Page 1 of 1

skyportals

Posted: Fri Jun 01, 2012 7:39 pm
by PaZ
physicl sky's light is much better than dome, or emitter at windows, but it's slower since rays are not optimized to "hit" windows.
Cant we have special emitters which simply "contrain" physical skylight into windows ?
Pratically it's like having emitters which emit a light with color and intensity caught in normal's opposite emisphere.

Paolo

Re: skyportals

Posted: Mon Jun 04, 2012 12:16 am
by Rickyx
In an old post this wish has been expressed.
The problems of using portals are that:
user can misplace them and cause crappy results
problems during animation

The focus on other kind of optimization were chosen.

Ciao,
Rickyx

Re: skyportals

Posted: Mon Jun 04, 2012 1:20 am
by PaZ
...but evidently that focus is not enough since emitters at windows renders sooo much faster than physical sky.
So, or we can get an optimized P.S. without portals (imho technically impossible), or portals are still the solution - like it happens in most engines.
Let's have portals with caveats, instead of nothing ;)

Paolo

Re: skyportals

Posted: Mon Jun 04, 2012 11:56 am
by Fernando Tella
PaZ wrote:Let's have portals with caveats, instead of nothing ;)

Paolo
+1

I think Next Limit argued that skyportals would bias the light solution. IMO it is as biased as IES, bump or noise in the image, but I don't guide the Maxwell boat, so...

Re: skyportals

Posted: Mon Jun 04, 2012 12:49 pm
by numerobis
+1 :D

Re: skyportals

Posted: Mon Jun 04, 2012 3:29 pm
by PaZ
well. I was a fryrender user before, and i stepped to Maxwell because of many reasons.
One of these is they really couldnd understand that users dont care about 100% pure unbiasedness, but care a lot about usability.
I had hard times in leading them to isolate caustics - "they are a coproduct of illumination! they are just there !". Ok, but if they make my render 300% slower and most of the times you dont even see them, what's the point ?
These engines are not meant to be scientific simulator only, but moslty production tools. As other said, AGS, bump, thinSSS are all out of goal, under this POV. So, if portals just make interior look better in a short time (because yo ucant compare physical light with a dome), personally i cant get the unbiasedness reason as valid.
Just my 2 cents ;)

Paolo

Re: skyportals

Posted: Mon Jun 04, 2012 4:43 pm
by numerobis
PaZ wrote: ...
One of these is they really couldnd understand that users dont care about 100% pure unbiasedness, but care a lot about usability.
...
er... did you follow the discussions concerning this topic in THIS forum?!?
not much different i think... :mrgreen:

Re: skyportals

Posted: Mon Jun 04, 2012 4:53 pm
by zdeno
useless fight ...

looks like there is something in deep core of maxwellrender that makes it impossible to use skyportal.
because if not - lack of this kind of feature is just shot in own foot.

hidding behind physical correctness is just funny when we have "hidden to GI" "hidden for camera" "hidden for reflection refraction" solutions

but if skyportals are possible suddenly then +1 from me too ;)

Re: skyportals

Posted: Mon Jun 04, 2012 5:24 pm
by Mihai
PaZ wrote:physicl sky's light is much better than dome, or emitter at windows, but it's slower since rays are not optimized to "hit" windows.
I don't think Maxwell is so stupid that it doesn't optimize this already though. It's not about putting these sky portals telling Maxwell where the rays are coming from which it would know from the very first SL anyway, it's more a matter of mixing very strong emitters (sun in this case) with more normal ones. Please try a test with Physical Sky but Sun turned off, then with Sun on and check the noise levels at a given SL. Before, Maxwell was optimizing the strongest emitters first, while the weaker ones showed more noise. Now it's more balanced which is why you see sunlight clear a bit later than before.