Minecraft Alpha 1.1.2_01 Lighting

2023-03-20

I used apitrace with MultiMC, following this tutorial to rip world meshes out of the game. Then I whipped up a quick program to render those meshes with their light values on each face. This version of Minecraft uses legacy vertex arrays (glVertexPointer and friends) for each chunklet. A chunklet is a 16x16x16 piece of the world. Each chunklet gets its own legacy vertex array. Each chunklet is baked into a Display List that will be used to draw it later:

glNewList(some_val,GL_COMPILE);
...
glTexCoordPointer(2,GL_FLOAT,32,tcptr);
glColorPointer(4,GL_UNSIGNED_BYTE,32,cptr);
glVertexPointer(3,GL_FLOAT,32,vptr);
glDrawArrays(GL_TRIANGLES,0,vCount);
...
glEndList();

The stride is 32 bytes, but the total vertex data only takes up 24 bytes, so the last 8 bytes of every vertex are just padding. Interestingly it uses a full 4 byte color for each vertex, meaning the game could easily support colored lights if someone wanted to mod that in.

Anyways I ripped these chunklet meshes out of the game and rendered the byte light value on each face. In the following images we can see the light propagation for each cube normal when exposed to zero sunlight:

Compare +yclose to +xclose. You'll see that the highest block face value (non-torch) for +y is 201, whereas that of +x is 120. So, clearly the face normal effects the final brightness of the face. You could call it "ambient light". If you dig into the source of the game with Mod Coder Pack, you'll find a function called func_1243_a in RenderBlocks.java which multiplies the brightness of blockfaces by ambient light. Its coefficients for each cube normal are as follows:

-x and +x: 0.6f
-y: 0.5f
+y: 1.0f
-z and +z: 0.8f

Since the coefficient of +y is 1.0, the +y values we see in the wide shot are the base brightness values, which come from an array generated on startup by the following code:

field_1042_i = new float[16];
float f = 0.05F;
for(int i = 0; i <= 15; i++)
{
    float f1 = 1.0F - (float)i / 15F;
    field_1042_i[i] = ((1.0F - f1) / (f1 * 3F + 1.0F)) * (1.0F - f) + f;
}

This produces the following array:

0.050000
0.066667
0.085185
0.105882
0.129167
0.155556
0.185714
0.220513
0.261111
0.309091
0.366667
0.437037
0.525000
0.638095
0.788889
1.000000

12
16
21
26
32
39
47
56
66
78
93
111
133
162
201
255