In addition to these options, you must also set the options specific to the maps that you wish to generate. Map-specific options are described in Choosing and Configuring Maps to Generate.
Setting the Format Options
The format options specify the size of the output images, the level of super-sampling, and the texture projection used to sample the rendermapped object’s surface.
To set the format options
1. Apply a RenderMap property to one or more objects as described [here].
2. On the Basic tab, set the output images’ size by doing one of the following:
- From the Format options, enter an X Res value to set the output images’ width, and activate the Square option to make the images’ height equal to their width.
or
- From the Format options, deactivate the Square option and enter separate X Res and Y Res values to set the output images’ width and height respectively.
or
- Click the Set resolution from clip button at the top of the tab to match the output images’ resolution to that of an existing image clip.
A pop-up explorer opens displaying a list of the scene’s image clips. Select the clip whose dimensions you want to match.
3. Set the Super Sampling level.
This value specifies the number of samples taken for each pixel. For sampling purposes, the pixel is divided into a grid whose size is determined by the Super Sampling value. For example, if the value is set to 3, the pixel is divided into a 3 x 3 grid, and 9 samples are taken and averaged.
4. From the UV Coordinates list, select the texture projection that RenderMap uses to match points on the object’s surface to points in the new images. If there is no texture projection applied to the object, click New to create one.
• You cannot rendermap an object unless the object has a texture projection applied to it. The texture projection determines which points on the object’s surface correspond to points in the new texture images.
• If you rendermap an object using a spatial texture projection, the output images will look as though they were generated using a planar projection.
To correctly bake a spatial projection’s surface attributes into a rendermap, apply an appropriate projection to the object (UV for example). The texture will still be projected by the spatial projection, but the RenderMap will sample the object using the UV projection.
5. If you haven’t done so already, activate and configure the maps (surface color map, normal map, and so on) that you wish to generate using the current rendermap property, as described in Choosing and Configuring Maps to Generate.
6. When you’re finished, click the Regenerate Maps button to generate all activated maps.
Setting the Sampling Options
The sampling options specify which set of UV coordinates RenderMap uses to sample the rendermapped object, as well as control the output image’s quality to a large extent.
To set the sampling options
1. Apply a RenderMap property to one or more objects as described [here].
2. In the RenderMap property editor, click the Advanced tab.
3. From the Sampling options, select one of the following sampling methods from the Method list:
- Area Weighted samples every polygon that covers a texel. The resulting color is computed by averaging all the samples. Each sample is weighted by the fraction of the texel that is covered by the polygon.
or
- Simple Sampling samples once in the very center of each texel (or subtexel if the Super Sampling value is greater than 1). Therefore, only the one polygon that overlaps the center is sampled, even if there are other polygons in the texel.
4. Select a Partial Texel option to specify how partial texels are handled in output images.
- Normalize fills in the entire texel with the color of the covered part.
or
- Blend in Background Color fills in the uncovered part with the background color specified on the Surface Settings tab.
5. Activate Jitter to randomly deviate each sample from its calculated position. This is useful in situations where regular sampling makes small artifacts more visible.
6. Spill into empty texels specifies how deeply the filled texels bleed into adjacent empty texels in the output image, in texels. A value of 0 results in no bleeding.
7. If you haven’t done so already, activate and configure the maps (surface color map, normal map, and so on) that you wish to generate using the current rendermap property, as described in Choosing and Configuring Maps to Generate.
8. When you’re finished, click the Regenerate Maps button to generate all activated maps.
Setting the Virtual Camera Options
RenderMap uses a virtual camera to sample the surface of the rendermapped object. You can control some of the virtual camera’s attributes to modify the output image.
To set the virtual camera options
1. Apply a RenderMap property to one or more objects as described [here].
2. In the RenderMap property editor, click the Advanced tab.
3. From the Virtual Camera options, set the Distance from Surface to specify the virtual camera’s distance from the rendermapped object.
Non-zero distances are useful when you want to include other scene elements in the output images. For example, you might have a wall with vines creeping up it. By increasing the Distance from Surface, you can bake the vines into the output images, even though they are separate objects.
Unless you want to incorporate elements other than the rendermapped object into the final rendermap image, it’s best to keep the Distance from Surface set to 0. Non-zero settings increase the time it takes to generate the rendermap, often significantly.
4. If your object is rendered using final gathering, and you find that the rendermap output image contains undesirable artifacts, increase the Final Gathering Smoothness value. The higher the value, the “smoother” (fewer artifacts) the result.
You should adjust your final gathering accuracy setting to get rid of artifacts in the resulting images. Once you’ve done that, adjusting the Final Gathering Smoothness value can help reduce the appearance of any remaining artifacts.
5. Set the View option to specify the virtual camera’s point of view:
- Perpendicular to Surface: the virtual camera samples the rendermapped object(s) from a position perpendicular to the object(s) surface, and from the distance specified by the Distance from Surface setting.
or
- Scene Camera: the virtual camera samples the rendermapped object(s) from the direction of the scene camera specified in the render options, but from the distance specified by the Distance from Surface setting. As a result, the RenderMap sampling ray may not originate from the scene camera’s position.
This is useful when an object is to be viewed from a single viewpoint and you want to include view-dependent effects like specular highlights or reflections.
If you are generating an Illumination surface color map, specular highlights and reflections are automatically forced off, making camera direction irrelevant. As a result, the View options are unavailable.
6. If necessary, activate the Ignore Rendermapped Objects option. When this option is activated, RenderMap casts a ray to find the surface to be sampled, but ignores the object(s) affected by the RenderMap property. As a result, the color information that is computed is taken from whatever object the ray hits, and not the object being rendermapped.
This option is useful for making sprites, and is generally preferable to the technique of putting a Constant shader with 100% transparency on the rendermapped object(s).
Keep in mind, however, that the rendermapped objects will still appear in reflections and through transparent portions of the surface when the shader is evaluated. If your scene contains reflections and transparency, using a Constant shader may still be appropriate.
7. If you are using RenderMap to transfer attributes from one surface to another, activate the Bidirectional Tracing option. This causes RenderMap to shoot each ray in both directions, if necessary, and choose the best of the two resulting samples. If neither sample is appropriate, both are rejected.
For bidirectional tracing to work properly, make sure that the Ignore Rendermapped Objects option is activated and the virtual camera View is set to Perpendicular to Surface.
8. Select or deselect the Front Facing Triangles and Back Facing Triangles options to include or exclude the object’s front and/or back facing triangles (relative to the scene camera) in the RenderMap calculation.
9. If you haven’t done so already, activate and configure the maps (surface color map, normal map, and so on) that you wish to generate using the current rendermap property, as described in Choosing and Configuring Maps to Generate.
10. When you’re finished, click the Regenerate Maps button to generate all activated maps.
Bidirectional Tracing Explained
Bidirectional tracing is best explained by the following example:
Let’s say you want to compute a high-resolution character’s normal maps relative to a low-resolution approximated character, for use in games.
This can be achieved by making the low-resolution character transparent (by activating the Ignore RenderMapped objects option, for example) so that the rendermap is relative to the low-resolution character, but the ray-casting captures the color/normals of the high-resolution character. However, this method can fail if the low-resolution character does not fully encompass the high-resolution character.
When this is the case and bidirectional tracing is activated, RenderMap first casts a ray towards the low-resolution character, but ignores it, attempting to capture the high-resolution character. If the ray misses the high-resolution character, or hits on the inside of the high-resolution character as defined by the surface normal, a new ray is cast from the rendermapped surface in the opposite direction.
If this second ray hits the inside of the high-resolution character, this other point is used as the cast position. If it misses the high-resolution character, it is treated as missing the object. For a normal map, this means a normal of 0,0,0 is returned (which, when biased, is .5,.5,.5).
Using bidirectional tracing is preferable to increasing the Distance from Surface value to capture portions of the surface that are outside of the low-res character. Increasing the distance from surface can fail in tight areas such as armpits because it “catches” the wrong part of the surface.
Bidirectional tracing causes additional rays to be cast, and may affect computation times.
