Sun with "physical lights"

[bkircher]

Is it just me, or is there a difficulty combining Sun-light with physical lights like IES or meshlights with powerdrawn/efficacy statements?

As soon as a Sun is in the Scene, the physical lights don't seem to lighten even the shadows much.

In earlier Versions of Indigo, one could simply crank up a gain parameter (or was it down?) and be done with it, but since we have all the great new physical lights, that has lost it's charme even if it's possible (didn't check, though).

In this scene, I've placed ~20 spotlights of "type 18" of the blendigo IES-Collection (I hope I recall that correctly). The Ball is a black phong and ~60cm or around that.


I'd love to check Indirect Lighting against Sun Light, so what am I missing

[dougal2]

I can see them... just behind the sphere.

There is a HUGE intensity difference between the sun and meshlights... they are difficult to balance.

Have you tried cranking the power_drawn up to some 1000s of W ?

[Whaat]

1) I don't think Efficacy scale works with IES lights (bug?)
2) I noticed that the brightness of IES lights seems to affected by the size of the mesh. Try having two identical IES lights of different sizes in the same render. The larger one will overpower the smaller one. I'm not sure if this was Ono's intent.
3) From my experience, normal meshlights with typical efficacy scales work fine in combination with sun and skylight.

I haven't done many tests with IES lights. Maybe someone else can confirm my statements.

[dougal2]

2) I noticed that the brightness of IES lights seems to affected by the size of the mesh.

this does indeed seem to be true.

I just made a quick test scene similar to bkircher's and using global scale factor on the scene to make the scene 100x smaller made the meshlights brighter.

[dougal2]

on further testing, it does appear that power_drawn and efficacy_scale are ignored with IES lights.

edit: it appears that blackbody Gain is ineffective too.

edit edit:
using

Code: Select all

<meshlight>
	<pos>0 0 0</pos>
	<scale>1</scale>

	<rotation><matrix>1 0 0 0 0 -1 0 1 0</matrix></rotation>
	<mesh_name>IESlocator|IESlight|IESlightShape</mesh_name>
	<spectrum>
		<blackbody>
			<temperature>6500</temperature>
			<gain>1000</gain>
		</blackbody>
	</spectrum>
	<ies_profile>
		<path>../ies/18.ies</path>
	</ies_profile>
	<efficacy_scale>
		<power_drawn>150</power_drawn>
		<overall_luminous_efficacy>20</overall_luminous_efficacy>
	</efficacy_scale>
</meshlight>

gives Light luminous flux: 600.00 lm

using

Code: Select all

<meshlight>
	<pos>0 0 0</pos>
	<scale>1</scale>

	<rotation><matrix>1 0 0 0 0 -1 0 1 0</matrix></rotation>
	<mesh_name>IESlocator|IESlight|IESlightShape</mesh_name>
	<spectrum>
		<blackbody>
			<temperature>6500</temperature>
			<gain>1000</gain>
		</blackbody>
	</spectrum>

	<efficacy_scale>
		<power_drawn>150</power_drawn>
		<overall_luminous_efficacy>20</overall_luminous_efficacy>
	</efficacy_scale>
</meshlight>

gives Light luminous flux: 3000.00 lm

[bkircher]

I just made a quick test scene similar to bkircher's and using global scale factor on the scene to make the scene 100x smaller made the meshlights brighter.

I think lights should get brighter as the scene get smaller, so the same luminous flux is applied to a much smaller area.

I also think we shouldn't resort to light using severall 1000Watts of energy, with global warning and all (sorry, couldn't resist)

I've made this second test with a 40 W light-bulb (I think that fits roughly), and with a 1m cube, I'd guess there should happen something inside.

2nd image is with the equivalent of a 28W fluorescent tube ...

If I wasn't that lazy this evening, I'd check with linear exposed images.

[dougal2]

in my last test this evening:

Code: Select all

<meshlight>
	<pos>0 0 0</pos>
	<scale>1</scale>

	<rotation><matrix>1 0 0 0 0 -1 0 1 0</matrix></rotation>
	<mesh_name>IESlocator1|IESlight|IESlightShape</mesh_name>
	<spectrum>
		<blackbody>
			<temperature>6500</temperature>
			<gain>20000</gain>
		</blackbody>
	</spectrum>
	<efficacy_scale>
		<power_drawn>3000</power_drawn>
		<overall_luminous_efficacy>200</overall_luminous_efficacy>
	</efficacy_scale>
</meshlight>

balances quite well with sunlight
(with reinhar tonemapping 2/1/6.5)

[bkircher]

Could it be that Indigo is 2 power of magnitudes off the mark as far as Sun brightness is concerned?

[dougal2]

hang on... is the cylinder's sun shadow visible ?

shouldn't be as it's masked by the box ??

[Whaat]

I think Indigo sun brightness is probably fine. The behaviour you are seeing is probably due to Reinhard tonemapping. If you were to try an interior scene with the sunlight coming in through a window, you would probably find that typical efficacy settings work as expected (but not with IES lights ).

[dougal2]

actually, there's no shadow problems... ignore that post.

[Kram1032]

the IES files already have included, how much light goes to which direction, so, the efficiency already is given in them (more or less, at least)

[CTZn]

So we have to calibrate the plane's size in MtI. Also the trick from 0.7 (MtIFull) was doing well, but my preference goes to manual tweaking.

Could it be that Indigo is 2 power of magnitudes off the mark as far as Sun brightness is concerned?

Maybe even more...

Code: Select all

Sky zenith luminance (cd/m^2): 10001.42853
Sun spectral radiance (Wsr^-1m^-3) at 500nm: 18024505540608.000000

from log.txt

[OnoSendai]

Hrmm ok, a lot of questions to answer.

" I don't think Efficacy scale works with IES lights (bug?) "

There is no need for the efficacy scale element if ies_profile is present. This is because the efficacy_scale just defines the total luminous flux emitted by the light, but the .ies file has that information already.
The efficacy_scale element is ignored if ies_profile is present.

"There is a HUGE intensity difference between the sun and meshlights... they are difficult to balance. "

Yes well it may be difficult, but I think it's correct.

"2) I noticed that the brightness of IES lights seems to affected by the size of the mesh. Try having two identical IES lights of different sizes in the same render. The larger one will overpower the smaller one. I'm not sure if this was Ono's intent.
"

Changing the size of an IES light has no effect on the total power emitted in my tests.

"Could it be that Indigo is 2 power of magnitudes off the mark as far as Sun brightness is concerned?"

It *could* be, but I think Indigo is correct. If you have actual comparisons with real life lighting situations or data, please feel free to share, otherwise you're just speculating.

[dougal2]

ok, that's fair enough.

does this explain why lights get brighter in a scaled down scene:

I think lights should get brighter as the scene get smaller, so the same luminous flux is applied to a much smaller area.

?