How come when I get out of FIB3 render radius it doesn’t change to voxel and it doesn’t emit any lights. It never acted like this when I was making this. robloxapp-20200920-1837048.wmv (1.6 MB)
Also it won’t even emmit when its changed to Voxel or shadowmap and it does this when im closer to the ground and I go away from it. Happens with Voxel and Shadowmap.robloxapp-20200920-1843140.wmv (1.8 MB)
Compared to shadow map with 257 point lights with shadows with the baseplate at smooth plastic to see specular more easily with 162 pastel blue metal balls there’s around a 62% performance decrease! But its not 100% accurate but should give a rough idea on how big the performance impact it is. Heres how I did the stress test.
I used roblox Summary fps counter and looked at fps. I used my own eyes to see what the lowest it went and what it usally went for and caculated the mean average. For example imagine I was getting 51-52 fps and a lowest of 41 fps with fib3 I would caculate the mean average of those 3 numbers. Also if I saw a .(1,2etc) with the fps I would ignore that. I set it to run for 30 seconds on each fib3 and shadowmap. I would also start determining the fps as soon as the balls hit each other. if my mean average was 26.33333etc I would round it down to 26.32. Specular and diffuse scale was off the range of the lights was at 20 and a brightness of 3 and the colour was white. And ambient and outdoor ambient was at zero with a shadow softness of .2 and a clock time of 14. Also I got the percentages by taking the shadowmap mean average and kept subtracting the percentage until I got close enough to the fps of fib3. Anyways lets get to the results.
With shadows.
FIB3 Mean average: 10.32
Shadowmap:27
That’s around a 62% difference comparing FIB3 with shadowmap.
Without shadows from pointlights
19.5
(Also I only caculated fib3 cause theres basicly no diffrence with shadows on and off with shadowmap)
That’s around a 29% difference comparing FIB3 with shadowmap. (Also the numbers I only got was 19 and 20fps)
Also I did not do a static stress test which is no balls or anything underneath it and nothing moves cause roblox caps it at 60fps which could make the results with shadowmap very inaccurate cause i’m sure it can go higher then 60fps but roblox caps it at 60.
You can download the place here to do your own stress test to see if you can get more accurate results. (Also the only application I had opened was roblox studio.)
StressTestFIB3.rbxl (25.2 KB)
The camera position should be what I was seeing and make sure your graphics setting is the highest it can be. Also the point lights have shadows disabled automatically so you gotta turn it back on.
Performance should defiantly be improved heavily. If a update rolls out with performance updates ill defiantly be checking using this little simple stress test I made to see how much has improved. Also again this is not 100% accurate this just gives a rough idea on the performance impact I would highly recommend doing it yourself and posting your results by replying to this or making a better stress test and general then the one I made. If someone wanted to do a more accurate result of this test Ideally it would be best if someone made a script that found the average fps of the first 30 seconds instead of relying on your own eyes to get a more accurate result and use a unlocked fps and reply to this with the results for a more accurate result.
3 Likes
Seconding this 100%. My game is already costly as far as performance goes, even utilizing workarounds like content streaming. I was excited to preview FIB3 in studio, but upon play-testing, the framerate is hardly bearable (even on my gaming laptop). I already sacrificed the possibility of having a mobile playerbase for my game’s quality, I don’t want to have to nix any and all lower-end devices on top of that to use this feature.
4 Likes
Will there be a way to change Fps cap (i want 120 fps for my 60hz display its so i can have up to date info as much as possible without vsync.)with FIB because well isnt this a engine upgrade?
1 Like
Very cool.
I’m really looking forward to adding fbi3 to my game.
3 Likes
ETA for a release if possible? Really looking forward to this.
2 Likes
Will there ever be a high dpi release? Because these graphics look great and all, but I’m running on a 4K display, and nothing on roblox client and studio looks sharp at all. It rather looks blurry and pixellated.
I remember seeing a post back in 2018 talking about this, but some staff responded about it requiring a newer version of Qt, and were not ready to drop older versions. Since roblox dropped support of many old OSes since then, are there any plans for adding 4K support in the near future?
3 Likes
Already mentioned in the topic.
That’s so cool! Keep up the good work! 
I think I’ve found a fix for the no-shadowed surface lights!
Here's how I did it.
I just stacked a lot of parts together, each with its own spotlight attached. The picture is the result.
Not only is it breathtaking, but FPS taking too.
5 Likes
If only roblox could simulate light bouncing…
3 Likes
Also, I heard they are looking to add global illumination. Well they tried but it’s really complicated and performance can be an issue. It’s said in this article written by zeux who is (correct me if I’m wrong) a technical director at ROBLOX. Future Is Bright - Comparison | future-is-bright
If you want just scroll all the way down and scroll up a bit to the part when he talks about global illumination.
3 Likes
The lighting is actually much of an improvement, even if it’s just for a beta. Here’s a preview of my game with Future lighting, both in 60,60,60 and 0,0,0 ambient settings.
Ambient set to 60,60,60
Ambient set to 0,0,0
5 Likes
Honestly, I can’t wait until this fully releases! However, why can’t it release right now? Everything is perfect!
If you're worried about removed features.
If you're worried about surface light shadows.
If you're worried about the light radius.
If you're worried about the autoexposure
The autoexposure from..."
ROBLOX: Next-Generation Voxel Lighting
Performance and bugs. Performance is really bad right now I even made a post here about it. It just gives a rough idea on the performance impact its not exact (if my tests were 100% correct it would be a 62% performance impact compared to shadow map). Also fib3 is fairly buggy, weird shadow artifacts at range 60, and my shadows dance, soooo that’s a issue. Also light leaking. Also in metal environments its really dark in fib3 and you got to get it stupidly bright in the settings but ill look way too bright on voxel (basically blinding while fib3 looks cinematic). Maybe if it they forced fib3 on all settings with (just not sun shadows) and not change to voxel all our struggles with it will be eliminated, but since its not optimized enough we got to live with it. I don’t think they can edit voxel to not give an unfair advantage, the voxels are way too big. And I don’t wanna struggle for lower settings to get an unfair advantage due to it being way too bright and cause it’s not optimized enough. I wanna use it really badly but I don’t really think I can due to voxel looking way to bright and fib3 look way to dark, the way you they can fix it is by making fib3 more unrealistic to fit voxel lighting but no one wants that. And if they won’t get rid of voxel entirely (only for artificial lights) when the default is fib3 I think barley anyone would use it, just basically showcases would use it.
Can you change it so when the shadows move it doesn’t look so shaky. Also I don’t have it locked on a grid when moving the ball.robloxapp-20201001-1924245.wmv (1.7 MB)
I think ill make it look a bit more smoother and better.
Uhh this is a bit strange.
The performance may be bad, but at a good reason. Also, the light radius will be improved later on after FiB 3 is released.
And,
Polls showing if people like the new lighting
1 Like
True but first, I’m pretty sure performance on other game engines are way better without even having to change it to a vastly different and way worse lighting system. Second of all those could of been used in showcases etc. But you got a good point. It’s just I really can see developers struggling making it look good on voxel when sometimes its stupidly difficult and almost impossible. No one should be struggling just so it looks good on two vastly different lighting systems and not give an unfair advantage. We should just place our lights and know there won’t be an unfair advantage on lower settings or be blinding. This is sorta why I see why it won’t be used that much due to those issues. Especially with the terrible performance currently. Right now it doesn’t seem that scalable and have good performance without needing to swap it with the voxel system, and that really should be fixed before release. The sun shadows are fine to get rid of on lower quality or distance, but it wont work the same with artificial lights.