shader model info

[Jakey757]

I think animation period, phase, and scale should be thought of like a graph. Period should be the base length of the graph, phase would shift the values in the graph left or right and scale would repeat the length of the period or cut it short. Animation function would simply change the shape of the graph.

Perpendicular and parallel angles used for reflections create a fresnel effect.

The orange in this example being the perpendicular and the green being parallel

I couldnt fit it in the first image but detail map scale applies after the shader UV map scale.

Rotation animation centers are based off of the UV scale in the gbxmodel/model tag

In this example for a animation function of zero you would have to use 1.00023 for x and 1 for y. For the cyborg gbxmodel you would use 0.5 for both as the base map UV scale is 1.

Alpha test vs alpha blended decal

By default shader model uses alpha test which treats alpha transparency as 1 bit alpha, this creates aliasing and hard edges but sorts much better with other transparent surfaces. The alpha blended decal flag forces shader model to use alpha blend which is the default transparent shaders like chicago uses, alphas are in 8 bit and this provides very smooth transparency but at the cost of sorting terribly against other transparent surfaces, for this reason its relegated to decal like shaders where they are against solid objects.

Radiosity note: If detail level is set too high, the mesh the shader is on could attempt to subdivide too much and fail to render lightmaps. Some maps will fail to render lightmaps even with original shaders because these had to be changed on a per scenario basis. (irrelvant to shader model but its part of the tag so i thought i'd add it)

Self illumination color lower and upper bounds are interpolated between linearly with the animation function. Setting a color source can allow you to set the lower and upper bounds in the object itself and choose to use the flags to change the interpolation.

Change color channel source C will have it's color values overwritten if the scenario is multiplayer, with the hardcoded colors. If the channel source is A, it will be overwritten by actor variant color values if spawned as an ai (this one i haven't proven yet, it just seems to follow some kind of convention bungie used).

For detail map functions
double biased multiply: double multiply the detail map with the diffuse map, brightens the detail map before multiplying (standard setting, detail maps with grey backgrounds and black details)
multiply: multiply the detail map with the diffuse map (white backgrounds and grey-black details)
add: add the detail map to the diffuse map (useful for detail maps with light colors. black backgrounds and light details)

Base map attributes for shader model
The RGB is treated as diffuse color, alpha is treated as transparency. Halo predates physically based rendering, so lighting and shadows are typically baked or authored into the base map.

Ending thoughts
Maybe get actual image examples from halo itself if they can be used.