o3d/converter/README.converter

The converter is a command line utility that will convert from COLLADA
to the form that the sample JavaScript scene loader (in
samples/o3djs/serialization.js) can read.

It is able to read COLLADA files as either DAE text files (and
associated asset files in subdirectories), or as part of a ZIP archive
(or KMZ file).  By default it outputs a gzip-compressed tar archive
(.tgz) which contains a JSON file and any associated assets (textures,
shaders, etc.) from the source COLLADA model.  By default we don't use
the .tgz extension, as it can cause problems with some webservers; we
put .o3dtgz on instead. The converter can optionally output a set of
flat files into a directory, which is the layout compatible with the
WebGL implementation of O3D.

In order to run the converter on Windows, you need the DirectX SDK installed.
Installing it requires administrator privileges.  You can get it here:
http://msdn.microsoft.com/en-us/library/bb219737(VS.85).aspx.

The command line arguments are as follows:

Usage: o3dConverter.exe [--base-path=<path>]
                        [--up-axis=x,y,z]
                        [--no-condition]
                        [--pretty-print]
                        [--no-binary]
                        [--no-archive]
                        [--convert-dds-to-png]
                        [--convert-cg-to-glsl]
                        <infile.dae> [ <outfile> ]

--base-path=<path>: This specifies the path elements to remove from
                    the beginning of the filenames placed into the
                    .o3dtgz file (to avoid including sensitive paths in
                    the .o3dtgz file).

--up-axis=x,y,z: This specifies the up-axis for the resulting model,
                 if you wish it to be different from the axis in the
                 COLLADA file.

--pretty-print: This simply formats the JSON file in the resulting
                .o3dtgz file to be human readable, instead of compacted
                into one line.

--no-condition: This will prevent conditioning of the shaders from
                COLLADA form to the form expected by O3D.  [This is
                useful only for pre-release data conditioned before
                the converter existed.]

The following four command line options are useful when converting
assets for the WebGL implementation of O3D:

--no-binary: Buffers, skins and curves will be written directly into the
             scene.json rather than into separate binary files.

--no-archive: Don't make a gzipped tar file, just flat files. Still
              takes the name of an archive file; for archive.o3dtgz,
              creates directory named archive/ and writes files inside.

--convert-dds-to-png: Convert all DDS textures to PNGs. For cube map
                      textures, writes six separate PNGs with suffixes
                      _posx, _negx, etc.

--convert-cg-to-glsl: Convert shaders using an external tool. Requires
                      python on PATH and either NVIDIA Cg toolkit
                      installed or cgc{.exe} in same directory as
                      converter.

<infile.dae|infile.zip|infile.kmz>: This is the input file in one of
                                    ZIP, KMZ, or DAE formats.  The
                                    DAE format expects to find assets
                                    relative to the .dae file.

<outfile>: An optional argument giving the name of the
           gzip-compressed tar archive output file.  By default
           this has the same basename as the input file, and has the extension
           .o3dtgz.