Baker Documentation

Surface Baking, sometimes call skinning, is the transformation of an object procedural texture, lighting or surface definition into a traditional texture map. Baker adds these possibilities to Carrara.

    To bake an object shader, select first the object in the scene, then choose the menu item Edit->Baking or press Ctrl+F. Note that several objects can be selected at the same time: Baker will generate as many maps as there're selected objects.

    3 map types can be exported. The first one, the Texture Map, transforms one of the shader channel into a texture map. Look for example  at this vase model.  It's shader  is a 3D  solid texture, both in the  diffuse channel and in the bump channel. To transform this  channel shaders into texture map, first be sure that the UV mapping of the object is a coherent one. Baker uses the UV map to transform 3D shaders into 2D texture maps, so if the UV mapping has flaws, like overlapping triangles, these will be visible into the baked map. Then in the Baking dialog, select the Diffuse and the Bump channels. The background color is used for unassigned pixels (the UV mapping might not cover all the UV space). If the object has several shading, 3 options are available: you can choose to export all the  domains into 1 map (Merge Shading Domains), to  create a different map for every shading domain (Separate Shading Domains), or to export only 1 of the domains by selecting it in the  popup menu.

These maps can then be used as traditionnal texture maps into the vase shader to fasten the rendering of the object.
The second map type Baker can generate is a Light Map. The light map is created using the current scene lighting and uses a raytracing engine (global illumination and NPR aren't supported) to render it into a flat texture map. The lighting can be rendered alone, or premultiplied with the diffuse color, the specular light, and/or the bump mapping. Here are 2 examples of light map generated by Baker:

The vase was then rendered using the lightmap instead of using a 3 spot lighing. The rendering is now more than 6 time faster than the real lighting scene.

The 3rd type of map is the Normal Map. The normals can be computed with or without the bump mapping, and can be determined in global or local space. A shader named Final Normal Map is available to read the map computed in a local space. This map can be used to store some geometry complexity from a model and restitute it as a bump map to a lower resolution one. This way the final object is much faster to compute. Here the first rendering was done with a real geometry, the second with a low polygon version of the object and a normal map that countains the details.

To read the normal map, Baker includes a special shader named Final Normal Map.

Differences between Deeper Normal Map shader and Baker Normal Map shader:
The Normal Map shader and the Final Normal Map shader read 2 different types of normal maps. The first one (include in Deeper) uses the information countained in the map to perturbe the existing normal on the object. It will add the read normal to the current one. The second one, called Final Normal Map and included in Baker, replace the object normal by the read normal. The object normal information are completly erased and a new normal extracted from the map is used instead. As a result, if a color is set as the child of the Final Normal Map shader, all the normals are forced to be in the same direction and the object display fake flat looking shadows.


Here are a few links where you can find more information on Normal Maps:

About the Normal Map used in Baker:
http://amber.rc.arizona.edu/lw/normalmaps.html
http://www.drone.org/tutorials/normal_maps.html
http://www.pinwire.com/article82.html

About the Normal Map used in Deeper:
http://members.shaw.ca/jimht03/normal.html