Clip Shadows

I am just starting to create shaders, and am using clip() to make part of my object transparent. My problem is that the clipped part still casts shadows. I am at a loss for how to make the transparent part of my object not cast shadows. Any help is very mush appreciated!

 

Anyone at least point me in the right direction?

 

I think Unity uses a separate pass with a tag called Tags { "LightMode" = "ShadowCaster" } to render shadows. To make "clipped" areas transparent you should just use or copy Transparent/Cutout shaders. Look at LegacyShaders/Transparent/Cutout/Diffuse shader source.

 

Yes, and look around the forum, this question has been asked many times.

 

After digging around in the vertexlit cutout for a while, I finally got it. If anyone is interested, here's the shadow caster pass

Code (CSharp):

1.         Pass{
2.         Name "Caster"
3.         Tags{ "LightMode" = "ShadowCaster" }
4.         Offset 1, 1
5.  
6.         Fog{ Mode Off }
7.         ZWrite On ZTest LEqual Cull Off
8.  
9.         CGPROGRAM
10. #pragma vertex vert
11. #pragma fragment frag
12. #pragma multi_compile_shadowcaster
13. #include "UnityCG.cginc"
14.  
15.     struct v2f {
16.         V2F_SHADOW_CASTER;
17.         float2  uv : TEXCOORD1;
18.         //float4 pos  : POSITION;
19.         float3 wpos : TEXCOORD0;
20.         float3 vpos : TEXCOORD2;
21.     };
22.  
23.     struct Input {
24.         float3 worldPos;
25.     };
26.  
27.     uniform float4 _MainTex_ST;
28.     float yBound;
29.     float xBound;
30.     v2f vert(appdata_base v)
31.     {
32.         v2f o;
33.         TRANSFER_SHADOW_CASTER(o)
34.             o.uv = TRANSFORM_TEX(v.texcoord, _MainTex);
35.         o.pos = mul(UNITY_MATRIX_MVP, v.vertex);
36.         float3 worldPos = mul(_Object2World, v.vertex).xyz;
37.         o.wpos = worldPos;
38.         o.vpos = v.vertex.xyz;
39.         return o;
40.     }
41.  
42.     uniform sampler2D _MainTex;
43.     uniform fixed _Cutoff;
44.     uniform fixed4 _Color;
45.  
46.     float4 frag(v2f i) : SV_TARGET
47.     {
48.         fixed4 texcol = tex2D(_MainTex, i.uv);
49.         clip(texcol.a*_Color.a - _Cutoff);
50.         if (i.wpos.x < 0)
51.         {
52.             clip(-1);
53.         }
54.         if (i.wpos.z > yBound)
55.         {
56.             clip(-1);
57.         }
58.         if (i.wpos.x < 0)
59.         {
60.             clip(-1);
61.         }
62.         if (i.wpos.x > xBound)
63.         {
64.             clip(-1);
65.         }
66.  
67.     SHADOW_CASTER_FRAGMENT(i)
68.     }
69.         ENDCG
70.  
71.     }

 

Thanks for sharing. I was also interested in this, and apparently the cutout shader(s) were indeed the place to look. I tried it (albeit I maybe don't need it any more as I've decided to go for an alternate solution for my project), and it indeed seems to work for clipping the shadows, but for some reason for me it had a weird striped side effect (see the "stripes" in the attachment). I wonder what the reason for that is. Anyone? Any ideas?

But I do have a tip for you too. From what I've read earlier I think having conditions like that in a CG section is inefficient as it's run by the GPU. Calling clip multiple times may not be too good either. Someone who knows better may confirm, further specify or correct me if I'm wrong, but I think it would be more efficient to put that whole clipping iffing into one line of code using the min function.

 

Not sure about the stripes, as they don't show up in my shader. Probably some weird interaction with whatever other code you have in the shader. As for the conditionals I switched it to

Code (CSharp):

1.     if (i.wpos.x < 0 || i.wpos.x > xBound || i.wpos.z < 0 || i.wpos.z > yBound)
2.     {
3.         clip(-1);
4.     }

which cleans it up a lot, but not sure how I would do it outside the CG (I really still have no idea how shaders work), or how to do this with the min function.

 

That is already much better: much cleaner, doesn't do multiple clips and doesn't necessarily need to go through all the conditions. I didn't mean that it would have to go outside the CG code, I just meant formulating the conditions as numbers (like xBound - i.wpos.x) and using the basic math function min for them. But what you wrote now might actually be better than the min solution - at least it's easier to read for a human. Which is computationally better may also depend on the probability of your conditions. It would be interesting to hear a comment from someone who understands the underlying architectural factors.

I did a very quick check about the stripes and actually it seems that it is not related to just having the shadow caster code in my shader, it's related to the combination of that shadow caster code AND using the shader as a replacement shader. Once again, really confusing.