How does dfgi perform in indoor scene?
The current GI solutions (DFGI and HFGI) are not suited for indoor scenes, because they are only one-bounce GI. If you are talking only indoor lights, I would recommend lightmass; the quality will be unmatched.
I was thinking about light leaking.
Outdoor and indoor combined.I don’t know how to explain it 
Read through the older posts on this thread then. I’m working on an indoor/outdoor combo as well, and I asked a lot of questions and got a lot of answers.
master branch 4.9
a lot of grid artifacts
I had no problem with DFAO + GI till 4.7.6. Everything was looking as it was supposed to be. When I switched to 4.8.1 with DFAO + HFGI as mentioned here, I am getting strange blobby AO on my grass meshes at certain distance. Looks like something is not working right. That blobby AO looks like it is camera facing. Near grass patch are not affected that much but far patches are looking real bad and it’s killing my eyes. I also have some house meshes that are having strange artificats on cieling plus side leaks. So all of these issues are totally ruining the experience.
I haven’t tested in the office yet but at home it’s really depressing to see these problems suddenly pop-up. I will put the images of my problem as soon as I get home. Also there is some strange thing going on with Atmospheric fog that I need to inspect it before jumping to a conclusion.
Here are the settings that I use for my outdoor scene. Feel free to correct me if I am doing anything wrong. As I have taken a backup of my project which was in 4.7.6, So I am on safe side but overall first experience with 4.8.1 not good at all.
r.AOMaxViewDistance=60000
r.AOViewFadeDistanceScale=.3
; Enable heightfield GI
r.DistanceFieldGI=1
; Disable surfel GI
r.VPLMeshGlobalIllumination=0
Best Regards
That is concerning as that is the new DFAO method which is much faster. What graphics card are you using? Any repro steps? Was that play in viewport or in game?
Make sure the grass meshes are not in the distance field. There are various hardcoded limits on how many objects can be handled in a certain region in space. Grass tends to blow these limits.
Also make sure the grass normals face mostly up. If they face down or sideways, they can pick up the occlusion of the landscape. Have a look at the Kite demo grass.
So I have some bad news.
Priorities have changed here at Epic and I won’t be able to work on dynamic GI anymore for the next 6 months or more. That means DFGI won’t be happening. Maybe I can still get Heightfield GI into a good shape performance wise as that’s a much smaller task.
Just wanted to set your expectations. It is a sad day.
ooh thats indeed bad news 
Could you tell us what the new priorities are, so on what you will work instead of DFGI? Just the fact that you can’t work on GI of course is bad news, but if you instead work on another great feature then at least I can get over it a lot easier :rolleyes: For example, if you at Epic decided to put more work in existing features and polish them instead of adding great new ones like GI, then this might not be totally bad news 
That is bad news. If you can would you mind sharing with us what the new plan is?
I just wanted to echo the sentiment expressed above. I would have a much easier time letting go of DFGI if I know your resources are going into an equally good feature.
Thanks for the warning though, you didn’t have to do that, but you did. Shows how much you care.
oh no…http://s626.photobucket.com/user/Quickrick/media/24f9rmh.gif.html. Well thank you for informing us.
Aww, man, so very sorry to hear that. But thanks ever so much for your work on DFGI! (it seriously saved my thesis project bacon!)
I’m sorry to hear that. 
Maybe Epic discovered a new way to create Global illumination 
AHR looks like it could be a good method. I was hoping to see where both of these go, but if this one can’t go through, all of the community effort can go to AHR to help push dynamic GI
I agree with you 
That’s too bad, indeed. I hope you will eventually get the chance to continue your efforts on DFGI! I’m also very curios to know what it is you are working on now.
.
Wow, this is the saddest day of my UE4 experience…
I guess I should my game more static then without thinking about bright dynamic future :c
Hey guys, thanks for your messages. The new focus for me is performance, especially on current gen consoles and mid spec PC. This means squeezing the last 20-30% out of many rendering systems.
Interesting… Thanks for the info. So we can expect 4.10 to have some nice performance improvements