Very hard to disagree with this though. Sorry that there are two threads. Maybe that’s because it is such an important issue that doesn’t seem to have been addressed though.
My understanding from some of the lecture videos by EG is that Virtual Heightfield Meshes are the intended replacement feature here (rather than Nanite necessarily). If that’s the case…then they needed to have been in a state where it is as-easily implemented in projects as tessellation and supporting all of the same situational uses before tessellation was deprecated. Working on new tech is great. Deprecating old tech before the replacement is ready? Very bad.
As for Nanite - Nanite is great and very neat technology, but it does not support landscapes and texture painting as of yet…which is absolutely essential (unless you fancy making your entire landscape in an external platform without Heightmaps, yikes). Regarding the UE stream as well:
Karis said that the algorithm that generates the nanite mesh is too slow for real-time.
So it’s unlikely that Nanite will be able to support landscape deformation in real-time on release anyways, unless something major has changed here that has not been publicly disclosed
I have to assume this is because UE5 Early Access is…well…early access. If that’s not the case though I’m not sure I can see myself using it for serious projects. For me, the idea of going backward to older versions of the engine for a more usable feature set is fairly disappointing.