Heya,
Wondered when the ETA for Universal Pipeline support for raytracing is (rough ballpark is fine)… I assume it’s coming because it’s not tied to HDRP at all API-wise, and it seems like a perfect boost for titles that are being ported from Universal / LWRP upward, adding much needed quality for ports with pretty much a few clicks. To me this is a very good improvement.
If it’s not coming then please insert confused.gif of choice 
(probably fastest way to shift folk from builtin to URP too)
Would be interested in this as well. Maybe the URP section would be a better place to ask for this?
Feel free to ask there too if you like.
I think we will get news when they will officially unveil RTXGI beyond stage demo of “upcoming features”
HOWEVER is the sources of the HDRP have the ddgi code, I want a pick at it, is there link?
The most interesting part for me is that Unity designed the raytracing API to be independent of any pipeline. Of course, there’s quite an amount of work to be doing to support it but I think it makes sense for Universal to have basic quality of life features hopefully ported down from HDRP:
The basics, that can turn even a simple Universal Lit scene into something pretty stunning, so you can target mobile and console within the same pipeline. If you know you will never go lower than HDRP then you can stay on HDRP and push quality even higher.
But I strongly believe that with Raytracing support in Universal pipeline, adoption rates from Builtin will soar, which reduces how much money Unity spends on keeping builtin around. Give it enough time and it’ll cover the cost of supporting Raytracing in URP, probably. Somehow. Hopefully.
Anyway get on with it Unity. Make some magic already.
And after URP gets as bloated as the old built-in pipeline, I suggest Unity start a new one that is actually fast. Lightweight Pipeline has a nice ring to it if they’re looking for name ideas.
Can you be more constructive ? Noticed a bitter trend to your posting. You alright?
EDIT: Maybe never mind all that.
I wasn’t very well, I’m better now. Thanks for asking.
Well I was just concerned. I mean you’re a regular but if you’re going to be angry, the unity forums can’t solve that problem.
Back to the original question of when the URP might support raytracing, I suspect the answer to that is there are no plans to support raytracing in the URP. Raytracing is still bleeding edge tech, and the URP is meant more for universal portability rather than bleeding edge graphics tech. The HDRP on the other hand is where you’ll find all the new bells and whistles.
So really this is a situation where most raytracing uses requires really high end hardware to be of real practical use for things like GI, or reflections. Shadows might come sooner, but as seen in games like World of Tanks and Modern Warfare, current implementations come with significant limitations.
Yeah. I do see the merit in investigating this stuff one day, but now isnt the time. Especially since plenty of people are still waiting for what they consider to be unacceptable limitations of core URP lighting etc to be overcome.
That’s the thing, I’m not sure it’s worth overcoming the limitations to reach what Builtin was. That’s old thinking.
A 200 dollar card offers more than acceptable RTX performance now, and in a year, I can’t see any non raytrace GPUs being released outside of mobile or embedded.
It seems like I’m way too early here, but the rate of change is you start now and in 2 years time you have something ready for market. Unlike how things have been so far.
LWRP was suitable for the slow stuff, Universal aims to be what? LWRP with more lights and some post effects? I don’t see Universal being useful unless it solves what people are really asking for:
Something that can do fast mobile stuff… yet when they port, they want a one button solution that brings out all the sparkles for PS5, next xbox, high end rigs.
I don’t see this being too soon even slightly. I see it as being quite late actually. Even porting the (raytrace) code from HDRP and tarting up would take a couple of years - long since past next gen console releases.
Really I just want Unity talking about this seriously. They made the API SRP neutral for a reason. Customers have asked to do the high end from the low end painlessly since SRP was introduced. Don’t give them something from 2 years ago in 2 years time. Give them raytracing in 2 years time.
Disagree. Dont want to argue about it though, will wait and see on that one.
Again I wait to see what Unity are actually going to do with this thing. So far all I really know is that there is an adoption problem and they decided to rebrand LWRP because they decided part of the adoption problem was down to a perception problem. And they are nowhere close to adding enough actual features to Universal to meaningfully shift perceptions.
Its clear plenty of people want this and this is one of the things that also causes perception problems and lower than desired adoption. Its a fairly complex subject though, I dont think RTX/DXR etc are a magic fix for this. I do quite agree that they could be part of the solution one day, but I think the core of Universal needs a bunch of stuff doing to it on a more general level first, including some stuff that would need to be there to get the most out of DXR anyway.
I do not agree with this estimate, it could be done much quicker than that if the pipeline is in a suitable state to receive it. But of course ‘tarting up’ is an open ended part of the mission that could take 5 days or 5 years depending on the details of what you had in mind.
Theres a lot I could say about a broader version of this topic, that I mostly wont because I’ve said it before. But I will say there is a bunch of other critical stuff that they should have been talking about beyond the initial announcements too, eg the whole GI situation. I dont think raytracing related announcements based on harfware like RTX on APIs like DX12 are the most pressing thing that people are demanding to know the future of, its much lower down the list and although it can be part of the solution in some areas it is not broad enough to fill all the essential gaps in the medium term.
Given where raytracing is now, from hardware to graphics APIs to how far Unity have got with HDRP implementation, I think they are targeting the right areas first and are doing things on this front in the right order. I’d give that some time to mature and see which parts of it get used in games by people beyond the niche subset of high-end uses/markets (eg non-gaming) that have the luxury of using all that DXR has to offer.
As for reasons why they make the API SRP neutral, dont forget that when they first started going on about SRP’s, they put a lot of emphasis on people making their own pipelines or completely modifying the ones Unity develop. So they still have to cater to that too, even though such aspects arent what most users who post on forums are into at all, they need Unity to do all the heavy lifting.
It depends on what you mean by “acceptable performance”. A GeForce RTX 2080 Super can’t run most RTX enabled games at 1440p 60hz, some titles it struggles hitting 30hz consistently. That’s a $750 GPU. The much more expensive 2080 ti and RTX Titan don’t fair much better there either. The $200 1660 Super won’t even get consistent >20hz frame rates running those titles at 1080p with any raytracing features on. Almost every title that uses raytracing also uses some form of TAA or computationally expensive denoising to hide the noise current implementations have due to the low ray counts even the top end hardware is limited to, and URP currently has no plans to support TAA.
I should note for the RTX cards I don’t count the semi-bogus DLSS benchmarks, since that’s often rendering the games at 900p or 720p (or lower!) and using temporal upsampling to hide the actual render resolution. I’m talking about actually rendering at 1440p.
Basically the only raytracing something like the 1660 can handle is some shadowing, and even then it’ll be limited to the same problems other real-time raytraced shadows have, like having no support for vertex shader deformation, and skinned meshes having a significantly higher cost to cast shadows from.
Yeah I must admit I am curious about hippocoders experience with RTX so far that has made them so optimistic about what its the best solution for, performance, timescales etc.
I have just revisited the ‘How the Lightweight Render Pipeline is evolving blog post’. Here are 3 statements from it which I think are actually reasonably compatible with what hippocoder and myself have been saying, even though we are not on the very same page over these issues. To start with, it can remind us of what timescales Unity have already suggested:
https://blogs.unity3d.com/2019/09/20/how-the-lightweight-render-pipeline-is-evolving/
So yeah, if you are lucky then the agenda scheduled for 2021 is where your Universal raytracing dreams could begin to turn into something, which might be more likely if stuff popularised by RTX/DXR has made it to other cards/graphics APIs has also flourished by then. I still wouldnt bet on it though, depending on performance constraints and what non-raytraced higher-end aspects of URP have improved.
Anyway bottom line for me is that I actually like the 2 pipeline separation, but I can totally see why developers being forced to make a choice between one or the other is often not the sort of choice and flexibility they were clamouring for. There are several possible different reasons for this, such as:
a) But the right choice for me is both pipelines and there isnt much tooliing/conversion to help with this - pain point!
b) I am not confident making the correct choice because there is a lot of stuff I need to read and learn first, including stuff that is still half-baked or has changed during the evolution of the pipelines.
c) I’m chasing moving targets with unclear roadmaps, sample projects of different sorts that end up only working with certain editor and pipeline versions. Unity are getting a bit better at eating your own dogfood but not over a sustained period of different engine versions and pipeline evolution.
d) I like choice, but not when I feel trapped by the choice and unable to easily change mind.
Some of these will naturally become less of an issue if things settle down and mature nicely. Others will have to be pro-actively tackled. Universal and the stated motives for it acknowledges a bunch of that, but many of the things it aspires to are not things I expect to get to try myself or hear technical discussions about for a long time to come. And just because URP is going to cover much more territory in future, doesnt mean the more exotic stuff will exist anywhere but HDRP. If the technique doesnt scale down usefully, I dont think it will belong in URP. If aspects of raytracing trickle down into the low-medium end successfully in the years ahead, it makes more sense.
if I were going to be targeting next-gen consoles in their early days, choice of pipeline I would make now still depends on the sort of graphics my game needs. If I wanted to make something that ‘looks like next gen eye candy’ the I dont think there is anything Unity could possibly tell me today that would make me choose anything other than HDRP now. And that would change till I saw the results of their 2021 plan for URP, so any project of that sort I start for next gen consoles in the next 18 months is highly likely to be HDRP based.
As long as they roll back a GI solution that does what enlighten did yesterday (ye old time).
Have a look at unreal engine people… Years ahead!.. Raytracing is mainstream and unity is only experimental HDRP only??? What about support for Standard Pipeline? How about DLSS 2.0?
I really doubt standard pipeline will ever get new features development now with the SRPs being “stable” (right…) and URP probably being in line to replace standard completely in the future years. IIRC the core raytracing graphics API is available regardless of pipeline so if someone needs it and has the knowledge they can always do their own implementation.
With other points I 100% agree.