This is not a Rant about Nanite. But showing the current state of Nanite in 5.1 If you say, Nanite can't do tessellation. It's true. Not in 5.1. But the team is exploring ideas. Check this: forums.unrealengine.com/t/ask-unreal-anything-rendering-june-15-2022-at-1pm-edt/577074/69 It's a super nice discussion with the UE team behind Nanite, Lumen.
The Nanite cluster view is how you can check to see the accuracy of deformation, as of 5.1 WPO is supported on the actual mesh, but not the clusters that determine the detail. So basically as noted in the 5.1 documentation under Nanite Virtualized Geometry, it works great with a mesh with a lot of holes, since the clusters themselves are not packed together in the DAG parents, whenever there is a gap clusters are forever separated at whatever distance. As long as you use the new keep surface area setting. To sum it up, put a wholes in your mesh to get more accurate clusters and more accurate WPO. Again I am not just saying this because of my research and deep understanding of Nanite, but it is also in the documentation . Don’t just take my word for it it’s there. And you can see more about the Nanite clusters DAG at the Nanite advancements in real time rendering paper. Clusters do not move with the WPO and the way clusters are grouped is fixed since the conversion of regular to Nanite mesh.
wait, Did I just see you faked a deprecated Displacement with WPO in UE5 (in the first 2 meshes) ?!!!!!!!!!!!!!!!! how the hell did you do that ? please make a tutorial on it.
Well, WPO is never deprecated in UE5. It was there for non-static meshes. But with 5.1, you can do that for Nanite meshes as well. Just google for WPO in Unreal (UE4 stuff) & it will work just fine for UE5 & 5.1 (for Nanite) as well.
I tried converting grass with WPO to nanite. Placing it inside the scene works fine, but it doesn't work fine in grasstype and disappears. And the swing direction is wrong in foliage mode (from left to right to up and down)
Pivot point texture is used in my grass shader. Grass with Character Interaction……Putting the grass directly inside the scene everything works fine. So I am considering whether I need to wait for the official improvement
The problem right now is that if all triangles are less than 1 pixel big, even if they fill the space, they disappear. Maybe they can group them in the next update
It seems as though you got an incorrect understanding of what Nanite does and extrapolated those missimpressions into incorrect comparisons. Nanite can never generate higher fidelity tesselation, it was never a tesselator. It can only go in the opposite direction of simplification and lowering fidelity. So Nanite or no Nanite, if you displace a mesh with insufficient tesselation at it’s highest fidelity, it won’t displace well. If you were expecting Nanite to adaptively tesselate your badly tesselated mesh so that it can be displaced well, then the problem is in your expectation. The real interesting test would be to displace a pre-tesselated mesh (before it is provided to Nanite) with Nanite enabled, but you have not shown that here.
Who talked about tessellation here. Displacement and tessellation are two things. I just talked about displacement. There was no any tessellated mesh here and it was removed in UE5. That was never the intention of this video.
@@Tertion Well. There is no correct or wrong way here. I remember the lead developer talked about a Voxel based method for displacing a Nanite mesh when we asked about WPO support initially. So, they are experimenting with ideas. Since it’s a whole new tech, there is no correct or wrong way to use it.
@@GDi4K The way you presented in the video gave the impression that you were expecting displacement to work well on un-tesselated meshes, and seems like other viewers also got the same impression (based on the likes of this comment). If that's not what you were expecting, then you can take the feedback to be about how this was presented, as opposed to about your understanding. What I said was that the more useful comparison that you didn't do would have been to use the same highly (uniformly)pre-tesselated meshes for both the Nanite and non-Nanite meshes, and then apply the same displacement on both to see how/if the displacement is maintained even as Nanite chooses lower LOD clusters as the camera moves away from the object. Instead, you've decided to create an inappropriate scenario for the Nanite meshes where they start off with insufficiently tesselated geometry to work with for being displaced, and then incorrectely concluded from the bad result you got that it's Nanite's fault for why the displacement doesn't work well for them. If I got that wrong and that's not what you actually did, I stand corrected, but it sure seems like that's what you did from how you presented it, so maybe you could have clarified those points better in your presentation
![Nanite: Everything You Should Know [Unreal Engine 5]](http://i.ytimg.com/vi/P65cADzsP8Q/mqdefault.jpg)
This is not a Rant about Nanite. But showing the current state of Nanite in 5.1
If you say, Nanite can't do tessellation. It's true. Not in 5.1.
But the team is exploring ideas.
Check this: forums.unrealengine.com/t/ask-unreal-anything-rendering-june-15-2022-at-1pm-edt/577074/69
It's a super nice discussion with the UE team behind Nanite, Lumen.
Oh man, the vibrant environments this will create
The Nanite cluster view is how you can check to see the accuracy of deformation, as of 5.1 WPO is supported on the actual mesh, but not the clusters that determine the detail. So basically as noted in the 5.1 documentation under Nanite Virtualized Geometry, it works great with a mesh with a lot of holes, since the clusters themselves are not packed together in the DAG parents, whenever there is a gap clusters are forever separated at whatever distance. As long as you use the new keep surface area setting.
To sum it up, put a wholes in your mesh to get more accurate clusters and more accurate WPO. Again I am not just saying this because of my research and deep understanding of Nanite, but it is also in the documentation . Don’t just take my word for it it’s there. And you can see more about the Nanite clusters DAG at the Nanite advancements in real time rendering paper.
Clusters do not move with the WPO and the way clusters are grouped is fixed since the conversion of regular to Nanite mesh.
pls more 5.1 overviews
Thank you.
I checked - everything is clean
This is really exciting. I'd like to see someone use the nanite and push the foliage to see the performance difference and if it's worthwhile.
There are some nice examples you Twitter with trees. I saw another one with animated octopus arms.
FINALLY!!! NANITE GRASS BABY
Does nanite work with pivot paainter 2?
wait, Did I just see you faked a deprecated Displacement with WPO in UE5 (in the first 2 meshes) ?!!!!!!!!!!!!!!!! how the hell did you do that ? please make a tutorial on it.
Well, WPO is never deprecated in UE5. It was there for non-static meshes. But with 5.1, you can do that for Nanite meshes as well. Just google for WPO in Unreal (UE4 stuff) & it will work just fine for UE5 & 5.1 (for Nanite) as well.
I tried converting grass with WPO to nanite. Placing it inside the scene works fine, but it doesn't work fine in grasstype and disappears.
And the swing direction is wrong in foliage mode (from left to right to up and down)
Pivot point texture is used in my grass shader. Grass with Character Interaction……Putting the grass directly inside the scene everything works fine. So I am considering whether I need to wait for the official improvement
Nanite displacement how it works ?
The problem right now is that if all triangles are less than 1 pixel big, even if they fill the space, they disappear. Maybe they can group them in the next update
Go to mesh editor of the tree and for each material with alpha channel select preserve area in nanite settings
works terrible right now, tradiional method foliage huge forest 90fps when I use nanite
It seems as though you got an incorrect understanding of what Nanite does and extrapolated those missimpressions into incorrect comparisons. Nanite can never generate higher fidelity tesselation, it was never a tesselator. It can only go in the opposite direction of simplification and lowering fidelity. So Nanite or no Nanite, if you displace a mesh with insufficient tesselation at it’s highest fidelity, it won’t displace well. If you were expecting Nanite to adaptively tesselate your badly tesselated mesh so that it can be displaced well, then the problem is in your expectation. The real interesting test would be to displace a pre-tesselated mesh (before it is provided to Nanite) with Nanite enabled, but you have not shown that here.
Who talked about tessellation here. Displacement and tessellation are two things. I just talked about displacement. There was no any tessellated mesh here and it was removed in UE5. That was never the intention of this video.
@@GDi4K but the stated "problems" you show in the video are the result of what @Arnon Marcus tried to say.
Yeah, I had the same taugh.... this is not how you are supposed to use Nanite....
@@Tertion Well. There is no correct or wrong way here. I remember the lead developer talked about a Voxel based method for displacing a Nanite mesh when we asked about WPO support initially. So, they are experimenting with ideas. Since it’s a whole new tech, there is no correct or wrong way to use it.
@@GDi4K The way you presented in the video gave the impression that you were expecting displacement to work well on un-tesselated meshes, and seems like other viewers also got the same impression (based on the likes of this comment). If that's not what you were expecting, then you can take the feedback to be about how this was presented, as opposed to about your understanding. What I said was that the more useful comparison that you didn't do would have been to use the same highly (uniformly)pre-tesselated meshes for both the Nanite and non-Nanite meshes, and then apply the same displacement on both to see how/if the displacement is maintained even as Nanite chooses lower LOD clusters as the camera moves away from the object. Instead, you've decided to create an inappropriate scenario for the Nanite meshes where they start off with insufficiently tesselated geometry to work with for being displaced, and then incorrectely concluded from the bad result you got that it's Nanite's fault for why the displacement doesn't work well for them. If I got that wrong and that's not what you actually did, I stand corrected, but it sure seems like that's what you did from how you presented it, so maybe you could have clarified those points better in your presentation
WPO motion in UE always going: *ɔ c ɔ c ɔ c ɔ c ɔ c ɔ c*
It would be nice if "SimpleGrassWind" in BP wasn't quite as "simple" anymore :D
👍