random nan pixels in render,
the only mention of nan in log is below, but they dont reference the actual maya error pixels i can see.
ignoring infinity/NaN image sample in pixel (1152,1731), rgba=(-nan(ind),-nan(ind),-nan(ind),1)
ignoring infinity/NaN image sample in pixel (1879,936), rgba=(-nan(ind),-nan(ind),-nan(ind),1)
no errors in log but warnings as follow. do any of these seem like a likely cause?
number of warnings, warning type:
00:42:52 52406MB | 1: [disp] %s: padded bounds could be %.3gx more efficient. given disp_padding: %.9g, recommended: %.9g
00:42:52 52406MB | 1: [disp] %s: padding is at least %.3gx smaller than it should be! given disp_padding: %.9g, recommended: %.9g
00:42:52 52406MB | 1: [ginstance] %s: %s needs to reference an object
00:42:52 52406MB | 1: [image] %s: invalid tag found
00:42:52 52406MB | 126: [mtoa.ext] %s Found no translator named "%s"
00:42:52 52406MB | 5: [mtoa.translator] %s: Maya node %s(%s) does not have attribute %s to match parameter %s on Arnold node %s(%s).
00:42:52 52406MB | 1: [rlm] lost connection to license server
00:42:52 52406MB | 930: cannot declare %s %s on %s - name already in use
00:42:52 52406MB | 2: ignoring infinity/NaN image sample in pixel (%d,%d), rgba=(%.9g,%.9g,%.9g,%.9g)
00:42:52 52406MB | 117: node "%s" is already installed
i've already fixed most of the bounds issues, but i assume this wont be a fix for the random arnold error pixels im getting.
is there something buggy about cylindrical area lights perhaps?
Blue is the NaN error color. Check your logs, there probably be some messages about that...
Would area lights intersecting with geo cause problems in arnold? I've reduced the error pixels but not eliminated them, by moving area lights(cylinder) around in my scene. I can't see anything in the logs mentioning the area light or the geo in question.. excepet for bounding box inefficiencies. (verbosity debug)