What am I missing here? Why is directionality making more light?

[shelshok]

I've got a simple room here with a 5000 lumen 1m² rectangle light 2.5mm above the floor. I would expect that with it so close to the floor, falloff and spread would be negligible, and the lighting analysis should show about 5000 lux for the lit portion of the floor directly below. Here, I've rendered it with directionality set to 0.0 and 1.0, and get two different results (the 0.0 version is as expected, but the 1.0 version seems to be getting twice as much light on the floor, and the rest of the room is brighter overall from the reflected light).

Is this not incorrect? Where is the extra light coming from?

[^Lele^]

Because you're concentrating the light flux, analogously to what you'd do with a lens.
It's perfectly natural.

[shelshok]

Oh I understand that it is concentrating the light, and I would expect similar behavior if the light on the left was further from the surface, but when it's right up against the surface, the same amount of light should be hitting it, should it not?

As I understand it, a 1 m² 5,000 lumen light should measure at 5,000 lux, but in this example, it is measuring at 10,000 when directionality is 1.0 and regardless of how the light is focused, there is clearly more light in the room on the right.

[shelshok]

For a more controlled experiment, I made all the surfaces white and closed the box (cut with a section not affecting light) so no light would escape, and the room on the right is still clearly brighter, which suggests the 1.0 directional light is actually emitting more light, right?

Edit: skp file attached.

[^Lele^]

Oh, i hadn't noticed the part where there was more light overall, i beg your pardon.
Could you share the scene above? I get quite drastically different results than yours when testing.

[Joelaff]

Interesting tests. Definitely seems incorrect.

I always have to reduce intensity of lights as I make then more directional, and it always seemed like I had to reduce them too much. Maybe it really is too much.

[shelshok]

Oh, i hadn't noticed the part where there was more light overall, i beg your pardon.
Could you share the scene above? I get quite drastically different results than yours when testing.

Here you go. Attached skp.

[shelshok]

And here's just one more test where I rotated the lights to see the overall effect on the room when not right against the floor.

[shelshok]

And the scene with rotated lights.

[^Lele^]

I haven't hidden away, I'm just waiting for a test from the devs.
I am told the light flux *is* identical, so the issue seems to be with the testing approach.
I'll share more as i'll get it.

[piotrus3333]

Just different GI I think. Quick test gave me much less difference between 0 and 1 directionality. BF+BF, GI depth 100, max ray intensity off. Sphere instead of box. grey mat 18%, white 80%, black 4%.
do you have access to PPT rendering mode in sketchup?

[shelshok]

Just different GI I think. Quick test gave me much less difference between 0 and 1 directionality. BF+BF, GI depth 100, max ray intensity off. Sphere instead of box. grey mat 18%, white 80%, black 4%.

If I look at your images, it's clearly darker in the parts not in direct light (bottom of sphere, behind man), when those areas should be getting roughly the same amount of light bounced back from the directly lit surfaces.

GI settings aside, my first example still shows that the area of the floor directly lit by the 1.0 light is at 10,000 lux, when it should be no more than 5000, and since the light is right up against the floor, both rooms are being lit almost entirely by the bounced light from there.

^Lele^ any news on this?

[^Lele^]

^Lele^ any news on this?

Nope, Vlado got busy, he never produced the test.
He was quite adamant, though, on it being exactly the same amount of emitted light.
I haven't forgotten, i just can't quite tackle him at the knees. XD