Same here. Now I’m having to import really high poly character meshes to make up for this, and since NANITE doesn’t work on skeletal meshes…
They seem to have gone with the “oh lol, don’t you guys have phones?” Approach.
Which is alarming, they should know which toolsets people use. Tessellation was not an ideal solution but widely used and very convenient on PC.
For real Epic, there needs to be a solution for texture-driven displacement in UE5. Nanite ATM is not it (there are way too many limitations and it doesn’t fill the same niche at all) and neither is VHFM (buggy, annoying to set up, requires virtual texturing).
Of all the things they have responded to…there have been literally hundreds of tessellation-related complaints and not a word on whether they plan to address this. One guy went to far as to start a petition.
Can we get some feedback on this at all, it is a real problem for landscapes and a lot of other things as well
ATM this almost leaves UE4 as the more desirable platform if this is a part of your workflow…and that is a shame, we shouldn’t be going backwards here
4 Likes
That’s absolutely unreasonable.
That implies pay at least usd 300 bucks for an indie license to do something that was native inside Unreal just a version prior.
I’ve even seen you suggest Zbrush on this thread, that makes no sense. Learn a new software to do something that was done natively in engine?
We need tessellation for a bunch of stylized animated materials. There’s no other solution at the moment available. It would be awesome if someone could at least point to a direction that doesn’t involve “don’t use ue5”
Well you just said the answer yourself, which is correct. If you still need tessellation, then don’t use UE5. It’s been well known for around two years now that tessellation was going to be depreciated.
The simple solution, for now, is to re-export your meshes with higher resolution or mess with your LODs. Use LOD0 for longer, LOD1 for longer, etc. If it’s going to be a nanite static mesh, export your million poly mesh that you would have used to make your normal or displacement maps. If it’s a skinned mesh, subdivide it couple times and bake in whatever displacement map you were using for WPO.
Other than that, you can use VHFMs. I do think it’s kind of lame that they didn’t fully flesh out the workflow for using VHFMs before removing tessellation though. In an ideal world, they’d have some kind of automatic in-editor-converter that could handle the legacy assets and automatically convert them to the new systems, but we don’t have that yet.
Again, you can still use UE4. It’s going to be a few years before UE5 has all the kinks worked out. UE4 didn’t really truly shine until around halfway through it’s life.
4 Likes
Ironing out major kinks is what the early access period was for.
Also VHFM is not a good solution to the tessellation problem at all in its current state.
If your answer is ‘just use UE4’ then we really are going backwards.
2 Likes
“Well you just said the answer yourself, which is correct. If you still need tessellation, then don’t use UE5.”
I’m sure people are sick of hearing this by now LOL
I’ve heard this “solution” many times before and I actually explored that option at one point myself last year after becoming really frustrated with the Tessellation situation. NVDIA rolled out RTXGI which is a really great addition to all of their other UNREAL based plugins, and I was willing to give that a shot. The only problem is that right before UE5 EA came out, I traded off my RTX 2060 and got a RADEON RX 6900 XT. AMD has always been my preference and the only reason why I purchased an RTX in the first place is because those GPUs were the only ones available for raytracing at that time. But this all changed by the time EA launched. My point is that due to this, I could no longer use NVDIA plugins and features.
For others like me who simply made the decision to go with a different GPU, unaware of what would come in the future, those individuals would have to choose between Lumen and Tessellation which is still pretty crappy. And honestly, I don’t think anybody was aware of how complicated things would be until they actually started using the software. This is why so many complaints about not being able to create terrain or use splats anymore only came after people had the opportunity to use NANITE. So saying things like we all knew what we were in for is not really the case.
I think we all had it in the back of our minds that even with the introduction of NANITE that EPIC would have a solution prepared to deal with this issue in engine along with the release. Not just for hobbyists but also for major studios that have adapted to the old way of doing things and now have to delay projects to change their entire workflow. Even the Modeling plugin that was incorporated into UE5 from way back when EA first launched, still as of right now cannot displace on color channels.
“It’s been well known for around two years now that tessellation was going to be depreciated.”
Its interesting because Hardware Raytracing has been deprecated as well for around the same amount of time, and yet it made it into the official version of UE5, working along side Lumen. That’s something which I was almost certain would NOT be the case. I even stopped depending on Hardware Raytracing because I didn’t want to get used to something that would be gone down the road. I assumed that Lumen would be a complete substitute for it.
And Lumen is definitely not finalized as there is much room for improvement, like better reflections or translucency for instance. These are things that work when utilizing Hardware Raytracing that currently don’t work well for Lumen, or at all.
Epic seemed to also be aware of this too I assumed which is why they didn’t remove Hardware Raytracing. Its just strange though that this way of thinking didn’t also apply when it came to the situation with Nanite and Tessellation, since Nanite is still not quite there yet either.
Maybe there are other factors that distinguish the two situations and its more complicated when it comes to dealing with mesh features? I dunno… I cant fathom how incredibly difficult it must be for these people to manage all of these features. However, in this case it just seemed like Epic wanted to make everybody have to depend on NANITE, their personal vision of how dealing with meshes in a game engine should be from here on out. The bit of Irony though (as others have pointed out) is that the biggest sell of this feature was supposed to make things much easier for the developers, but instead its actually put quite a heavy burden on us.
I feel like both Tessellation and NANITE could have worked along side each other, Tessellation simply being disabled on NANITE meshes and developers being the ones to make the creative decisions in terms of what is best for their projects when managing the optimization process.
2 Likes
I agree with you on that.
Also I cannot speak for anybody else but I personally only want Tessellation in order to smooth my skeletal meshes. Its really tedious having to manage 2 different versions of a character skeletal mesh (one low poly for animation and the other a high poly representation for unreal) when I could just use the low poly mesh and Tessellate it in UNREAL. Even if Nanite eventually supports WPO (I think 5.1 actually does) or skeletal meshes, there is no telling if things will work the opposite way later on down the line, with NANITE being able to add in triangles and do surface smoothing as well.
Following the release of UE5 EA however, I actually figured out how to deal with the terrain issue. I work in Blender so its really convenient for me but I don’t see why what I’ve done cannot be replicated by someone else using their own preferred 3D modeling tool.
My solution may not help people who’s workflows are confined only within UE5 though, and when I suggested what I had discovered to people last year, I was told that this is just too many steps.
But I couldn’t figure out any other way to do it 
And maybe people have come to this same solution already? I don’t know… What I do know is that you cant get good results just popping open an external program, painting some textures on a mesh and calling that terrain. And the “baking out” nonsense as some have suggested is ridiculous PS2 era stuff resulting in you losing a bunch of texture quality. You would also be forced to displace evenly across all surfaces of your landscape which is not realistic from a visual standpoint. The way that people are suggesting everybody do this takes the creativity out of every aspect of building your world.
Having to stack a bunch of meshes on top of each other just to create simple dirt paths, puddles or small patches of land is INSANE! Its WAY more time consuming than how things were before where we just painted the land variations. I have no clue why someone would think that something like this is even a good idea or convenient for an artist because its not.
Anyway in order to make this work I needed more flexibility than that and also to be able to take advantage of NANITE as well. This means that using color channels is essential which is exactly what I did here: Custom Terrain Using Nanite With out Tessellation
Doesn’t change the fact. People just don’t want to accept it.
It was on optional toggle, tessellation was rooted in deep. Apples and oranges. HW tessellation is super dated and inefficent tech anyways, so it’s good they have ripped the bandaid off. Hopefully they enable WPO with nanite meshes, then the world will be at peace again.
Yes and no, you can still paint out meshes with brushes. Not positive, but you might be able to autopopulate foliage brushes on the terrain using layer weights and grass nodes, but it’s been a long while since I’ve messed with any of that. Hopefully they will rework the terrain system at some point.
So again, if you’re really needing tessellation, a 15+ year old and very dated tech, stick with UE4 for now.
2 Likes
I dream of the day Nanite and WPO work for landscapes
Landscapes are overrated in new bright era of nanite.
But indeed would be great if (when) landscape gets those blueprint nondestructive tools to level of something like terragen or gaea has. And then it produces nanite quality mesh out of it. So please add real nodes for generating terrain to current ones.
My dream for landmasses is that i use splines to mark: “here be mountain, here be forrest, and here be dragons”, then click generate.
Yes such stuff would be awsum.
And then how will it allow for displacement on surfaces made with texture painting?
Right…WPO or tessellation.
I believe you can still attach nanite meshes to the skeletal mesh. Is this not what was done in Valley of the Ancients?
Yeah this is some sort of bad joke that tessellation is gone.
3 Likes
It’s already being worked on - nanite landscape - the main branch, so it will come with 5.1.
Dunno if “Nanite + landscape + WPO” needs additional work, although WPO is reportedly being already supported on the main branch too. Together with masked materials?
5 Likes
Thats such good news!
Yes I was very happy to hear this. Hopefully with WPO as well, since without texture-driven displacement from the landscape material I think there wold be not much benefit to that.
They have confirmed they are working on propriety Nanite tessellation though in the AMA which is great news. At the moment, making Nanite assets has to be done externally which is a HUGE workflow issue if you need to change something about the asset after import (since any changes to the texture require the mesh to be re-displaced…and UE no longer does that natively since Nanite does not support displacement).
So hopefully we will see that in 5.1 or soon after.
1 Like
omg yeah, wouldnt WPO with Nanite Landscapes sorta work as a bandaid/replacement for tesselation?
Yes it would, for landscape.
For mesh, given Epic stated WPO would be usable, I could very much live with that too.
1 Like
This may be an obvious question, do we have any stated release window for 5.1 ? Just wondering in case i can plan around it.
Likely sometime between now and the end of the year, I’d assume.
1 Like