Cornell Box

Our forum gallery for displaying your finished renders.

Moderator: coordinators

Forum rules
Don't forget to upload your finished work to our gallery too.
http://www.luxrender.net/gallery

Re: Cornell Box

Postby ingenious » Wed Feb 29, 2012 10:48 am

Hah, wait a second. Are you saying that the image in the very first post has been rendered using a portal that doesn't cover the whole region through which light can enter the scene? Depending on how LuxRender handles portals, the image might be totally wrong. As far as I see from the source code, if there is a portal, lights will always only sample the portal, resulting in a biased image in this case.
ingenious
 
Posts: 60
Joined: Tue Mar 16, 2010 3:43 pm

Re: Cornell Box

Postby A-man » Wed Feb 29, 2012 2:46 pm

No, I don't think you it's get it. I thought you put portals where you DON'T want light, hence, most of the light sampled the outside of the roof of my Cornell box :(
i7 3770k 3.5 GHz | 16 GB RAM | Radeon 7970 3GB
User avatar
A-man
 
Posts: 720
Joined: Sat Apr 24, 2010 4:24 pm

Re: Cornell Box

Postby ingenious » Wed Feb 29, 2012 4:08 pm

A-man wrote:No, I don't think you it's get it. I thought you put portals where you DON'T want light, hence, most of the light sampled the outside of the roof of my Cornell box :(


This I get. My worry is that light may not be sampled at all through regions not covered by portals. It seems to me from the light source implementation that you have two options:

1) You don't have portals in the scene -> light is emitted from the light source in the whole environment
2) You have at least one portal in the scene -> light is emitted from the light source only towards the area covered by the portal

If I'm not mistaken, you have case 2), so light is not sampled at all through most of the roof opening. Hence, the rendering is incorrect. If the portal covers the whole roof, this wouldn't be a problem, and the only problem you'd get is inefficient rendering. If the portal doesn't cover all regions through which light can enter the box, then you have a real problem.
ingenious
 
Posts: 60
Joined: Tue Mar 16, 2010 3:43 pm

Re: Cornell Box

Postby slime » Wed Feb 29, 2012 7:01 pm

Image

Image
slime
 
Posts: 45
Joined: Thu Nov 03, 2011 12:28 pm

Re: Cornell Box

Postby A-man » Wed Feb 29, 2012 8:00 pm

I'm so sorry about that guys. I've modified the scene and am rendering it, I'll upload the blend/lxs files if anyone wants, but don't feel obligated to help :p I'm also increasing the resolution of the water mesh because I can see edges in high-res rendering.
i7 3770k 3.5 GHz | 16 GB RAM | Radeon 7970 3GB
User avatar
A-man
 
Posts: 720
Joined: Sat Apr 24, 2010 4:24 pm

Re: Cornell Box

Postby slime » Wed Feb 29, 2012 8:30 pm

Lol, yeah upload the new ones we'll start again.


Sprocket is going to cry when he's seen this


A-man wrote:Most of the light sampled the outside of the roof of my Cornell box

slime
 
Posts: 45
Joined: Thu Nov 03, 2011 12:28 pm

Re: Cornell Box

Postby sprocket » Thu Mar 01, 2012 12:05 am

slime wrote:Lol, yeah upload the new ones we'll start again.


Sprocket is going to cry when he's seen this


I would cry.. but I did a little experiment yesterday with portals... I edited the portal so its just filling in the window, normal pointing inwards. then rendered that vs the original (without internal objects or scattering....

so either I got it totally wrong, or it still ends up the same??
Untitled-3.jpg


A-man wrote: I'm also increasing the resolution of the water mesh because I can see edges in high-res rendering.


buggar!
User avatar
sprocket
 
Posts: 340
Joined: Sun Jan 03, 2010 7:59 am
Location: Australia

Re: Cornell Box

Postby neo2068 » Fri Mar 02, 2012 6:42 am

There's another thing with this scene. The short box has double vertices and, IMHO, the water shouldn't be in the glass boxes because this interior interface won't be handled properly.
i7 860, 16 GB RAM, NVIDIA Geforce GTX 560 + GTX 460, Windows 7 64bit, Blender 2.70
neo2068
Developer
 
Posts: 559
Joined: Sun May 03, 2009 2:11 am
Location: Germany

Re: Cornell Box

Postby sprocket » Fri Mar 02, 2012 10:51 pm

neo2068 wrote:There's another thing with this scene. The short box has double vertices and, IMHO, the water shouldn't be in the glass boxes because this interior interface won't be handled properly.


yes ive run into that before. should be a glass <-> water interface, but the cube is a glass <-> air interface. Dont know how hard it would be to work the water surface mesh. I know using a fluid sim was a pain because i did it in this one...viewtopic.php?f=15&t=5325&
User avatar
sprocket
 
Posts: 340
Joined: Sun Jan 03, 2010 7:59 am
Location: Australia

Re: Cornell Box

Postby neo2068 » Sat Mar 03, 2012 3:01 am

sprocket wrote:
neo2068 wrote:There's another thing with this scene. The short box has double vertices and, IMHO, the water shouldn't be in the glass boxes because this interior interface won't be handled properly.


yes ive run into that before. should be a glass <-> water interface, but the cube is a glass <-> air interface. Dont know how hard it would be to work the water surface mesh. I know using a fluid sim was a pain because i did it in this one...viewtopic.php?f=15&t=5325&


I think, I didn't describe it correctly. I the original scene the water surface is inside the glass box.

screenshot.jpg

I tried to solve that problem with the new boolean modifier and it worked quite well. The best way would be that you apply the boolean modifier on a simple plane before you subdivide it and use the displace modifier with the cloud texture.
screenshot2.jpg
i7 860, 16 GB RAM, NVIDIA Geforce GTX 560 + GTX 460, Windows 7 64bit, Blender 2.70
neo2068
Developer
 
Posts: 559
Joined: Sun May 03, 2009 2:11 am
Location: Germany

PreviousNext

Return to Finished Work

Who is online

Users browsing this forum: No registered users and 1 guest