Search Unity

  1. We are migrating the Unity Forums to Unity Discussions. On July 12, the Unity Forums will become read-only. On July 15, Unity Discussions will become read-only until July 18, when the new design and the migrated forum contents will go live. Read our full announcement for more information and let us know if you have any questions.

Official Enlighten deprecation and replacement solution

Discussion in 'Global Illumination' started by Jesper-Mortensen, Jun 19, 2019.

Thread Status:
Not open for further replies.
  1. elbows

    elbows

    Joined:
    Nov 28, 2009
    Posts:
    2,502
    A rather one sided discussion that has been completely devoid of transition help so far. I understand the unfortunate engineering/enlighten commercial relationship realities but please dont promise to help us transition when no help is actually on offer, when there is nothing to transition to, and when there is not even much in the way of discussion, with no official information forthcoming for many months.

    Even when DDGI (Now known as RTXGI) was shown at Unite Europe, I dont think we were given any context for it, nobody said whether this would be the only future solution or whether there were other plans too.

    Users interested in realtime GI for HDRP who might be confused about the current realities, or find them hard to believe, can at least see this new post from Sebastian Lagarde which confirms what we already thought we knew: #9
     
    Salvador-Romero and ftejada like this.
  2. elbows

    elbows

    Joined:
    Nov 28, 2009
    Posts:
    2,502
    I'm also aware that I'm in no position to make demands.

    But I will say that the absolute bare minimum I will accept from Unity on this front is that you actually be ready to show and tell us what your replacement solution(s) are, no later than GDC 2020. I dont expect it to be ready for us to use by then, but for planning purposes the information void must be filled.
     
  3. Rich_A

    Rich_A

    Joined:
    Nov 22, 2016
    Posts:
    338
     
    Salvador-Romero likes this.
  4. konsic

    konsic

    Joined:
    Oct 19, 2015
    Posts:
    995
    2020 will be weird with lighting in engines. Hardware is there for RTR, DXR. Cryengine is hardware and API agnostic. Enlighten is mostly proprietary.
    Still performance is not there yet to ship a product for wide systems.
     
  5. Rich_A

    Rich_A

    Joined:
    Nov 22, 2016
    Posts:
    338
    A point of reference about best-in-class lighting and shadows for Nintendo Switch:



    Seems reasonable to target any non-DXR, URP solution against that,
    and HDRP against whatever the next-gen of consoles (with integrated raytracing hardware) can handle.

    And a warning of what ports designed for high-end PC and consoles and taken to Switch can look like: (although using Unreal in this case)



    I think there will be a very clear divide between games targeting Switch/Low-Mid Range PC/Mobile, and games targeting NextGen Console/High-End PC.
    I hope that URP/HDRP accurately reflects that split and is appropriately optimised. For example, HDRP should just formally not support mobile, switch, TVs, Mac, tablets etc, to allow the HDRP development team to focus on targeting high-end PC and next-gen console.

    I also think HDRP was released way too early, it should have been held back a year for more internal development. Its pointless to have it without the new lighting solution and without some previews of next-gen consoles available - it could have been built from the ground up for them. This would also have given URP more time in the spotlight.
     
    Last edited: Dec 5, 2019
    DGordon likes this.
  6. Bordeaux_Fox

    Bordeaux_Fox

    Joined:
    Nov 14, 2018
    Posts:
    589
    I think the lack of information provided by Unity is just a prove, that they do not have a clear vision for the new GI replacement. Or that new system is in a "at risk" state.
    Otherwise why hold information back if that is not the case? So for me, Unity is no longer a good partner if you need a strong realtime GI solution with AAA graphics right now.
    Despite HDRP is nice, I would underline the other statements by users that HDRP is very useless with no proper Realtime GI. Because what is the benefit of volumtric light when it gives you no bounce? When you cannot have changing your sun light and have proper lighting response. Also how long Unity know SRP is coming and in order to launch that, a new realtime GI system by Unity is needed? It is like "So here we have HDRP. Oh wait a minute. Enlighten is to expensive and also does not fit our needs. But we did not develop a new GI in the time HDRP was developed. Let's start now and having a gap year."

    I would really like to know who was responsible for that fundamental bad decision.

    But watching the latest blog posts by Unity, I think they give up more and more for serious gaming. The new big thing for them is architecture and car demos ...

    And I highly doubt that new GI will be good when even cannot develop proper contact shadows and decals. These things are still unstable even in the RC.
     
    Last edited: Dec 6, 2019
    Salvador-Romero likes this.
  7. elbows

    elbows

    Joined:
    Nov 28, 2009
    Posts:
    2,502
    Thats not my conclusion at this stage, for now I just add lack of information to the list of symptoms of their big timing mistake which you mention.

    I never think that - its an attempt to expand into other markets, not give up on the gaming one. Unitys most obvious competitor has been doing pretty much exactly the same thing over the same period of time. I know some people dont like it because they think it has clouded Unitys priorities, but I dont agree, especially as I use Unity for non-gaming stuff myself, and because Unity made the same sorts of mistakes back when games were their sole focus.
     
  8. Bordeaux_Fox

    Bordeaux_Fox

    Joined:
    Nov 14, 2018
    Posts:
    589
    I don't think so. For their architecture stuff, they always have new updates but when it comes to GI, they have nothing. It is not just the new stuff. Also think about the lack of care for Enlighten before they decided to drop it.

    So all in all, no wonder game developers feel Unity has some new other priorities.
     
    Salvador-Romero likes this.
  9. konsic

    konsic

    Joined:
    Oct 19, 2015
    Posts:
    995
    What about complete rendering overhaul with C# job and Burst? Not the coding itself, only using it for background rendering, lighting and GI.
     
  10. Bordeaux_Fox

    Bordeaux_Fox

    Joined:
    Nov 14, 2018
    Posts:
    589
    I think that would delay everything even more because these technologies are still in preview and need to grow stable.
     
  11. elbows

    elbows

    Joined:
    Nov 28, 2009
    Posts:
    2,502
    Most of the features they added for non-gaming purposes were low hanging fruit that did not involve insane amounts of engineering effort, or were part of things that had gaming uses too like HDRP.

    It was just the same with VR support - people who didnt care about VR moaned that it took away resources from the areas they wanted to see prioritised. I had different thoughts on that because I do not see this stuff as a zero sum game.

    Anyway, none of this changes the fact that Unity dropped the realtime GI ball in a manner that deserves ongoing criticism. This and their continued inability to give us a useful roadmap would have caused me to ditch Unity by now, if there was an alternative product that met my needs. There isnt, so I am still here.
     
  12. Bordeaux_Fox

    Bordeaux_Fox

    Joined:
    Nov 14, 2018
    Posts:
    589
    I wonder if they have it on their bill how much they loose when clients who need realtime GI will go away. Or somehow their internal statistics showed them only a few devs were using Enlighten at all.

    I think the only advantage Unity has over Unreal is that in Unity the usage of Enlighten is free when in Unreal you need an own contract which is out of question for indies and small teams.
     
    Marc-Saubion likes this.
  13. elbows

    elbows

    Joined:
    Nov 28, 2009
    Posts:
    2,502
    I expect it was a combination of technical and commercial relationship reasons that the Enlighten thing came to an end. In terms of user demand for features, I expect Unity always knew Enlighten wasnt a hugely popular solution, and also saw many requests for decent realtime GI over the years. I suspect they long since concluded that decent realtime GI would give them a competitive advantage, but other priorities, the immature state of new render pipelines, and engineering issues with actually delivering with the sort of GI people want, and various aspects of Unitys internal culture and decision making processes, have lead to very poor timing on their part. Part of the problem is the lack of 'perfect' realtime GI solutions in general (not Unity-specific), yes there are other alternatives out there that can serve the needs of certain projects, but nobody has really nailed realtime GI in the way other requirements of realtime graphics engines have long since been solved to the general satisfaction of most.
     
  14. Bordeaux_Fox

    Bordeaux_Fox

    Joined:
    Nov 14, 2018
    Posts:
    589
    I think nobody is demanding from Unity to reinvent the wheel for realtime GI. We just want to have at least one system providing bounced light and update probes. Of cource different projects have different needs. But that is Unity's part to deliver a tweakable solution with enough settings and also easy support to extend it. It also begins with document things like the lighting data asset. Of course there is no really breakthrought and you cannot rely on RTX if you want to support a wide range of customers. So yes for the new console generation and high end RTX pcs but we should not exclude the not so lucky customers from rich graphics. It is also the question how much frames a RTX based solution will eat. Alot of players are not willing to trade huge amount of frame rate for some better reflections and ambient if they see no difference at all.
     
    EricWilliams and Lars-Steenhoff like this.
  15. konsic

    konsic

    Joined:
    Oct 19, 2015
    Posts:
    995
    Can raymarching method be used for calculating GI ?
     
  16. dgoyette

    dgoyette

    Joined:
    Jul 1, 2016
    Posts:
    4,215
    I've been under the assumption that I need Realtime GI because I have fairly dynamic lighting conditions. For example, flickering lights, where the main light in a room occasionally flickers, which should result in the room being very dark for a moment, before returning to normal brightness. My understanding is that if I tried to do this with baked lighting instead of GI lighting, the walls would still be extremely bright as though I hadn't turned off the lights at all. But I wanted to make sure I wasn't overlooking something.

    Is there any approach, using baked lighting, to having the main light in a room change in intensity?

    For example, here's me turning on and off the mixed lights in my scene after having baked lighting. There's a small amount of change to the lighting on the walls, but it mostly looks full strength. That all seems like what I would expect from baked lighting, I just want to make sure I'm not missing something simple that could free me from my realtime GI requirements.

    BakedLighting.gif
     
    ftejada and Rich_A like this.
  17. neoshaman

    neoshaman

    Joined:
    Feb 11, 2011
    Posts:
    6,493
    It's already done, every voxel GI are raymarching (that's why we use voxel), every ssgi are raymarching too (at the pixel level). The only difference with raytracing is that ray tracing use different type of primitive and resolution (a structure of triangle and intersection test with rays). These two technique are compatible (you can raymarch globally then raytrace locally, and vice versa).
     
    DragonmoN and konsic like this.
  18. Noisecrime

    Noisecrime

    Joined:
    Apr 7, 2010
    Posts:
    2,058
    You could always bake the scene two or more times to get different baked lighting conditions, then at runtime switch between light maps. This will work for simple ‘boolean’ type operations ( main light on/off ) but quickly becomes convoluted for anything more complex. indeed it may just prove impractical based upon the type of project and scene environment.

    I’ve not actually done this myself yet (just researched it was possible) so I’m unaware of any issues such as performance cost of performing the switch. As I said it’s probably perfectly usable for simply turning a light on/off occasionally, but I’d want to test it for say a constant flickering light, checking performance, garbage collection issues, memory usage etc.

    Also you might want to test if you would you lose dynamic emission from materials without GI.

    Edit link for light map switching https://forum.unity.com/threads/how...-for-night-and-switch-them-in-runtime.319920/
     
    Last edited: Dec 29, 2019
    dgoyette likes this.
  19. dgoyette

    dgoyette

    Joined:
    Jul 1, 2016
    Posts:
    4,215
    Thanks. I wasn't aware that you could have multiple lightmap sets in a single scene. That definitely sounds promising. I'll test this out.
     
  20. konsic

    konsic

    Joined:
    Oct 19, 2015
    Posts:
    995
    What about voxel GI using C#job system? Rewriting SEGI with C#job system in SRP.
     
  21. neoshaman

    neoshaman

    Joined:
    Feb 11, 2011
    Posts:
    6,493
    The gpu is faster because massively parallel, probably there is some usecase on low end and slow update, but that would still be for coarse render. Voxel are volumetric so increase of resolution is cubic, if you increase length by two, the volume increase by 8, you would run out quickly of any benefit. Also raymarching itself is not complex and it's a great candidate to parallelization, hence gpu, the problem is more less coherent access than a cpu, but even then gpu are one order of magnitude faster to computation through parallelism. Then you have to take into account bandwidth and latency bottlenecks to transfer data between the cpu and gpu, etc... it doesn't look worth it as a generic solution, and any specific solution mean you will probably use it for one narrow use case that happens to be a sweet spot of the constraint, if it exist at all.
     
  22. MaxWitsch

    MaxWitsch

    Joined:
    Jul 6, 2015
    Posts:
    114
    Has anyone some informations if DDGI will be avaliable as Preview or beta, or when it will be released?

    Global Illumination is something that would give the HDRP a professional touch.
     
  23. jjejj87

    jjejj87

    Joined:
    Feb 2, 2013
    Posts:
    1,117
    The same here. Can we get prospects of realtime GI implementation?
     
    id0 likes this.
  24. Zergling103

    Zergling103

    Joined:
    Aug 16, 2011
    Posts:
    392
    I was gonna inform everyone still subscribed to this thread that I saw in a presentation (in case you haven't seen it already) has satisfied most of my skepticism towards Unity's intention to use raytracing for GI:



    The simplest way to describe it is (if I understand correctly): realtime light probe grids.

    Performance wise, it appears that rather than measuring the performance impact in samples-per-pixel, you'll be measuring it in samples-per-probe. And probes are much much sparser. This seems very elegant for simulating diffuse light propigation, though it still requires hardware support for raytracing.
     
    Rich_A likes this.
  25. Bordeaux_Fox

    Bordeaux_Fox

    Joined:
    Nov 14, 2018
    Posts:
    589
    Ok, yesterday I was told from Unity that DirextX12 get out of preview with 2020.2, so maybe then we can expect they will also release that probe system shown in the video, because stable DirectX12 could also mean DXR will be finished.

    Nevertheless, this cannot be the only solution because it requires the right gfx card. So realtime GI will be locked to Windows 10 with Nvidia and rich customers ... Maybe by that time AMD will have joined to the ray tracing game, too.

    So they also have to came up with a pure software solution, like Enlighten, only better, faster and tweakable for every platform. (And for free of course, no license issues.)
     
    Last edited: Feb 13, 2020
    Rich_A likes this.
  26. Rich_A

    Rich_A

    Joined:
    Nov 22, 2016
    Posts:
    338
    DXR lighting is absolutely the correct single lighting solution for HDRP:

    - The coming next-gen of consoles will set the baseline of non-mobile gaming hardware for the next 7 years
    - Releasing on these next-gen consoles will itself be a huge bonanza of wealth for Unity devs (like any new platform)
    - Windows 7 is EOL and will be even more EOL by the time these games are actually all released - 2021
    - We have to assume AMD will be on the ball with DXR since they are providing the actual console hardware

    A single, DXR-only lighting pipeline for HDRP would be excellent. Whatever enlighten-replacement system URP gets should remain with URP.

    This will put the pressure on Nintendo Switch (and Switch developers) though, since the gulf between platforms will be enormous.
     
    OCASM likes this.
  27. alexandre-fiset

    alexandre-fiset

    Joined:
    Mar 19, 2012
    Posts:
    717
    We need a solution that can provide dynamic GI on a broad range of hardware, which Enlighten provided. I strongly disagree with DXR being a viable replacement solution under HDRP environment.

    The vast majority of developers will continue to support PS4, Xbox One and Switch for years to come. We need a dynamic GI solution that at the very least work on PS4 and Xbox One, and ideally on Switch and virtual reality.

    DXR is a technology that will truly shine only on the next-next generation, at a point in time where the current and next generation will support it.
     
  28. Rich_A

    Rich_A

    Joined:
    Nov 22, 2016
    Posts:
    338
    If you require last-gen support, just use 2019.4 LTS. You'll have support for that until early 2022.
     
  29. alexandre-fiset

    alexandre-fiset

    Joined:
    Mar 19, 2012
    Posts:
    717
    No support for new platforms such as PS5 and Xbox Scarlett.
     
    Salvador-Romero and Noisecrime like this.
  30. dgoyette

    dgoyette

    Joined:
    Jul 1, 2016
    Posts:
    4,215
    The other issue with that is packages like HDRP don't get much in the way of backporting of features to earlier versions. So sticking with 2019.4 most likely also locks you into no HDRP improvements for years, as well as other packages. That's a lot to have to give up.
     
    Noisecrime likes this.
  31. Rich_A

    Rich_A

    Joined:
    Nov 22, 2016
    Posts:
    338
    If you want support across a universe of devices... low end PC, high end PC, last-gen, current-gen (current as of 2021).... maybe a Universal Render Pipeline is the solution? ;-)

    HDRP and DXR should be built hand-in-hand, and Scarlett/PS5 should present the baseline customer hardware specification for it in Unity 2020 onwards.

    Otherwise the solution might be to build for HDRP, then port and downgrade to URP for last-gen.
     
  32. Noisecrime

    Noisecrime

    Joined:
    Apr 7, 2010
    Posts:
    2,058
    The problem with this attitude ( which comes from Unity themselves ) is that two years of support is simply no longer sufficient to release a product on. While Unity have been pretty good in addressing serious issues across multiple previous versions of Unity, less critical problems will not be fixed potentially leaving a developer in a situation where they can no longer support a product due to say a specific OS update, or build requirement ( iOS and Android have gone through many of those over the years ).

    This situation is getting worse as the churn of new features from Unity is increasing at a rapid pace. Meaning you can be left in a situation where either you simply let your product die ( frustrating your customers ) or have to completely remake it every few years to fit the next great Unity feature that deprecated a perfectly good previous one. It certainly doesn't help that Enlighten was third-party, though i'm not entirely convinced on Unity's track record that things will be improved with using in-house solutions.

    Unity really need to be committed to a feature for much longer or some additional level of LTS to provide developers confidence that a product can be support for a decent amount of time.

    I mean if i were to start say a AA title development today with 2019, its highly unlikely it would be released by the time 2019.4 LTS support ends. That to me isn't a great selling point and I certainly wouldn't be happy starting a project now that used Enlighten, but as there is no replacement I can't imagine how you'd even start to approach the Art assets at the moment.


    With regard to DXR, I love all this new ray-tracing technology and the push towards it, since if it comes off it will simplify massive amounts of rendering technology, getting rid of all those annoying workarounds and tricks we have to use today. However I'm yet to be convinced this is a long term commitment from gpu manufactures or 3D API's and can't really see it going to mobile any time soon if at all. Thus reliance on such technology without a software fallback seems like a folly.

    ah, another frustration I have. ( nothing personal you just seem to be hitting them all ;) )

    This split between high and universal is so annoying to myself. Having been used to the old legacy system I was secure, confident in the knowledge that I could create a project that would be able to hit the widest possible range of hardware, with the greatest potential scope of features. I was able to target mobile and desktop simultaneously and simply enable/disable graphical features that were either not supported or too demanding for lower end hardware.

    That has all but disappeared with HDRP and Universal. Now you have to choose one at the start, make the wrong decision and you are screwed. Want to improve the product in a few years time, you're screwed as there is nothing left to leverage.

    Deprecating existing features and then only supporting them in HDRP is not a solution.
     
    Last edited: Feb 14, 2020
  33. alexandre-fiset

    alexandre-fiset

    Joined:
    Mar 19, 2012
    Posts:
    717
    HDRP has been sold to us as the high definition renderer for PS4 and Xbox One. Heck, the first demonstration of it was running on a Playstation 4.

    A console that currently has no owner / market maybe your priority, but it is not the one of the majority of HDRP users.

    The maybe-cancelled System Shock 3 as well as our own game are made with HDRP to actually ship in 2021 on all major platforms. HDRP is not supposed to be just a thing to render cars in DXR, it is a render pipeline to ship products to customers. How many Xbox Scarlett and Playstation 5 customers there is?
     
  34. nikescar

    nikescar

    Joined:
    Nov 16, 2011
    Posts:
    168
    Amen. If only we could all be so lucky.
     
    florianBrn, Waz and AcidArrow like this.
  35. Bordeaux_Fox

    Bordeaux_Fox

    Joined:
    Nov 14, 2018
    Posts:
    589
    So they confirmed in the latest blog post about HDRP that the new and only one Realtime GI system is coming for 2021.1. They did not say anything new about the features we can expect:
    "Note that HDRP does not support Enlighten, but we are committed to delivering a real-time Global Illumination (GI) replacement solution in 2021.1."
    Meanwhile, Unreal Engine has got Screen Space GI. It really hurts that Unity is so slow, meanwhile other engines evolving their Realtime GI solutions.
     
    valarnur and protopop like this.
  36. OCASM

    OCASM

    Joined:
    Jan 12, 2011
    Posts:
    329
    There's ray tracing too and for screen space GI you can just do something like the Marty McFly shader for Reshade.
     
  37. sergiusz308

    sergiusz308

    Joined:
    Aug 23, 2016
    Posts:
    255
    Brackeys just said that HDRP has realtime GI:


    So what's the deal? Video is clearly sponsored by Unity, nothing bad in that in general, but how this relates to the whole discussion here?

    My project is to be completed in 2020, so in simple words: what are my options for realtime GI in 2019.3, after all?
     
  38. dgoyette

    dgoyette

    Joined:
    Jul 1, 2016
    Posts:
    4,215
    2019.3 still supports Enlighten for Realtime GI in HDRP, though I think it's possible that brand new projects created under 2019.3 won't, by default, show it as an option. Upgrading from 2019.2 to 2019.3 I believe will allow you to use Enlighten in 2019.3, or maybe there's some other way to allow Englighten for brand new 2019.3 projects.

    From 2020.X onward, your only choice for Enlighten Realtime GI is to remain on the standard renderer. You won't be able to use URP or HDRP. You also won't be able to upgrade a 2019.3 project to 2020.x and preserve Enlighten support.

    URP never supported Realtime GI, I think, so that's irrelevant to this.

    So, your practical options are:
    • Stay on Unity 2019.3 with HDRP 7 if you want Realtime GI.
    • Stick with the standard renderer in 2020.X if you want Realtime GI beyond 2019.3.
    Supposedly Unity is expecting an HDRP-compatible enlighten replacement for 2021.1, but I would expect that to be experimental for a while, and possibly not ready until 2021 LTS. We'll have to see how that goes.
     
    OCASM and sergiusz308 like this.
  39. sergiusz308

    sergiusz308

    Joined:
    Aug 23, 2016
    Posts:
    255
    Thanks for clearing things up.
     
  40. Bordeaux_Fox

    Bordeaux_Fox

    Joined:
    Nov 14, 2018
    Posts:
    589
    Have to add a third practical option, choose another engine if you looking for HD graphics and a good Realtime GI. Right now, there is no benefit to start a project that needs a strong lighting solution with Unity if cannot fallback to baked lighting. Maybe I come back 2022 for any serious project with strong lighting but at the moment it is not worth to deal with everything in Unity is under construction and "half" baked. It's a dead end.

    Why should somebody stay with buildt in pipeline when other engines give you far more? Why stay with an older HDRP version just to maintain Enlighten when other engines give you more?
    Right know, the only drawback is to put the complete Unity knowlegde and bought assets to the bin. But that is better than limiting your artistic freedom with old tech and wait time. No thanks.

    You can go on with the old Enlighten workflow but please then do not go to the forums and asked about light leaking, broken probes and worse performence.

    Not to say for procedural and large worlds, this will also strongly fail. Good luck for trying a rather small indoor scene with thin walls at least.
     
    Salvador-Romero and Rich_A like this.
  41. dgoyette

    dgoyette

    Joined:
    Jul 1, 2016
    Posts:
    4,215
    That's completely fair, and seems reasonable for someone starting out a new project. In my case, I'm a bit over two years along with my game, with plans to release this year, so jumping ship to another engine isn't in the realm of possibility.

    Ultimately it comes down to what each project needs, and what its release timeline is. I don't think there's any way to spin the Enlighten deprecation as anything other than a complete fiasco, but I'm sure there are projects that won't be released for a couple of years than can wait out its replacement. And personally, it would be a very hard decision to drop Unity for Unreal, leaving behind a lot of domain knowledge, over a single engine feature. But for those who have less invested in Unity, I agree with your idea of considering a different engine.
     
    Salvador-Romero and Noisecrime like this.
  42. Rich_A

    Rich_A

    Joined:
    Nov 22, 2016
    Posts:
    338
    I think Unity deserves some defense on the decision to separate HDRP and URP, which I think was the correct one. I'm involved right now with heavy optimisation for Switch, and the effort behind that is conceptually similar to the effort of going from HDRP to URP in an established project. For example, all kinds of camera effects need to be switched off, lighting heavily dialed back, all fog turned off...

    Any kind of realtime GI would just never be performant on mobile devices (I include Switch as a mobile device). On my title it will probably be necessary to completely disable all realtime lights.

    Looking to the next 3-5 years I think its clear that a big gulf will open up between high-power PC and Console using RayTracing, and battery-powered Mobile and Switch. Focusing on RTX as the single realtime GI solution for HDRP is justified.

    The gulf between mobile and desktop/console has been understated for so long because of the very weak current-gen of consoles. With the next-gen, porting titles to Switch, even if running them at 540p, will probably just become impossible - if those titles truly took advantage of next-gen hardware.

    Having two distinct render pipelines makes sense - as long as as many things are shared in common, so that switching between the two is at least possible. The posts from Jason Booth touch on a lot of aspects here.

    We also need top-tier Optimisation (mesh optimisation, LODs, mesh baking, material batching) included in the Pro/Plus variant of Unity. I think it makes sense for these to be paid components since they are mostly essential for commercial products. Currently the offerings on the Store in these areas are sub-par. The best Mesh optimisation tool is Simplygon which is no longer fully compatible with Unity.

    People have raised the example of Unreal, but it performs terribly on Mobile. Look at the port of Mutant Year Zero on Switch - all of the visual quality of Unreal is wasted there.

    On top of that choosing Unreal is a non-starter for many because the requirement to use C++ is an insurmountable obstacle, and the 5% of gross sales revenue means that after a publisher and platform cut there is very little left for the Studio.
     
    OCASM likes this.
  43. OCASM

    OCASM

    Joined:
    Jan 12, 2011
    Posts:
    329
    Fourth option: code your own GI solution based on the specific needs of your project.
     
  44. alexandre-fiset

    alexandre-fiset

    Joined:
    Mar 19, 2012
    Posts:
    717
    I believe "cutting dynamic GI from your game design ideas" is the only "solution" as of now and for the next two years, unfortunately.
    • Coding your own solution is a high risk, low reward endeavour
    • Sticking to built-in pipeline does not result in Enlighten working on PS5 and Xbox Series X
    • Moving to URP or HDRP is even worse as Enlighten support is already deprecated
    • Realtime raytracing is pre-alpha and not scalable
    • Waiting for the new stuff to arrive means waiting until 2021
    • Switching to a new engine is not something you do based on one unfortunate event
    I'm just kinda pissed at Unity for their lack of consideration toward their userbase... but did my griefing and just removed dynamic day & night cycles from my game.
     
    Salvador-Romero likes this.
  45. Bordeaux_Fox

    Bordeaux_Fox

    Joined:
    Nov 14, 2018
    Posts:
    589
    When I speak about Realtime GI, I target to the consoles an pc's. I mean they are the platforms where most of the games would shine with dynamic lighting, like open worlds, action, horror games, etc.
    If you target to mobile or Switch you mostly do not get that complex and why would you want a HD dynamic light game for low hardware? It is simply not possible without a baking and faking approach. And mostly you cut down even more. No Android or Switch gamer cares about indirect bounce anyway. That are not the target group with high demanding graphics. Of course you have pretty titles like racing games, but they just bake their maps and it looks good.

    Coding everything on your own is fair but also makes the usage of a REALTIME engine pretty laughable. Realtime GI is a part of a REALTIME engine that has to provided. If not then the engine just falling back. You could always argue the licensing cost, but at the end you have to remember the features you are using had to be developed by people who needs to get payed. If you do not like it, then stick with Godot.

    And switching to Unreal is maybe not a thing for just one generalist working alone on a title who just knows C#, but definetly an option for a team with their own programming department. Despite the fact that you also have other engines that using C#.
     
    Last edited: Mar 9, 2020
  46. alexandre-fiset

    alexandre-fiset

    Joined:
    Mar 19, 2012
    Posts:
    717
    Not really true. We had Dynamic GI in our game as it really helped switching lights on and off dynamically. The Switch has a memory "problem", but not really a rendering one. You can definitely dynamically light the interior of a spaceship or house on Switch if you are smart about how you use GPU time.
     
    EricWilliams and Salvador-Romero like this.
  47. Bordeaux_Fox

    Bordeaux_Fox

    Joined:
    Nov 14, 2018
    Posts:
    589
    If you talking about Enlighten, I think it is more CPU bound then GPU. That's why you have the Realtime CPU Usage option in HDRP asset. Nevertheless I would like to see an interior scene without light leaking.
     
  48. OCASM

    OCASM

    Joined:
    Jan 12, 2011
    Posts:
    329
    "Low reward endeavour". If that's the case then it's clearly not very important. IBL like CryEngine should be enough for most cases.

    To be honest my impression is that most users don't care about the Enlighten removal.
     
  49. neoshaman

    neoshaman

    Joined:
    Feb 11, 2011
    Posts:
    6,493
    EricWilliams and OCASM like this.
  50. Salvador-Romero

    Salvador-Romero

    Joined:
    Jan 21, 2014
    Posts:
    54
    Hi,
    I just want to say that I differ on that. I develop for mobile with dynamic lightning (as an important feature), and so far I managed to get good results. It is maybe harder and you have to be creative, but doable.

    At least until they drop Enlighten, of course (o:

    Cheers!
     
Thread Status:
Not open for further replies.