Humans Sketch + Cascaded Shadow maps

I like the idea of 2 different body types with the Human race.

As for the graphics issues, sadly that is not an area I excel in.

CSM isn't hard once you have single shadowmapping working. You can do the following to approach CSM step for step:

1. Be able to calculate the shadow-camera to tightly fit a part of the view frustum, best to use the near/far clipplane as input

1.1 calculate frustum corner

1.2 project frustum corner on shadow-camera plane (assumption: direct lighting)

1.3 min/max projected corner to determine positioning and view port

2. Be able to render the shadowmap only on a restricted area of the texture

2.1.utilize the viewport settings of the camera & scissor test

3. Rendering of the CSM (assumption: 4 layers)

3.1. subdivide the frustum equally at first (later tweak it)

3.2. for each subdivision, calculate the shadow camera(1.) and render it to a quarter of the shadow map(2.)

4. Using the CSM(assumption: only in the fragment shader, once it works you could transfer parts of it into the vertex shader if necessary).

4.1. Determine the depth of the pixel

4.2. map the depth on the two closest layers (e.g. depth=0...1 , layer 1 = floor(depth*4), layer 2=max(3,ceil(depth*4)) )

4.3 for each layer calculate the offset (shadowmap quarter), then get the shadow texels

4.4 calculate an alpha value to represent the relative position of the pixel depth value between the two layers (layer1<=pixel_depth<=layer2)

4.5 blend the two depth values, tweak it later to optimize the transition between the two layers.,

Here's an article parralel split shadow mapping, very similiar to it, maybe the same ?

Wow, very informative, ill pass this along to telanor and see if this helps. Shadows are kicking our asses right now just like the point light was!