Hi All,
I would like to clarify some things about IES light intensity in Octane, it would be really appreciated if an Octane developer could comment. From my own tests, IES files plugged into the 'distribution' node of an emitter seem to normalize the candela values listed in the IES file to a range of 0.0 to 1.0. It makes no difference if the values listed in the distribution web, or the overall multiplier field, are 1, 1000, or 2000. In the rendering below, each box has the same geometry, the same emitter sphere (with general visibility set to 0), and the same material and emitter node, only distribution differs. Do I have this right? If so, then doing accurate lighting would require reading the lights maximum intensity from the IES file (in candelas), and entering that as the emitter's power.
Another question is about the units Octane uses for light intensity. The documentation for mesh emitters lists the power node a 'the wattage of the light source'. I think this should be actually be the radiant intensity of the light source, measured in watts per steradian. If it were actually just watts, then a light with a half-sphere distribution would necessarily have twice the intensity of a full sphere, but the test renders show the same level of illumination on the test boxes. Checking 'surface brightness' has no effect on this behavior.
I would like to see Octane move to photometric units instead of radiometric units. There's just too much extra work required to figure out the efficiency of a light source, and these days with the LED dominated market, it's easy to get absolute photometry for any luminaire. It's even more complex with coloured lighting, since the efficiency of a light source also depends on what part of the visible spectrum a light source covers.
Sorry if this all sounds pedantic but this stuff was confusing me for a long time, and I think there is potential for Octane to be a useful lighting simulation and design tool, on top of creating renderings (I'm an architect!). With custom camera response profiles or lookup tables, we could even produce false colour photometric studies for buildings instead of having to build separate models for AGI32 or Dialux...