[UE4} Medieval Interior
polycounter lvl 10
Hey all,
Spent a few weeks working on this in UE4. Still needs a bit of work with the lighting, atmosphere, and god rays. Was hoping to get some feedback or crits so i can wrap it up. Thanks!
Spent a few weeks working on this in UE4. Still needs a bit of work with the lighting, atmosphere, and god rays. Was hoping to get some feedback or crits so i can wrap it up. Thanks!
Replies
Yeah i was going for really grungy, but it could be a number of things such as detail textures or even the anti aliasing as well.
Overall this looks killer. If I had to nitpick:
The reflection on the metal rims of your barrels seems a little strong/smooth. Makes them look significantly cleaner and newer than everything else in here.
I agree with PogoP a little bit on the noisyness. I think going for grungy is cool, but like PogoP says...might be with your diffuse/albedo. Maybe an example is how the water stains/streaks on your walls are really visible in areas with no light, like directly above the windows near the ceiling.
I can tell from these shots that there is a specific direction the player is supposed to go..especially in the first shot. Might be kinda cool to lessen the amount of cracked concrete and rock chunks there are in the the path leading across the screen. Maybe clean that part up just a tad to really guide the eye across the path.
Again, nitpicky things. Good stuff.
A lot of this lies in how much you are trying to push in the diffuse; in parts it's so overwhelming you can't quite read what the normals are doing. I'd pull back some of the detail in the diffuse, maybe even start with solid colours and put more detail in the roughness and let the normals do more heavy lifting.
I think what is working exceptionally well and what really works well on initial viewing (squint test) is the composition of the shots and the colour scheme (I love the blueish fog receding in the background).
Really great job so far!
It looks like a few of your lights, specifically the torches don't cast shadows. The lack of contact shadows is hurting the readability of shapes.
PS: Press F9 for a screenshot without the beta number/name (also you can configure a shortcut in the preferences to run the High Resolution Screenshot utility)
What i think is happening here honestly is just the power of how UE4 handles physically based materials and reflection environments, which means the diffuse itself has very little to no lighting information (such as pre baked AO or specular highlights) and the deferred renderer makes seams less noticeable because of this since it is primarly color information and the reflection environment that is blending.
I noticed this on other objects as well with seams, they are only really noticeable if the cross border seams are very different and even than it doesn't stand out as much as i would have expected.
With that mesh i tried my best to match the bricks, so its probably a combination of PBR and the procedural wet drops i have that accumulate on Z faces of the normal and meshes.
So yeah, no fancy procedural blending here, just new features of UE4 making things easier.
I agree with the above on the noisy textures. It feels almost like it's so much you can't really enjoy one part of the images, instead everything is fighting for your attention.
1. Did you manage to over-ride UE4's "metalness" shader workflow? (ie, did you work with a full colour spec, or the metalness mask?)
2. If you did use the metalness workflow, how did you find it on a large scale environment?
You can't. Moving to deferred pipeline entirely, effectively cut possibility of using custom lighting, which requires forward rendering path.
MY EYES THEY HAVE NO REST
I have to agree with what people are saying, there isn't quite a focal point, everything pops up to the eye. But hey, if this is meant to be played, not just to be looked exactly as it is, I guess it could work!
Brilliant work, seems like it may have given a ton of headaches!
Are you working on this alone?
Energy preserving, physically based deferred. They've even stated "forget your custom lighting art direction, all physically based!"
Which, I mean does prevent some things. But mostly it's just extra work for little apparent affect. At least that's my own opinion.
Btw how did you get hold of ue 4?
As for the focal point thing, I have to admit this is something I overlooked. The primary reason being entire level is meant for gameplay (and cinematic) so I built the lighting scheme based on what I thought looked best on enemy placement and gameplay flow more so than portfolio work.
As for how I got UE4, like I mentioned this environment is for a actual game being worked on by a team and is not for portfolio work.
I'll post a gameplay vid in the near future to help with feedback. Thanks again!
Okay i understand that, but there really is no reason not to allow full specular ranges as an option, the math change is so simple it's actually scary that it isn't already in.
the line of math would currently look like:
what it's doing here, for the benefit of those who don't know: (source here is the UE4 tech paper, i'll grab a link later) is giving any black areas in the metalness mask a specular value of 0.02, and any white areas a specular value of whatever is in the albed. it's then energy conserving the diffuse (albedo * (1 - specular)) and making the diffuse black wherever there is metal present.
to change it is incredibly simple, allow a coloured specular input, and then change the code to:
I guess what i'm trying to say here, is that the excuse of "everything is deferred, no custom lighting" doesn't hold any water in this particular argument. sure, maybe i should have rephrased my question before and asked why there isn't coloured specular available as an option instead of custom lighting. i understand why there isn't any access to lights in a deferred pipeline. i don't understand why there is no option for fully coloured specular. it basically feels like epic is saying "this is your workflow, no deviance". and that will hurt you in the long run.
your opinion is wrong. the metalness workflow is actually less work to do, but it also gives less choice for artistic direction. as for the quote... having a metalness workflow is NOT the same as being physically based.
of course, but why wouldn't you leave that choice up to the developers who're actually going to be using the engine?
Just describing their thinking behind why they did what they're doing. I think Fortnight supports colored spec, so I don't know why they don't allow others to do so. But there was someone definitely going for the "all physically based" and thus no alternate color spec route over at Epic.
Maybe if enough people complain they'll enable it for everyone. As you pointed out its not like they can't, it's just that they don't want to. And of course any custom specular, or any other such thing, is more work, of only a little. I think that's their reasoning behind disallowing such.
side note- calm down almighty_girl. I think epic know what they are doing, considering it's still in dev. Stop ruining so many thread with arguements.
Using a metalness map as apposed to a specular map, provided your shader takes energy conservation into account, its not any more or less physically accurate. Metalness is simply a more efficient (memory wise) function as you can pack a metalness map into 1 8 bit channel vs a a 24bit image for a full color spec. A full color spec map is more flexible and gives you more control over the specific reflectivity values (though how many assets actually need that extra control is likely pretty small).