ptrender - renders a point cloud file into an image file

SYNOPSIS

ptrender [options] pointcloudfile channelname imagefile

DESCRIPTION

ptrender turns any channel of a point cloud file into a 2D image by either projecting it onto a uv set (stored in the point cloud itself) or by splatting the points given the camera matrix stored in the point cloud file.

In order to use ptrender to unwrap a function onto a parameterization, the point cloud file must contain u, v coordinates stored in the P value - in other words, the bake3d call used to generate the point cloud should look something like:

    bake3d(bakefile, "pointcloud", point(u, v, 0), ...

ptrender can then map any channel into the plane [0,1]x[0,1] by reprojecting every single sample into the corresponding pixel.

ptrender can also be told to reproject samples in the space of a camera. The camera used will be the same one stored in the point cloud file (which typically will also be the rendering camera used during the generation of the point cloud file in the first place). Hence in this case the bake3d call used to generate the point cloud should use a position value visible from the camera, which may be as simple as just P itself:

    bake3d(bakefile, "pointcloud", P, ...

Several algorithms are provided for the splatting process itself. Diffusion algorithms are provided which are very robust in cases of sampling that can have widely different densities with no particular structure; also, the hole filling used takes care of cases where leaking of black samples can happen if the parameterization has discontinuities (by simply filling in the gutter with samples from neighboring pixels). For camera based projections a Z based splat and an area estimation algorithm based on point sizes are also provided.

ptrender supports multiple output formats via the same display driver system used by prman.

OPTIONS

-depth depth

Specifies the bit depth of the output image. Valid arguments to this option are: byte, short, and float. Note that depending on the choice of display driver this option may be ignored or overridden. If not specified, the default depth is "byte".

-dspy driver

Use the specified display driver. This determines the format of the output file. If not specified, the default driver used is TIFF.

-dspyargs arguments

Passes the specified arguments to the display driver. The arguments must be a single string which contains a parameter list with inline declarations, exactly as would appear in the RIB stream. In other words, it should contain values that look exactly like what can be passed on the Display line in a RIB file. As an example, here is a valid command line which sets three parameters known to the TIFF driver:

        ptrender -dspy tiff -dspyargs '"float[2] origin" [10.0 256] "float[2] OriginalSize" [1024 1024] "string compression" ["zip"]' \
            sphere.ptc Ct sphere.tif

Note the usage of single quotes to protect the double quotes and square braces contained with the argument string from being parsed by the shell.

-project mode

Controls whether to use uv or camera based projection. Valid arguments to this option are: uv and camera. If uv projection is used, the point cloud file must contain u, v coordinates stored in the P value, and the points will be projected onto the plane [0,1]x[0,1] using those coordinates. If camera projection is used, the points will be projected to the image plane using the world to NDC space transformation stored in the point cloud file. If not specified, the default projection mode is uv.

-size width height

Controls the format of the output image. If not specified, the default size is 512 x 512.

-splat mode

Determines the algorithm used to splat points onto the image plane. Valid choices are:

• none: points are drawn as single pixels directly on the image plane, left over areas are left black
• diffusion: points are drawn as single pixels are drawn directly on the image plane, left over areas are filled in using diffusion
• smooth: like diffusion, except that a filter kernel is also applied
• zsplat: points are splatted with precedence based on their z values. The size of the splat will be determined via the radii stored in the point cloud file. This splat mode is most useful when used with the camera projection mode
• area: points are drawn as rectangles on the image plane with their area determined by their radii stored in the point cloud file

If not specified, the default splat mode is "smooth".

-verbose

Turns on verbose output. Off by default.