I was sad to hear about the closure of Lionhead Studios a couple weeks ago. Not only because game studios closing is always a sad affair, but also because I had high hopes for Light Propagation Volumes. As far as I understand, Lionhead were the only ones working on LPVs, so I am now wondering if there is any future for this technology or if I should drop it from my project.
I really liked the results from LPVs, but if there is no guarantee that they will work in future versions of the engine, it seems quite risky to rely on them. Not to mention that the current implementation has quite a few bugs and limitations that I was hoping would disappear with future updates.
It would be very helpful to know if there there any plans within Epic to pick up development of LPVs or other real-time global illumination solutions that work on PS4 and Xbox One.
There was nothing coming from Lionhead, what they had added was already integrated.
There is no future for LPV, it doesn’t work that great and there are other methods that are more promising. Most likely there won’t be a truly great dynamic GI solution for consoles. Once PC’s get more powerful the current solutions that are too slow will be able to perform well enough.
The thing is, I don’t necessarily need a “great” solution. I just need any solution at all for open world dynamic lighting on PS4 and Xbox One. If we had access to SVOGI or if there were a way to occlude sky lights, that might already be enough for me. But it doesn’t appear to be possible.
Could you elaborate on that? Do you mean there were no new updates from Lionhead before they shut down?
I agree that there’s potential, even if it’s limited and only applies to some projects. But unfortunately I’m not able to make any modifications to the current version of LPVs, nor is anyone else in my team. Unless Epic or someone else is still developing it, I won’t be able to use them. (Especially since there appears to be a new LPV bug in 4.11.)
I see, so all recent changes were made by Epic and/or the community then? That would be good news. There have been changes after all, even in 4.11, so someone must still be working on it.
LPV in Cryengine has similar limitations as it does in UE4, I think the only significant difference is that it probably has cascades which allow lower quality in the distance so that you can cover large areas.
I believe support for lighting from emissive materials was added only recently as well.
Either way, it’s interesting to hear that LPVs in the CryEngine already support cascades. The lack thereof in UE4 is by far the biggest headache with my current project, since the LPV simply cannot cover the whole visible area without a severe cut to quality.
Can anyone comment on how hard LPV cascades would be to implement and whether there’s any chance we might see it in a future update?
I think emissive lighting using LPV was in there since the beginning, if not then it was added not long after release. I don’t know for sure that Cryengine supports cascades, but I’m guessing it must since they use LPV and they have large levels.
I’ve used LPV with CryEngine extensively and the implementation was no different than that in UE4. The range was very limited but if you tweaked it correctly it would barely be noticeable things were updating as you moved through a level. It’s old tech ofc and for old tech it still has its uses. Speed for one. Now with the new VR hype, that argument is very much valid again. Quality wise it doesn’t hold a candle compared to SVOGI/VXGI (doh). I read on the CryEngine forums, they will soon be experimenting with volumetric lightmaps to facilitate partial pre-calculations for SVOGI/TI. Here things have died on the dynamic lighting front after the surfel GI experiment. I hope it’s just silent and things are very much cooking out of sight.
This is the part I’m struggeling with the most. No matter how I tweak the LPV settings, it’s always extremely noticeable where the volume ends. Of course making the levels smaller or obstructing the view of the player would fix the problem, but unfortunately that’s not an option for me.
According to these two papers, there seem to be LPV cascades in CryEngine 3, but I’m not entirely sure if that’s the “official” version of the engine or something else.
Either way, this is precisely what I would love to see in UE4. Maybe I should try finding a programmer who’s willing to look into it, but I have no clue how complex that task would be and how much it would cost.
Hmm, now I’m not sure either anymore There was a bunch of stuff related to LPV’s in CryEngine that never worked either but overall the core worked fine. So yeah lets assume it had cascades. There’s absolutely nothing you can do in UE4 to soften the transition at least to acceptable levels? Good examples in those images. Can’t believe that’s just an oversight on Epic’ behalf. It’s way too noticeable.
LPV lighting used to fade in more softly, but ever since UE4.9 there’s always a hard transition. As far as I can recall, fading out has always been a hard transition though, so it has always been an issue. Plus, even with the “soft” fading, it’s very noticeable in extreme cases. The cubes in the background of the left video illustrate this quite well:
I fear real-time GI simply isn’t a priority for Epic right now, so these things don’t get a whole lot of attention. Which is unfortunate, because I still think that LPVs can produce really beautiful results at low cost. There are just a couple of kinks that need to be ironed out…
You could try VXGI. I know it has a reputation to drain your fps quite rapidly and in part this is true. However, if you are content with something on a LPV-Quality level, I reckon VXGI on very low settings will be enough already. This performance hit should be more or less equal.
Yeah, I was planning to look into VXGI if LPVs don’t work out. I was told that it doesn’t work on consoles at all though, which would be a deal breaker. However, if it’s only a matter of performance and can made to work by lowering the quality, that would be fine.
Oh ****. That’s a bummer for sure then. Well, guess here goes praying some dev has a good day and decides to have a go at those cascades then :rolleyes:
There was a bunch of stuff related to LPV’s in CryEngine that never worked either but overall the core worked fine. So yeah lets assume it had cascades. There’s absolutely nothing you can do in UE4 to soften the transition at least to acceptable levels? Good examples in those images. Can’t believe that’s just an oversight on Epic’ behalf. It’s way too noticeable.