+Generate a precomputed contribution photon map containing a fraction of
+\fInphotons\fR photons (specified with the \fB\-apP\fR option, see
+below), and output to file \fIfile\fR. Only light sources may contribute,
+and are specified via their modifier(s) with a subsequent \fB-m\fR option
+(see below). Each precomputed photon accumulates contributions from the
+\fIbwidth\fR nearest photons into bins (specified with the \fB-bn\fR option,
+see below). If \fIbwidth\fR is less than the number of bins, a warning is
+issued, since some bins will be zero.
+
+.IP
+Ideally, \fIbwidth\fR should be a multiple of the number of bins. If too
+many bins (default 50%) are still zero, a warning about an excessive empty
+bin ratio is issued, indicating potentially biased contributions, and that
+the number of photons and/or the bandwidth should be increased further.
+However, if very small contributions are indeed expected in some bins,
+this warning should be taken with a grain of salt.
+
+.IP
+Precomputed contribution photon maps can be used by \fIrcontrib(1)\fR to
+very efficiently evaluate binned light source contributions (e.g. from sky
+patches) at multiple points by looking up the single nearest precomputed
+photon. In contribution mode, \fImkpmap\fR is therefore designed to behave
+similarly to \fIrcontrib\fR. To ensure the latter obtains binning parameters consistent with those used in the precomputation, \fImkpmap\fR outputs an
+option file \fIfile.opt\fR which must be passed to \fIrcontrib\fR
+via its \fB@\fR option.
+
+.IP
+The \fIcomp\fR parameter specifies a compression ratio for the binned
+contributions in the range [0, 1], where 1 corresponds to maximum compression,
+and 0 implies no compression at all. Compression is recommended as it gives the
+user control over the size of the binned contributions, which are saved in a
+separate file for each modifier; depending on the number of photons and bins,
+these files can become very large. Note however that \fImkpmap\fR uses a
+\fBlossy\fR wavelet compression, which can introduce artefacts at
+higher compression ratios if the lighting is complex (e.g. exhibits high
+gradients). In typical scenarios, a compression ratio of 0.7-0.8 yields an
+acceptable compromise between accuracy and space savings.
+
+.IP
+With this option, \fImkpmap\fR uses a modified photon distribution
+algorithm that increases the odds that all selected light sources contribute
+approximately the same number of photons. Because of this, \fImkpmap\fR cannot generate a contribution photon map in combination with others in a single run,
+and other photon maps specified on the command line will be ignored.
+
+.IP
+\fBNOTE: This option requires out-of-core storage, which is currently not
+enabled in the default build, nor is it supported on Windows. On platforms
+that do support it, precomputed contribution photon mapping can be enabled at
+Generate a transient photon map with the given propagation velocity.
+This is a variant of the global photon map with the distinction that
+photons can be located in space and time. When a transient photon interacts
+with a surface, its time of flight, subject to its velocity, is recorded along with its position (technically, this is internally expressed as the photon path length to simplify the spatio-temporal distance metric used for photon
+lookups).
+
+The velocity is expressed in units of 1/sec in relation to the
+scale of the simulation geometry. For reference, the speed of light in a
+vacuum is 299792458 metres/sec (or 3e8 if you're not Stephen Hawking).
+The velocity is currently treated as constant in all media, and is
+consequently not adapted for the index of refraction in \fIdielectric\fR
+and \fIinterface\fR materials. Relativistic effects are also not accounted
+for, since Stephen Hawking doesn't use RADIANCE anyway.
+
+Transient photon maps are particularly useful to study the propagation of
+light in extreme slow motion. See also \fIrpict(1)\fR for details on how
+transient photons can be rendered at specific points in time.
+
+\fBNOTE: This option is experimental and requires in-core (kd-tree) storage,
+which is enabled in the default build; it is currently not supported with
+(c) Fraunhofer Institute for Solar Energy Systems,
+.br
+(c) Lucerne University of Applied Sciences and Arts (Hochschule Luzern),
+.br
+(c) Tokyo University of Science.
+
+.SH ACKNOWLEDGEMENTS
+Development of the RADIANCE photon mapping extension was supported by:
+
+.RS
+\fIFraunhofer Institute for Solar Energy Systems\fR funded by
+the German Research Foundation (\fIDFG LU-204/10-2\fR: "Fassadenintegrierte
+Regelsysteme (FARESYS)"),
+
+\fILucerne University of Applied Sciences and Arts (Hochschule Luzern)\fR
+funded by the Swiss National Science Foundation (\fISNSF 147053\fR: "Daylight
+redirecting components" and \fISNSF 179067\fR: "Lightfields for spatio-temporal glare assessment"),
+
+\fITokyo University of Science\fR funded by the JSPS Grants-in-Aid for Scientific
+Research Programme (\fIKAKENHI JP19KK0115\fR: "Three-dimensional light flow in architectural spaces") under the supervision of Prof. Nozomu Yoshizawa.
+.RE
+
+In particular, the author thanks Prof. Stephen Wittkopf and Dr. Lars Grobe
+for resuming development at Hochschule Luzern, and Prof. Nozomu Yoshizawa
+for the opportunity to collaborate with the Tokyo University of Science.
+Many thanks also to the many individuals who tested the code and provided
+valuable feedback. Special greetz to Don Gregorio, PAB and Capt.\~B!
+
+.SH "SEE ALSO"
+rpict(1), rtrace(1), rvu(1), rcontrib(1),
+.br
+\fIThe RADIANCE Photon Map Manual\fR,
+.br
+\fIDevelopment and Integration of the RADIANCE Photon Map Extension: