Thanks I greatly appreciate the explanation. Just one last question: Is the pre-exposure cached range locked to 32 ev100 stops?
Just wondering if there may be any benefits (memory/performance/precision?) to setting it lower if I knew for example that I was only ever going to need a range of 8 stops of EV100.
Cached lighting range is locked to 32 stops, because that’s the range of a 16 bit float (Float16). It doesn’t matter for quality or performance if you use only a part of that range. You can read a bit more about it here:
I wonder whether that is caused by materials on the opaque object not rendered correctly in hit-lighting. When you enable r.RayTracedTranslucency.UseRayTracedRefraction, opaque hits after a ray is refracted are lit using hit-lighting instead of the regular GBuffer lighting pipeline. So if the material look different in hit-lighting, it can cause opaque pixels behind refractive objects to look different.
Lumen has several subsystems that trace rays but those rays won’t intersect translucent objects. Lumen reflection rays can intersect translucent objects when certain settings are enabled but GI related rays should always ignore translucent objects. So the situation where indirect lighting is blocked by the glass shouldn’t happen in theory. You can check how the surface cache of the opaque object looks like using “r.Lumen.Visualize 1”. The right most tile will visualize surface cache. You can toggle visibility on the glass jar. The opaque object should look the same in that tile whether the translucent jar is visible or not and it shouldn’t be yellow or pink in the visualization. If you set r.Lumen.HardwareRayTracing.LightingMode to 1, you can see in the middle tile how the object looks like when lit by hit-lighting. The caption of the middle tile should change from “Reflection View, HWRT” to “Reflection View, HWRT, Hit Lighting” when you set the cvar to 1.
I am using Ray Traced Lumen with Megalights, and the lighting mode is set to Hit-Lighting.
I am also using the following cvars for my cinematic scene:
r.RayTracing.Nanite.Mode=1;
r.RayTracing.Culling=0;
I thought the surface cache is only important for software-based lighting and not important for ray-traced lighting.
I already fixed the pink mesh by increasing the lumen mesh cards or by separating the complex mesh into small parts. But I can’t fix the yellow meshes, there are lot of meshes that are small showing with yellow color in my scene. Unable to fix yellow even with increasing scene detail from 10 to 1000. if yellow mesh is the reason for dark GI with ray traced reflection, then how can i correct the issue so it don’t cull these small meshes from ray tracing?
Here are the view modes:
I also tested by increasing the size of mesh from build setting using Build Scale. but it still yellow in surface cache
I also tested the material,
the current glass material uses Translucent Blend Mode with Thin Translucent Shading model. and lighting mode at surface forward shading. This gives the result, which you can see in the reply above.
and this is the result with changing the material to Default lit and blend mode to masked. basically changing it to default lit shading model changes nothing in lumen view mode’s reflection view. it only change the result when i change the blend mode from tranlucent to masked and set opacity mask value to 0. you can see result below.
am i seeing these issue because ray traced refraction is still under development?
well… it is a raytraced refraction issue. no clue where where those rays shooting. legacy translucency works and delivers clean refraction bounces thru glass. the reflection is kinda weird and there’s obvi no gi in the bottle but you could bake an ao map or bake it into a vertexcolor channel for this prop and be done with that.
I hope they fix it in 5.7, for the time being, I can use raster refraction with ray-traced translucency.
Hello I’m having an issue with lumen and Light functions in directional light. Somehow the directional light leaks a tiny bit through interiors and generate some artifacts. Does anyone have encountered the same issue? it happened in 5.5 and 5.6
I am using Unreal Engine 5.6 and trying the new Ray Traced Translucency. I am using Lumen for lighting and Hardware ray tracing. High-quality translucency reflections are also enabled, and all bounce settings are maxed out in the post-process volume, along with Lumen GI and reflection settings. Order Independant Translucency is also enabled from project settings and sort policy is set on “Sort By Distance“. and all of these objects are separate meshes.
I am having an issue with new Ray Traced Transparency. Reflections are getting absorbed with light, and even without light, they’re barely visible. seems like a bug. but if i am missing some CVAR for new ray traced Translucency, please let me know.
In Case this a Bug, then for the time being I will need to stay with raster, but i am also having an issue with raster translucency. object in rooms looks fine with translucency. But objects which are inside cabin behind the glass (another translucent object) show a flat black color instead of reflections on extreme angles (in this image it’s tweezers back top face). But as soon as I pull these objects out of the cabin and there is no glass front of these translucent object, the side angle will start showing reflections. seems like Raster Transparency also struggles with GI and reflections when placed behind another translucent object. Please let me know if there is a solution for this problem as well.
same problem as your bottle. currently the last raytrace hit shader for glass is too dark. hence why it doesn’t look right.
just for the heck of it i put a light into a raytraced pill bottle. i removed the interior faces of the bottle. it just halfs the refraction bounces without further impact and does not do the double refraction that would naturally occur on a “thick” (double layered) surface. the trace seems good, just the last hit is too dark.
that’s without refraction and good looking translucent traces.
maybe you should build a testcase to figure this out. this is all full but stock settings, btw.
issue resolves as soon as i switch to substrate. but raytraced refraction are same in the substrate causing issues.
right now, the values that I am using are:
Max. Primary Hit Events = 64;
Max. Secondary Hit Events = 64;
It only happens with ray-traced refraction. Only primary hit events are causing the problem; anything higher than 1 starts to darken the glass as I increase the primary hit events value.
yes. i see that too. the first hit event is the glass itself and the refraction. it’s entering the raytracing scene. everything after that it bounces around in the raytracing “scene” or exits out the back. when it exits it, the shading is wrong aka too dark. you can clearly see that in my test aswell. it comes out the back between the 2 other bottles where it is dark. i have no fix for that, currently. i gotta dig thru shaders? or we wait for epic to find a way to correct that. : )
I hope they fix it soon I am also going to post it into substrate thread
Bug report:
If level has landscape with grass type. Then unreal engine crashes as soon as I enter this CVAR “r.RayTracing.Geometry.Landscape.GrassType=1”.
follow up on the translucency glitch. i think it might be lumen related, indeed. i “accidentally tested” this setup (checking some other forum post). it’s baked gi. stationary lamps for megalight shadows and stained glass. it goes thru the ball. into the mirror and out the window without errors. refracted colors are “accurate” and the translucency works correctly thru the whole trace.
i love the semi caustics , btw. only works with raytraced lamps tho, no mega shadows, yet. : )
(edit)
i might aswell throw this one in. just benched it with everything in there. lumen on. nanite meshes. raytraced translucency. it hit the glass ball, the glass pane, the mirror and the window. everything is shot and stained but i still got a flawless green box trace. also no occlusion errors behind the glass pane. hmm…
so… i dunno what sadiq did with his scene that it doesn’t work. maybe the glass mesh gotta be seperated somehow.
Refractions issues is lumen related but reflections issue was shading system related, it works fine with substrate; it was showing an issue in my scene with the legacy material system because i was still on legacy material, but as soon as I enabled the new substrate material system, it worked fine. The ray-traced reflection started working without any issue. but the refraction issue is still there with the substrate.
but when i compare the reflections with path tracer i think ray traced still looks little bit dull and there is still room for improvement. other than that reflections are working as long as you are on substrate material system.
okay. i don’t use substrate. (doesn’t run too well on my craptop). the “legacy” material is still working just fine, as tested right there. obviously reflections are not visible cause the material is transparent without the “opacity boost”, but… the refraction is kinda “flawless” in this scene.
i wouldn’t call raytracing legacy, btw. we just getting started to push some broad strokes.
might be a setup issue with your pills and the shelf, then or… i dunno, tbh.
The issue happens with transparent glass not mirror and you need a light source behind it to reproduce the issue. as you can see in my example cabin shelf is a closed space. and it has a light source behind. if disable the light behind glass, ray traced reflections will appear again and if I move out the small objects inside cabin the reflections and GI will improve even i have light source in room but seems like translucent object behind translucent object also cause issue.
Regarding pills and Gi: did you have every pill as a separate object? Because all pills stacked in a bottle are a combined mesh in my scene.
I am having few issues with open scene:
BUG: landscape spline actor breaks and have gaps in the lumen scene view while looks fine in lit mode geometry wise but lighting will be incorrect. I have Nanite enabled on the mesh that i am using in landscape spline actor. But issue resolves when i disable nanite on source mesh.
