View Full Version : Portal Light Vs Physical Light.
wizlon 11-20-2007, 03:54 PM Here is a test, portal light vs physical light, The arealight on the left of image 1 is rendered with a Portal light and the arealight on the left of image 2 is rendered with a physical light.
Which is physically correct?
The portal light, appears to be the correct one, whereas the physical lights reflection looks additive and incorrect. That said the portal lights refection appears slightly too weak.
Both are driven by a mib_cie_d - temp 4000.
| |
Sphere♥
11-21-2007, 06:39 PM
Many factors could be at play here, your color space (gamma), your materials, scene size, etc. But it looks like the physical light's falloff is good but also happens to be brighter than the other; color temperature is one thing, what about intensity?
floze
11-21-2007, 08:11 PM
The physical_light has a cosine attribute which might be the reason for this, because it weakens the intensity towards the opening angle. Try turning it down to like 0.001 (0.0 would equal to 1.0 I guess.. that's how the shader is built), and see if that would get it closer to the portal light behavior.
wizlon
11-22-2007, 01:06 PM
Thanks both, looks like Physical lights reflections are not correct.
Here's a response from Master Zap :
"The portal is the "correct" one under the assumption you are trying to emulate a fixed luminance surface (i.e. with a diffuse per-point emission profile)
I think the "visible" luminance (i.e. how bright it looks to camera and reflections) of the "physical light" is approximated, which here ends up brighter. The "physical" light use a traditional distance-squared (from *center*) as it's intensity falloff.
Whereas the portal actually base it's entire calculation on the visible luminance of the surface, and using a proper integral over the subtended solid angle to figure out the illuminance from it, i.e. much more accurate for the near-field case. (Traditional cost-theta-distance-squared is ok for things at a distance from the light, not "near" it, with "near" being stuff like touching the edges of the rectangle etc.... basically anything closer to the light than the light diagonal)"
And some further tests and thoughts.
What I'am really now missing from the Portal Light is the ability to change the Cosine Exponent as with the Physical Light, this would be a huge benefit and offer the greatest control when using portal lights as light cards.
Would this be something someone (Zap/Mental Images) could consider adding?
I guess the Cosine Exponent in this sense simulates barn doors as used in traditional photography.
wizlon
11-22-2007, 03:43 PM
http://mymentalray.com/forum/showthread.php?t=1128
CGTalk Moderation
11-22-2007, 03:43 PM
This thread has been automatically closed as it remained inactive for 12 months. If you wish to continue the discussion, please create a new thread in the appropriate forum.
vBulletin v3.0.5, Copyright ©2000-2009, Jelsoft Enterprises Ltd.