SunSky, HDR maps, area lights and gain values

[scl]

I am trying to come to grips with setting up proper/accurate/realistic lighting. I am setting the sun direction vector based on the correct sun position based on latitude/longitude and time of day. This works really well by itself.

When I add other lights (using area lights, IES date etc) to simulate things like street lights, are the gain values for the SunSky and area lights to be left as 1.0 in order to simulate the real lighting conditions? In other words, are we messing up the lighting if we start changing the gain values? What about the other parameters such as Importance?

Do similar consideration apply when we use HDR maps instead of SunSky? What about when we combine HDR Maps and SunSky with other discrete area lights?

Also, for area lights, if IES data is used, do the Power and Efficacy parameters need to be adjusted independently from the IES?

[jeanphi]

When I add other lights (using area lights, IES date etc) to simulate things like street lights, are the gain values for the SunSky and area lights to be left as 1.0 in order to simulate the real lighting conditions?

Yes.

In other words, are we messing up the lighting if we start changing the gain values?

Yes.

What about the other parameters such as Importance?

Importance only changes the sampling frequency, not the light intensity, so if you know a light contributes more to the overall lighting you can raise its importance to help reduce the noise faster. This requires the use of a lighting strategy taking advantage of the parameter.

Do similar consideration apply when we use HDR maps instead of SunSky?

It really depends on the HDR calibration. With an HDR, you should adjust the gain so that the it matches the desired intensity.

What about when we combine HDR Maps and SunSky with other discrete area lights?

Once the HDR gain has been adjusted, you can combine it with any light.

Also, for area lights, if IES data is used, do the Power and Efficacy parameters need to be adjusted independently from the IES?

Yes, IES properties defining the lamp power are not used.

Jeanphi

[scl]

Wow, that was quick! Thanks for the response. If my understanding is correct, best to leave the gain values at 1.0 if we have used correct SunSky and area lights. For Area lights, I have to make sure that I set the correct Power and Efficacy parameters that match the IES and light data.

Regarding HDR maps, is it correct to say that if I knew the latitude/longitude and time of day for when the HDR was created, I could leave the gain value at 1.0 and not model the SunSky at all? If so, is there any advantage in also modeling the SunSky in addition to such an HDR, where the position of the sun would also be properly placed (and hence, be the same as those for the HDR)?

What do you mean by adjusting the HDR gain to match the desired intensity? Are you talking about faking it so it looks bright enough? Also, when you say that once the HDR is adjusted, it can be combined with any other light, are the gain values of these other lights to be left as 1.0 if the goal is to simulate real lighting conditions?

Finally, assuming that all these lights were properly modeled and using gain values of 1.0, what other parameters in LuxRender would I need to consider for the purpose of recreating real lighting conditions? For instance, should I be concerned about Turbidity?

[SATtva]

Regarding HDR maps, is it correct to say that if I knew the latitude/longitude and time of day for when the HDR was created, I could leave the gain value at 1.0 and not model the SunSky at all? If so, is there any advantage in also modeling the SunSky in addition to such an HDR, where the position of the sun would also be properly placed (and hence, be the same as those for the HDR)?

95% of HDR images (regardless of their source) aren't properly calibrated, so most of the time you have to raise EnvMap's gain significantly to image be even noticeable. By significantly i mean values like 1000-10000.

It's not clear to me what you meant by "modelling the SunSky". Are you talking about using an envmap in conjunction with the simulated Sun (note: not SunSky, but Sun -- you have to disable the Sky component, otherwise the result will be wrong)? The advantage of this is you'll have sharp sun shadows, because most HDRs don't have the necessary dynamic range and sharpness, making shadows too soft. A method to calibrate HDR with the simulated Sun is described here.

For instance, should I be concerned about Turbidity?

Turbidity simulates atmosphere fog, affecting SunSky light scattering and wavelength. If you wish to depict a sky in the hot day or above big city, raising turbidity should make it more realistic. (Please note: turbidity has nothing to do with volumetric scattering, it just affects appearance of the SunSky and its illumination of the scene.)

[Qantourisc]

Another imporant factor: don't ferget to set the correct gamma for the HDRI.

[Meelis]

How can i match the sun and the sun on hdr position? How to i uv-mapp the lat-long hdr inside sphere, so i can tune sun position to correct place?

[SATtva]

I use environment camera and adjust Lux sun orientation until it's matched with the HDR background.

[Meelis]

Thx SATtva
This worked out well.