What keeps you from developing a tech similar with nanite?

Everytime this topic come up, the answer have always be the same.
Unity main consumer & studio is mobile and vr. So it has to focus on tech that work for their main audience. I mean, just look at the disparity between hdrp and urp usage.
Hdrp have dozen more features. Look wayy better. But people still hate it and asked unity to axe it every other month because they think it take dev time out of developing for urp :joy:

2 Likes

As far as I know, main demand for the Chinese market for such feature is not for games, but for industrial applications (eg: rendering complex CAD models).
As mentioned in this thread, the gaming market there is mostly mobile, and even if mobiles can be powerful devices, games are very complex (UI, size of the world,
) and still need to run on a wide range of devices and cannot be played with ice cooler stick to them.
With industrial applications, most run on PC, or on « mobile » hardware which can be more selective for embedded/in-car screens, product configurators or maintenance/training apps on tablets.

Multiple HDRP games have shipped on Steam Deck (Harold Halibut, Lego Builder’s Journey, Syberia The World Before), and most shipped with versions before U6 where in the meantime we have done many performance improvements at different places of the pipeline. Just STP alone should help running various advanced effects on these amazing portable device.

1 Like

Yep, it’s possible. Not easy. Especially with larger environments. Slime Rancher 2 supposedly runs but not super well.

Would love a good scalable demo to accompany that new terrain system to show how to actually optimize those scenarios (Steam Deck would be a good target for this).
The 2022 HDRP terrain demo runs terribly.

True, infact has anyone ask for HDRP to be axed this month?

tbh the only reason I don’t care for HDRP is because you can’t target low end and high end in one project and be able to scale things via turning off or tuning settings yourself
 and ironically you still end up having to use third party addons because those developers seem to have more attention and passion for the features they add and also more onto of actioning requests and fixes directly
 unity add stuff and then move on and forget it maybe eventually picked up and resolved in a proper issue report.

They also promoted URP (I mean Universal implies something right) as the BIRP alternative and guess what other addon developers made stuff for only that SRP, could name a few that only support URP, the actions for splitting up the pipeline and dividing the market and developer attention has had far reaching consequences. So seems to me it will only be the Unified pipeline that potentially fixes this entire mess and Unity get back to making things that aren’t just for very specific pipeline and instead makes things that scale better.

Sad that because there are things in HDRP (like https://docs.unity3d.com/Packages/com.unity.render-pipelines.high-definition@14.0/manual/images/HDRP-MaterialSample-ShadowsTransparency.png) that I wish I could use in URP for higher end setting, but I will not bother touching HDRP for a project frankly the last time I checked a HDRP project for the ocean shader the performance was just bad (probably just terrain/tree/detail rendering ancient tech Unity still use that was the issue with the scene but the thought was like might aswel just UE if you’re thinking of using HDRP and targeting that kind of detail) Anyway it good they are working on improving these things but this Unified pipeline should have been something started like 4 years ago.

This game running at 30 fps on a Steam Deck is NOT a good look for HDRP.
brave_lA3qgiUDCP

Edit: I see it gets to the mid 40s. Really not a good look, still.
image

Does it use specific HDRP optimization tricks like updating global light cascades on demand, managing shadow atlas manually and using ONLY ONE camera.
Cuz HDRP sadly is not production ready without target optimization for this specific SRP

In all my tests HDRP is much slower than URP, i would never consider using HDPR to make an actual game, as can get similar looks in URP with much less performance impact.

That makes perfect sense, adding a mode for lazy importing massive data is so far from optimal that i was always wondering how it is even in question to use something like Nanite, than just use the optimal optimizations like batching and direct instancing.

I really like this guys channel in general as he just rips into UE5 optimization.

Even just disabling nanite and using the actual assets as is can have performance improvement in a final game.

His big issue with UE5 is that TAA (Temporal Anti Aliasing) is essentially a keystone optimization technique which all other optimizations rely on.

Lumen, hair, foliage and fog look terrible with TAA turned off. The problem with TAA is it causes ghosting and smearing when the frame is moving which tends to make the frame blurry. They argue UE5 is heading a bad direction where every future game is going to be blurry, smeared and have ghosting because there’s now no alternative to using TAA.

1 Like

We need proper virtual textures tech available - that would give Unity serious ramp up in overall visual quality and fidelity.

Nanite - this is a marketing gimick. First hand knowledge from one of the U5 dev teams, working on AAA title- it’s a huge pain in the ass, most of the time not worth the effort, increasing massively optimization effort to make it work right.

Nanite is win for artists, who can dump zbrush meshes into editor, skipping all optimizations and call it a day.

Lumen - Unity has promising APVs, need to play with it more to see if it goind to hold up for a more complex environments than just trenches in a grassy fields.

Virtual Texturing, c’mon Unity, you bought Granite in XIV century, it’s time to wrap it up and release into the wild - enough of this mobile crap.

I will second this. Texture Memory is probably the largest bottleneck for lower ended devices as it limits baked assets.

Very fair point.

Human eye detects repeating patterns very efficiently, breaking the illusion. Virtual textures would allow to push detail at scale that would remove that problem, w/o costly shaders, across all platforms.

I do really wonder what’s going on with this - all major engines have that for years now, yet Unity, despite owning the tech, stubbornly ignores it :thinking:

Individual mesh complexity does not contribute to the overall visual fidelity that much as textures, lighting and motion do.