You can use the extra information in deep raster images, such as Depth, Normals, and various mattes, in more advanced compositing operations. See rendering deep raster images .
Merge deep raster information in separate files
The renderer can store the extra information in a single deep-raster
file format (
.pic), or in multiple “shallow” files. You may need
to merge color planes together with the deep raster information, for
certain operations (like Z Composite, which
Zdepth to be present in the input).
For deep rasters that have been saved over several files, you will need to merge the deep raster planes and the color planes into a single sequence. See how to import an image as a plane of another image .
Adjust render passes
In your surface shaders, you can export float (or vector3) variables such as diffuse, specular and ambient. You must write the appropriate values to them; they are not automatically filled in.
With scalar (matte) information, you can use the mattes as masks for color correction operations, or other filters like blur, to adjust portions of the image.
With vector3 (color) information, you can subtract the pass color from the image color (C), manipulate it, and then add it back into the composite. You can also use vector3 mattes as masks, just as scalars, but the mask will be applied as a 3 channel mask (Mask.R, Mask.G, Mask.B) and affect the different color channels differently.
You can do similar operations on shadow passes, allowing you to darken, blur or add noise to shadows.
Z depth compositing is useful for compositing volumes or particles. For example, compositing falling snow around foreground elements.
To do Z depth compositing, each COP needs either a depth (
Pz) or point (
Use the Z Composite operator to composite the images, instead of a traditional “ over ”. In addition to transparency, Z Composite looks at the Z depth of each point to see if it should be in front of or behind the other image.
The depth information is also updated with the new Z positions, so you can do multiple Z composites in series correctly.
If you have many Z composites to perform and some have transparency, put all the Z composites dealing with transparent images at the end of the Z composite chain, if possible. This will minimize problems with Z compositing and transparency (since only 1 Z value can be stored, and transparency requires one for each transparent surface, plus the final opaque surface).
Relighting involves simulating the effect of adding extra lights to the rendered image. Houdini uses the depth and angle information encoded in the deep raster to calculate the interaction of the new lights with the surfaces in the image.
To do 3D lighting, you must have Point and Normal planes in your image.
Use the Lighting operator to position a single directional or non-directional light in the scene, or do volume lighting with the parameters on the Lighting node’s Atmospheric tab.
You can generate a normal map from a point map with the Lighting operator, but the results are not as smooth as when you render out a real Normal plane.
To add multiple lights, input the image to multiple Lighting nodes and turn off the Add to original color parameter in each. Then use an Add or Screen operation to combine the output of the nodes.
Depth of field
You can use the depth information encoded in the deep raster to adjust the apparent depth of field in the image, but de-focusing parts of the image outside the focus range.
To adjust the depth of field, the image must have a Z-depth or Point plane.
Append a Defocus COP; set the Per Pixel Defocus parameter. Adjust the size to get the desired amount of blur.
See how to preview different planes in the node tile’s thumbnail.
Use the Fog operator to add various atmospheric effects to your network, such as normal fog, layered fog, and heat waves. The image must have a Z-depth plane or Point plane.
For normal fog, you can specify the color of the fog and its density.
For layered fog, you can specify the above plus the height that it starts/ends at, plus a drop-off region to make the transition smooth.
You can add 3D noise to the fog to make it more varied (though this takes quite a bit longer, as it does ray marching through the noise).
You can add an image to the Fog node’s second input, which varies the fog color based on the second image’s color.
Apply an environment map
World space pixel culling
Use the Geo Key operator to partition, or cut out, areas of the image based upon their 3D position or normal direction.
For images with a Point plane, use clipping planes and distance from a point to isolate or remove elements.
Use a static direction vector to isolate all normals pointing in that direction. Use a point to isolate all normals pointing towards that point.
To get a smooth transition, you can use a drop-off area or angle.
All points removed close to a given point.
All points kept facing a certain direction (up).
All points removed in front of a plane.