| 1 |
|
.\" RCSid "$Id$" |
| 2 |
|
.TH RTPICT 1 3/19/2018 RADIANCE |
| 3 |
|
.SH NAME |
| 4 |
< |
rtpict - generate a RADIANCE picture or layerd image using rtrace |
| 4 |
> |
rtpict - generate a RADIANCE picture or layered image using rtrace |
| 5 |
|
.SH SYNOPSIS |
| 6 |
|
.B rtpict |
| 7 |
|
.B "-n nproc" |
| 100 |
|
.I "irradiance.hdr" |
| 101 |
|
and some other output types become irrelevant (i.e., 'r', 'x', 'R', and 'X'). |
| 102 |
|
If one or more of the associated output files already exists in the |
| 103 |
< |
destination directory, it will be overwritten with the new data. |
| 103 |
> |
destination directory, an error will be printed and the command will abort. |
| 104 |
|
.SH EXAMPLES |
| 105 |
|
To render a scene with four processes: |
| 106 |
|
.IP "" .2i |
| 112 |
|
.SH NOTES |
| 113 |
|
Users should seriously consider using |
| 114 |
|
.I rpiece(1) |
| 115 |
+ |
for multiprocessing |
| 116 |
|
when irradiance caching is employed with a shared ambient file. |
| 117 |
|
If |
| 118 |
|
.I rtpict |
| 119 |
< |
is used in multiprocessing mode with a shared irradiance cache, |
| 120 |
< |
it is trying to compute almost |
| 121 |
< |
the same part of the same scanline in different processes, which |
| 122 |
< |
results in many redundant calculations. |
| 123 |
< |
The |
| 124 |
< |
.I rpiece |
| 125 |
< |
program avoids this by working on separate tiles in each |
| 126 |
< |
process, with less overlap in the indirect irradiance calculation. |
| 127 |
< |
Either program will benefit from an "overture" run of a |
| 128 |
< |
single-process rpict to create the ambient file using a low-resolution |
| 129 |
< |
rendering that is discarded, but this will not completely solve the |
| 130 |
< |
problem for |
| 131 |
< |
.I rtpict. |
| 119 |
> |
is used with |
| 120 |
> |
.I \-n |
| 121 |
> |
greater than 1, it tries to compute neighboring parts |
| 122 |
> |
of the same scanline in different processes, which can |
| 123 |
> |
result in many redundant calculations and cache entries. |
| 124 |
> |
.I Rtpict |
| 125 |
> |
will automatically run a randomly-ordered overture calculation |
| 126 |
> |
to mitigate this problem, but it is not a perfect solution. |
| 127 |
> |
.I Rpiece |
| 128 |
> |
avoids redundant cache calculations by working on separate tiles in each |
| 129 |
> |
process, which brings other benefits as well, such as enabling the |
| 130 |
> |
.I \-ps |
| 131 |
> |
option for image plane sampling. |
| 132 |
|
.SH AUTHOR |
| 133 |
|
Greg Ward |
| 134 |
|
.SH "SEE ALSO" |
| 135 |
|
getinfo(1), mkpmap(1), oconv(1), pfilt(1), |
| 136 |
|
pvalue(1), rad(1), rcode_depth(1), rcode_normal(1), rcode_ident(1), |
| 137 |
< |
rpiece(1), rpict(1), rsplit(1), rtrace(1), rvu(1), vwrays(1), |
| 137 |
> |
rcrop(1), rpiece(1), rpict(1), rsplit(1), rtrace(1), rvu(1), vwrays(1) |
