I hope to the Lord Jesus that it comes out in September; I can’t wait till October!
Same , I am to excited already
I remember trying to build the lighting in the landscape demo and failing twice. I have 16GB of RAM, an i5 4670k and a 780ti. I should be able to build the lighting for a 5k x 5k map.
Improved Mesh Import Please!!!
If 4.5 will let me concentrate on modelling instead of bloody second set of UVs, I will be so, so, so bloody happy
The improved code hot reload is worth it on its own for me.
Yeah i have to say seems to be the first example of Epic misleading Devs, We all remember the fan fair release of the Epic terrain example (that i should also add had nothing in it other than the terrain a few rocks and a minimal amounts of simple low poly tree’s) but yet they gave no information about how much memory was needed to build such a simple example with lightmass. Which after speaking with others seems that if you have 26GB memory your JUST get that example lighting baked. THAT IS UNACCEPTABLE. Clearly lightmass is not usable for even small terrains when adding other baseline geometry for any game type, All that needs to happen is for lightmass to be able to break terrains into grid sections with each section guard banded by maybe 5-10% to make sure lighting is continuous between section processed separately.
I dont care if it takes me 10 hours to bake a terrain if need be, all i care about is actualy being able to do that, even with only 4-8 GB meomory. is the biggest fail of Epic so far since UE4’s release, and a pretty big one at that. Needs immediate resolution before the word gets out UE4 can’t even build static lighting for a reasonably small area, let alone it’s realtime lighting. #Disapointed
[;142842]
Yeah i have to say seems to be the first example of Epic misleading Devs, We all remember the fan fair release of the Epic terrain example (that i should also add had nothing in it other than the terrain a few rocks and a minimal amounts of simple low poly tree’s) but yet they gave no information about how much memory was needed to build such a simple example with lightmass. Which after speaking with others seems that if you have 26GB memory your JUST get that example lighting baked. THAT IS UNACCEPTABLE. Clearly lightmass is not usable for even small terrains when adding other baseline geometry for any game type, All that needs to happen is for lightmass to be able to break terrains into grid sections with each section guard banded by maybe 5-10% to make sure lighting is continuous between section processed separately.
I dont care if it takes me 10 hours to bake a terrain if need be, all i care about is actualy being able to do that, even with only 4-8 GB meomory. is the biggest fail of Epic so far since UE4’s release, and a pretty big one at that. Needs immediate resolution before the word gets out UE4 can’t even build static lighting for a reasonably small area, let alone it’s realtime lighting. #Disapointed
[/QUOTE]
I agree, it seems extremely amateurish especially for a company like Epic. What developer thought it would be acceptable to force people to buy 100GB of RAM to light a simple map and how could code like that have made it into the Unreal Engine baseline? I am trying to build a simple level with 1 castle in it and lightmass is consuming all RAM in sight before it crashes. I only have 8 GB, but I have no idea how much I would need for such a simple example. Why on earth would it in multiple GBs of RAM? needs to be fixed ASAP and there are bug reports from at least April. Is it as least on your radar for being fixed? What alternative do we have other than wait for it to be fixed? Is there another way to light a level without lightmass?
I just use Dynamic Lighting with LPV + DFAO - Never have to bake a lightmap for anything, and works great for me in large scenes (I have a gtx 660 ti currently), even the more demanding scenes (LandscapeMountains demo) I get 40+ fps. usually 60. The performance hit of Dynamic lights in an outdoor environment is very minimal, plus you don’t need to rebuild, personally I will be using dynamic for all outdoor scenes in my game.
+1
I’m not against the suggestion that it would be nice to be able to build lighting regardless of your RAM, but i wouldnt use static lighting for a world larger than 2K. Even a 4K map filled with buildings(and also interiors of some of those buildings), foliage, rocks, etc. can be more performance heavy than dynamic lighting for the end user, considering the RAM they’ll need for those lightmaps.
So for large outdoor levels is it best to use an unbounded lighting volume and dynamic lighting for everything?
I’m excited as well! As a clear-cut amateur, I’m hoping to find it even easier to learn than it already is. My biggest hope is for improvements on landscape, so they don’t have to rely on an external program that will cost me more money…
but perhaps that’ll come at UE 5.5 
You can build landscapes directly in UE4, and I personally don’t bother with the World Machine ****. Yea it can produce some good results, but the cost and time it takes to learn isn’t worth it for me. Also, I have never had a problem with lightmass crashing on large terrains, though I don’t think I ever created a terrain as large as what is being described, and I always use a lightmass importance volume around my playspace (which is only a portion of the landscape). I have 8GB of RAM and an i5 2400 btw.
Regardless of your skill level with WM or your level size, Lightmass should not need nearly THIRTY GB of RAM to build a level of that size. What’s really sad is that the level doesn’t have **** going on in it. It’s literally just a few rocks, very low poly trees, a few bridges and a frozen lake. God forbid someone tries to make a decent nature scene.
I understand that LPV/DFAO is an option to consider for larger scenes, but as it stands, it ISN’T an option; it’s the only way to have lighting in large scenes. There are benefits to static and dynamic lighting, but one of them is not viable whatsoever for large scenes. I’ve heard many times from cynical devs that UE is only good for indoor/small levels…and right now that seems to be pretty accurate. The landscape painting setup is beyond obtuse and lightmass is just…not good. I generally don’t like to rant about engine problems (moved from Cryengine to UE, so God knows I’ve experienced problems with CE), but needing that much RAM to build lighting is bull****.
Yeah, looking forward to real time lighting improvements. Not using lightmass at all… don’t wanna wait hours for compiling shaders, and we have huge outdoor areas
Thing you never hear a women say: Shopping is overrated
Thing you never hear a guy say: I enjoy trying out new pants
Thing you never hear a developer say: I enjoy making lightmaps.
On a serious note, has anyone experiencing, with both lightmass, & full Dynamic, & how much quality difference in an indoor scene?
We have a gigantic house in our game that plays a huge role, so we’ve experienced both lighting systems with an indoor scene. I noticed that things that have extreme specularity (sinks, fridge, etc) go haywire with how much light they reflect, but that’s about it. Have you experienced a lot of problems?
Ack! Apologies, I didn’t realize the requirements for that content had hit something unreasonable. Let me discuss with our devs what we can do to improve . Our dev machines generally have 32gb of RAM which means unfortunately means cases like often don’t surface as often during our internal development process. To help catch more often internally we’re expanding the diversity of our PC configurations to better represent what active developers are using with the engine. We’ll make it right, sorry to have caused angst and left unaddressed for quite some time!
I am looking forward to UE5.0.
I’m really looking forward to html5 and mobile improvements - mainly compatibility in the mobile area (WP8 sometimes maybe?) and performance improvements, build package size optimisation, ability to strip unwanted/unneeded features, html5 support is currently the worst out of all supported platforms, the engine’s js source is just too big to being useful. Yes i know, I can only dream! 
It’s going to happen, but in a more distant future, until then I’m waiting patiently for 4.5.
Thank you! I’m sure most of us having issues with it would be happy knowing it is on your radar and will be addressed in a future version even if we can’t have it today. Hopefully the devs will agree it is a problem that doesn’t scale very well and can do something like idea of using a grid pattern or something to process sections at a time. Admittedly I don’t know what it is doing behind the scenes, but I would think consuming 1GB should be plenty for just calculating lighting.