After watching the tutorial on fundamentals of lighting, I feel a much greater sense of clarity around how to select right light sources and set their strength. However, I feel stumped trying to setup the lights for a room like this where it's not clear what is illuminating the room. It's the original reference of image I'm using to build the 3D model of the room.
I wonder how the experts on this forum like @jlampel handle these situation
may sound simple but I would choose the proper kind of light source and then place it in the location where the light is...
such as if there is a ceiling light put it where it goes on the ceiling...
hope this helps
Thanks @marksmith ! I tried to place ceiling lights of different kinds and some HDRIs but nothing results in this flat saturated lighting. Here is the one created using photography softbox just below the camera in the scene.
I suspect the photographer had their camera flash turned on causing this flat lighting of the scene but I don't know how to simulate that correctly. Should I simply place a point light and turn up its power?
in that case I would try an area light in a position that the camera person would use...
Unfortunately there's no real way to objectively reverse engineer the brightness of a light source from a photo without tons more information. In this case I think it's save to say the light is coming from a window, so your thought to use an HDRI is a good one! To bring it to the right value, add a sun lamp that's about the brightness you'd like, match the HDRI to that, and then delete the sun lamp. To add a flash, add a small point light just above the camera like you already thought of. You can look up the value of a camera flash in Lumens and use that if you're using the addon and want to be precise.
Hope that helps! Sounds like you're already 98% of the way there.
Thanks @jlampel I have started to use your technique to set the strength of the HDRI. It's really clever!
On the topic of estimating the strength of the main light source and its direction
If we pick a paint PBR material for the wall with a known orientation and the camera location is known as well, shouldn't the BSDF model of the material let us roughly estimate the strength the light source?
Another way I am thinking is to compare the diffuse_direct + indirect maps with the reflectance map inferred from the image.
I wonder if someone here has already tried to create an add-on like that.
shouldn't the BSDF model of the material let us roughly estimate the strength the light source?
Yes! However, that would only be true if the exposure of the camera is known since a bright light with a higher exposure might look the same as a dim light with a low exposure, everything else being equal. You could also estimate the size and distance from the subject based on the shadows it casts, but to tie all that together would (I think) take some serious machine learning.