why the back of the specular reflection is all black when arnold renders the normal? Even if exit to background is enabled.

why the back of the specular reflection is all black when arnold renders the normal? Even if exit to background is enabled.

2689081964
Enthusiast Enthusiast
212 Views
5 Replies
Message 1 of 6

why the back of the specular reflection is all black when arnold renders the normal? Even if exit to background is enabled.

2689081964
Enthusiast
Enthusiast

why the back of the specular reflection is all black when arnold renders the normal? Even if exit to background is enabled.

  1. when I used the normal map.
    It seems that it can't calculate the ray depth for the parallax effect generated by the normal map.
    Other renderers may
    So I tried to find a way to get arnold the right result, but I couldn't.
  1. The test scenario is just a plane plus the first picture in Google search normal map.
    And aistandardsurface with a metal degree of 1 and a roughness of 0 is used.
     
     
     
     

    82b0a2af4e8904c7dea2a5cc6a4fd6.png

image (1).png

2689081964_0-1745571347210.png

2689081964_1-1745571358657.png

2689081964_2-1745571367900.png

2689081964_3-1745571378120.png

 

 

2689081964_0-1745572621552.png

 

 

0 Likes
Accepted solutions (1)
213 Views
5 Replies
Replies (5)
Message 2 of 6

2689081964
Enthusiast
Enthusiast

2689081964_0-1745571930697.png

this is new arnold render,but the result is still not correct.

2689081964_1-1745571983383.png

And this is to open mikk, and the result has not changed.

0 Likes
Message 3 of 6

2689081964
Enthusiast
Enthusiast

2689081964_2-1745572046305.png

It's not just me. This is a youtube video tutorial by a great artist.

arnold seems to render normally only in a small range of normal or bump, and once it exceeds a certain angle, it will be displayed in black. This is completely different from other renderers.

2689081964_3-1745572152332.png

So this is the correct result, or the result of other renderers.

2689081964_4-1745572168963.png2689081964_5-1745572177857.png

And these two are arnold's results, and they are both wrong results.

 

0 Likes
Message 4 of 6

2689081964
Enthusiast
Enthusiast

2689081964_6-1745572335328.png

2689081964_7-1745572375829.png

 

I don't know whether other renderers use some approximate algorithm to simulate, maybe it is not necessarily accurate, but at least the results rendered by other renderers or mayaviewport2.0 are very good and correct in my opinion. If you add a normal with a hemisphere to a plane and add a real ball to the scene, you can see the results of mayaviewport2.0 at the same angle, and the reflection results of the real ball and the plane (with normal map) are exactly the same. If everything should be presented in real geometry, the meaning of normals will disappear. The meaning of normals is to present high-mode details for fewer polygonal meshes and deceive vision. However, the normal angle supported by arnold is too small, and a slightly stronger discovery map will cause problems. That is to say, basically downloading a set of pbr maps from online materials can't be restored correctly. For example, downloading the normal of a brick wall casually can have beautiful effects in other renderers, while arnold will have inexplicable shadows.

0 Likes
Message 5 of 6

2689081964
Enthusiast
Enthusiast

a

0 Likes
Message 6 of 6

2689081964
Enthusiast
Enthusiast
Accepted solution

所以事实上其实arnold的结果并没有问题,光线追踪渲染器理论上确实会得到这样的结果。

所以事实上是我错了,虽然从理论上来说mayaviewport的结果似乎看起来不错,但是确实是有问题的即使这个结果作为人类来说看起来似乎很合理也好看不少。

 

原因是一些视角下法线贴图改变的法线会导致视角变成在假想法线的平面背面,导致了黑色的镜面反射。maya的viewport预览的法线效果虽然不会有黑色,但其实也并不算正确,它只是把越过法线假想平面的视角继续计算然后找到对应的hdri角度的像素显示出来而已。

2689081964_0-1754904139951.png

 

0 Likes