record, to try and get a smooth 60fps video.
I'd love to get a hands on demo, to see how well it runs locally.
And do keep us informed about the "commercial project" as and when you can.
Nice job
music?
Wooow! Yeah!!!🥰😆😆😆😂😝😎😄😄😁. Im excited
THIS IS AWESOME! Can you release this as a demo? nothing polished...
If I find some time, I'll definitely pull something together :)
@@AlistairHume that would be awesome! Many have pcvr but not the skills to setup a quick tech demo of nanite in vr in UE5.3
Nice to not have to worry about making imposters or LODs. I'm sure that will still be applicable in some circumstances, but I think it'll eventually be one of those things in the pipeline that goes the way of manual UV unwrapping or entering skinning weight values using a table instead of a brush. With each game engine advancement, more busy work is removed which liberates the artist more. My only concern is that the first couple of commercial games that use Nanite tend to make environments way too noisy and busy just because they can, however, it doesn't always equate to good design.
I don't understand why people who make these types of videos don't compile these demos and upload them so we can all try them.
Will this become available for development of Android standalone games as well?
Not any time soon.
Can rtx 3050 run unreal engine 5.3??
Great Video. Thank You. 👍 . Is it possible to get this Scene/Datafile from the City from you or is it CopyRight/Commercial ? 🙂
To memory, a few of the textures can't be shared. But if I have time in the future, I will look into re-authoring the scene and getting it up somewhere :) No promises in the short run, but it would be cool to see it out there
is that lumen or forward rendering?
It's neither - it is deferred rendering (more flexible than Forward if you know what you're doing), but doesn't use Lumen. It uses baked lighting. Some Lumen experiments to come soon, using 5.4 and a very beefy GPU :)
Quest 3 is a really good headset. I use it almost every day.
This looked great, the performance looked pretty bad though, was it just a capture issue?
Big time capture issue - it was running nicely at 90fps with the equivalent of Epic settings. I'll play around with a better capture solution next time:)
@@AlistairHume Ah i see, thank you for letting me know.
10 months passed till this video, and no next-gen VR game on the horison :( :(
@vrgoskert One is in the works in my Studio :) It is open on my PC right now, but we're not ready to announce yet. Also a few others I've seen behind the scenes. They'll come, but because it is smaller studios taking them on, they'll take a little time
@@AlistairHume Yesss! Thanks, than: "We are here, and we are waiting." by Optimus Prime 🤩
Watching your video make me think about one thing. If you gonna record some voices, your voice would be suitable for announcer/narrator.
Really great sci-fi city scape looks fabulous and the interior design is very pleasing 🌟good to see you testing the nanite with an RTX 2080 to proove that developers with older GPU's can have a go at working with these types of environments 🌟I have a 3080 coupled with a Ryzen 3950x and was very much in need of the nanite development to even consider a personal project that I would like to develop. 🌟Subbed and look forward to your videos expanding on this particular creative project >> great work and channel here perfect 👌💯👀🎯😎🌠
👍🏾👍🏾👍🏾 subbed
Vr nanite starfield. Nasapunk in vr. With orbiting planets, moons, global illumination from stars.
I was expecting another "Look how amazing this looks using my $1500 4090 card" video. Nice to have a more realistic representation of what most people are running.
Nice video! What systems specs are you running to be able to run this at 90 fps at 4k?
Every VR related video that you have uploaded so far is f**king amazing, I consider that is my bench mark of vr tech, but this time Meta Quest 3 + RTX 4080 + UE5.3, all these power house together I was expecting something more from you. Apart from Nanite capability there is nothing much i can see. You have throw some way too old City wid extremely limited details as u said. Real clarity, next gen graphic and stuff, i did not see any of those. None the less, thanks for your effort and surely waiting for something more in next video!
Hello mate >> yeah he was running it on a 2080 in this video >> just to see how well that would perform for those with older builds 👍
VR need great games
It's important to note for VR devs that Nanite does NOT currently support forward shading
@JD-od6tw this is true, and will sadly probably remain true. Hopefully deferred will be a good option for certain types of games for this and future generations, but it requires a bit of a strong focus on AA to make sure the image finishes well. I usually suggest Forward Rendering unless devs are wanting to push into new visual techniques, and want to add a lot of model detail into their worlds. IMO both methods are viable. Forward is awesome for well scoped projects, and deferred (despite its drawback) gets us a little closer to new levels of detail. Fun times :)
It looks ok but it still needs FAR more detail to be realistic in VR.
Bold move using a 2080. As a dev I'd recommend targeting more powerful hardware so you're ahead of the curve, not behind it. That way when you ship you aren't already dated based on low cost development investment. Get a 4090 brother. Plus it's a writeoff. Cheers.
The progress in price/performance of GPUs is slowing down, so I wouldn't make games for 4090-level hardware right now, unless you plan to ship your game in 2030.
@@aapje Prices for everything have gone up... it's called inflation. 😉
@@rhadiem It's not even just that, but new production processes provide less gain. If people don't have as much reason to upgrade, for whatever reason, they are going to buy a new GPU less often. You're not suddenly go see the big mass of $300 buyers, go for $1000 cards.
@@aapje 👌 :)
That lag is so bad it breaks my heart, I guess we're just not there yet for VR 😥
Amazing! Can you please work for Rockstar Games and bring GTA to psvr2? :)
I have some friends working at Rockstar - I will scream at them to add VR next time I catch up :)
No offence I couldn't really see any changes. The issue for me I think is whatever you can see in the VR headset doesn't translate across to video or seeing it on a video screen. Alot of the data in VR is unique to the lenses and the headset, so all that depth and upgrade done is not translatable on screen unless you have the headset on. Nice video 📹 though I just wish I could of seen what you can see.
Foveated rendering intensifies.
GPU screaming intensifies...
Very dece video Ali
Idk i don't really care much for nanite at all. For me, post processing, shading, and rendering methods, such as SSAO, SSR, GI, upscaling, etc, are more important to me. God, especially SSAO - I'm not sure i know of a single VR game rendering using multiview, that has some form of SSAO.
I see no reason to use UE5 over just man-handling something more lightweight like unity or godot, and writing my own rendering/shading methods.
Especially with the file sizes... I'm not midas.
Argh! Why you don't use darkmode on youtube! I watched this video at night and now i am blind..
every single ue5 game released so far has been a shit show of performance. Problem is devs think nanite is a lazy way to drop massive unoptimized models into levels, it just does not scale well across different hardware. The dream is not there yet.
Not to mention unimpressive image quality.
Seems overhyped af
What are some games that use it? I play a ton of VR games and wondered which ones use this. Seems like most games are poorly optimized
@@Shadow__133mostly due to 95% of game devs using megascans (real life 3D scans just look ugly in games/VR).
from scratch models with a large amount of stylization looks way better.
more vibrant and better saturated coloring etc
Robocop looked and played well
@@Hadkek The cutscenes looked good, the game ok but as I said, unimpressive. Just normal.
And the performance was not the best, with some huge lags on certain areas (sniper scene on the mine/dump/construction map).
All we see are tech demos for the last 10 or so years. So far when it comes to games the best graphics are probably The Last of Us 2. Spider man 2. So if they don’t translate into games what’s the point.
PS5s don't have the same processing power/GPU bandwidth as a 2080. No where near.
Gimmie dat city
make a discord server, i'd like to join
It's time to upgrade UnReal engine to Real engine
Need a play tester?
Why are you whispering bro?
A VR developer waiting for Nanite to work in VR without knowing that "Nanite" is actually 14 years old and originally called Unlimited Graphics?
but star citizen looks miles better and that's been out way longer.
The performance looks really bad, i hear VR is a pain in Unreal
You lost me at 2090
Always a foggy, smoggy environments, never a clear day. I wonder has any game developer/cgi artist ever seen or visited a place without smog/fog? Or maybe it's the grimy glasses they wear that makes them see everything as foggy?