And not a single on of them worked for my water. I have been trying to make sine wave shader water for ages and have had shadowing issues because of normals and can’t find a fix.
(please note, this is not a shadow maps issue, it’s not a ray trace shadows, or any of that crap.)
I am guessing they removed the AbsWorldPosition (excluding materials offets) node too? not surprised.
The only thing that has even slightly improved my normals so far has been hooking a constant–>COS–>normals and keeping the constant at 0. it’s improved on opaque and transparent modes but if I add metallic sheen to the material it goes haywire.
I watched videos on youtube, I watched ben clowards DDX DDY videos as well as some british dudes.
I need a fool proof way of recalculating normals on sinewave water, and something that is not 50 nodes long.
As you can see in my previus picture, the lightning on metallic set to 1 is still trash.
in my guide I posted in OP they have a world position before material offsets, where the heck did that go?
The plane picks up other shadows fine, it’s the self shadowing that is broken.
for example, if I put a block over it, it will pick up the blocks shadow perfectly and distort it with the waves in a reasonably nice way.
I think you probably have to use world position all the way ( including the normal correction ), or not at all. I have no idea how to do that, I’m afraid…
but the problem is that whatsoever lies beneath the X line on sine (here I’m using cosine to avoid the .25 offset) is not shading correctly regardless of whether or not I am recalculating normals.
EDIT
if I choose double sided, because the positive on the first side works, the other side is negative (concaved because its protruding on the other side) so the other side has shading issues.
math? find a way to compute an angle lerp in one quadrant. the normal angle is identical at the peaks of the curve. ofc you can scale the normal with wave height and renormalize the length.
So dabbling around last night I found that I have two problems.
the first one is that if the emissive color is darker than, say, 180 = I get horrible shadows in the crevices of the backside of the mesh. If I kick the color up, the shadows return to normal.
I tried this with lumen and screen space reflections.
the second is that the multiplication after sine or cosine breaks the shadows on the top layer, while if
I use a power node instead (does not matter what the power is) it fixes the shadows on the top and the emissive still breaks them on the bottom.
the problem with the power node is without some form of smoothing it is very blocky because of the exponential timing. multiplication is smoother.
One noticable difference is that exponential does not go negative, it stays flat and then increases while cosine with multiplication goes -1, 0, 1 (or .25 offset in case of sine).
The shadows are much improved in the actual level if I place a plane or box, the box seems to do better because its not utilizing unreal’s “two sided” option which I suppose is too much to ask for the engine to work with that and these janky sine waves.
However, the shadows still look “noisy” I kicked shadows to highest, lumen to highest, SSR to highest
ray cast shadows, and literally no difference.
Also I have noticed that when the wave is at an angle, looking at it from the top looks fine, looking at it from the “side” of the angle looks fine, but looking at the angle slope coming towards me has weird color/shadow reflection noise.
I don’t seem to get any of this when I have small waves, so maybe the solution for Notreal engine is to
use little waves/scale mesh up?
