To HDRP now, later, or not

Our game is about a year in the making now, with most of the core functionality in place, art direction mostly set, but with visuals generally being the area most behind. It's a 3D stylized Rpg.

We have two main areas I can think of impacted by HDRP that are linked to gameplay. Ie we have minor to considerable code/technical design linked to them. Our water system Crest, and our terrain shading using Microsplat.

Eye candy stuff I don't so much care about. Sky, scattering, fog, etc.. Doesn't seem high risk to me.

All told though my gut tells me it will be cheaper to switch over at a later time. Where there will be fewer unanswered questions and of course a more stable HDRP itself. I'm having a hard time thinking of why switch now, from an all around perspective technical and business.

As to if, the big advantage I see is simply not getting stuck on something Unity obviously wants to ditch asap. We use DOTS extensively, I don't see anything related to built-in getting much love going forward.

We do render a lot of stuff, our game kind of has a focus on scale. So it seems like in theory HDRP might help us there. But that's too early to know yet, both in terms of our game and HDRP. Would our game look better in HDRP? I think that's doubtful, to an extent that it would actually matter. So I keep going back to not wanting to get stuck on a major piece of old tech as the main reason to at some point switch.


Any advice/thoughts/things I have missed welcome.

 

I think it's hard to advise someone on this without knowing the specifics of your project. However, I made the decision to move to HDRP last December, after some consideration. In my case, the very specific reasons I wanted to be on HDRP included:

• Ability to use ShaderGraph and Visual Effects Graph
• Support for more lights

I also had a kind of vague hope that things would just look better in HDRP out of the box, but I'm not sure that's really the case.

Anyway, I think it's important for you to enumerate the specific things you want out of HDRP, as well as list the things you'll be giving up. Realize that even with HDRP coming out of preview in 2019.3, there's still a fair number of features that just aren't there yet. And be very aware of the deprecation of Enlighten, which will be a big issue if you're using Realtime GI lighting in your game.

Also consider whether your game would be better suited to URP instead of HDRP. There are some tradeoffs there, but depending on what you want to get out of HDRP, URP might offer you the same things.

Anyway, this probably hasn't been a whole lot of help, but maybe you could list the features of HDRP that motivate you to migrate to it. If it's just out of concern that the built-in renderer will be going away, I'm not sure that's a big concern.

 

There is no substitute for playing with the pipelines yourself, either in the context of a copy of your project, or a different project just for learning and experimentation.

The built in renderer wont get much love and wil be abandoned eventually, but that wont happen very soon and switching for the sake of it at this stage doesnt make that much sense.

 

For stylized 3D, definitely worth evaluating URP (ex LWRP) as well as it'll have lower initial overhead and you probably wont be heavily utilizing HDRP specific feats anyway on that type of a project. But yeah, it should be easy enough to test some scene on both to see how it's going to work out.

 

Considering this:

And this:

I'd suggest the Universal Pipeline. Cheaper to move to, and URP is much more stable and will get much more love on the Asset Store than the HDRP at the moment. For instance, Both CREST and Microsplat seem to support the URP, with no mention of coming HDRP support, so there is that as well.

We're using HDRP because we're targeting early post-2021 release and we care about volumetric, true physically-based shading, SSAO, Motion Blur and so on.

 

Our game is stylized but we need realtime GI and volumetric. The I don't care so much comment was more about those things don't need to be fully there right now but they do eventually. We can punt on things not tied to gameplay until later. But existing gameplay that integrates tightly with rendering related stuff, that has to work now.

The only third party assets we use are Crest, Vegetation Studio, and Microsplat. But it's all multiplayer at some scale with advanced usage of all three, and on top of that mostly jobified. So significant engineering already there. If/when we moved to HDRP all the gameplay tied to those has to continue working.

Crest I have the highest confidence in. We could probably get the functionality we need working now, and the Crest team I'm sure will have HDRP ready.

Vegetation studio we could probably make work, I'd rather not do the work myself but I see that as an option. I think the author will support HDRP, but how soon not sure.

Microsplat is problematic. The author doesn't like HDRP, I don't see him making it work before he has to. And the HDRP terrain shaders they basically just copied the built in approach. So for us where we use a good number of textures for some features like farming, cave entrances, etc.. That built in approach sort of sucks performance wise.

 

Then stick with built-in because Enlighten dynamic GI is not supported in 2019.3+ versions of Unity and no production-ready alternative is planned until 2021.

They say that they are phasing it out in 2020.1 but we're already having a lot of troubles in 2019.3 with HDRP and Enlighten... and they are not answering to any mention of these bugs in the forum and through the bug tracker. AMong these problems are point & spot light bounce not working, which is the whole point of dynamic GI.

 

To be clear I don't have an issue with HDRP in particular, my issue is that unity has split the shading pipeline into many very difficult to support pipelines without providing an abstraction layer between them unless you use a shader graph. Shader Graphs are not viable for what I do, and should generally be for small custom shaders close to the art, not shader systems which attempt to solve large configurable problems like terrain shading.

HDRP is still under development, and changing rapidly, so supporting it means having things break all the time. Meanwhile, despite LWRP being "production ready and unlikely to change", it's still rapidly changing as well. To top it off, Unity has directly expressed to me that they "don't document shader code because it changes too fast", which makes it even harder to support multiple SRPs across multiple versions of Unity and multiple versions of the SRPs. In reality, it's a huge mess created entirely by Unity, and they were prewarned that this would be the outcome and so far have continued to ignore it.