[Aldaryn]

I didnt wanted to post this pic, until its finished, but I'm stuck with a strange problem.
Take a look at this image, it was a light setup testrender 20min, and it looks exactly like I've planned. Took some time to set the sun's position right, but I've got a decent ray:

This test was dome just before I've finished the door, and suddenly, when I was done with the geometry, my direct light through the door disappeared.


The diffrence between the two images is only geometrical, and all I did was replacing the crappy proxy door geometry with highly detailed ones. No open edges, no inversed normals, and exactly the same shaders in both renderings. And, after lvl 20, still no visible direct light.
Also, if I remove the tainted yellow glass, scene gets back to reality:


Any thoughts on this one?

Ps.: sorry bout the image host, but imageshack seems to be down for some reason.

[Aldaryn]

...meanwhile I've tried with different glasses, and still "no direct light". I really don't know where did I screw this up...

[rivoli]

there's a known problem with direct sunlight going through glass (i thought you knew it, you even posted in one of those glass related thread:

http://www.maxwellrender.com/forum/view ... ss&start=0

can't be the same problem you're having here?

[Aldaryn]

Yup, I know there's somthing bugging bout the scene scale and caustics, but I've checked, and there's really no difference in that way between the two images. In the first one, the caustics rendered perfectly. Only modification I did to the glass shader was to remove the bump, but with or without bumpmapping, it looks the same.

[rivoli]

ah ok, i didn't get there was glass there even in the first image, sorry for the silly post then.

can you post a wire of your glass geometry?

edit:
i think i can spot some direct caustic there in the second image, almost invisible, but there is something. both on the rug and the wall.

[tom]

aldaryn, do you have a big plane as a base?

[Aldaryn]

Here are the wireframes:
The whole scene
door and glass geometry

As you can see, there's no big ground plane, and the geometry seems fine to me. The glass is just an extruded shape for now.

[andronikos916]

in case you have not.... checked:

scene scale
caustics
glass thickness

cy,
Andron

[Hervé]

looks very cool and promising Aldaryn, but I have no idea.... other that what's being already said....

I was thinking about glass thickness also... try 1 mm, just to see...

[Mihai]

In your initial image, besides the big glass panel, did you also have these little things:



If not, try removing them, but keep the big glass panel, see what happens....

[Aldaryn]

Sorry Mihai, but for some reason I cant access imageshack. I really dont know why. My browser returns with a "not found" error message for www.imageshack.us links. Another bug in my system ...

This is a really annoying issue, couse I can't view almost all of the gallery uploads here.

[Mihai]

But can you browse to the main site itself? www.imageshack.us

Uploaded it again:

[Aldaryn]

Thanks Mihai!

Imageshack is now OK again. I don't really know what happened, but several friend of my cvonfirmed that imageshack was unavailable from hungarian IP-s for some time. Strange...

With, or without those small disks there is no visible direct light caustics, however:
I've replaced the geometry with one, standrad box primitive, and my caustics came back! But when I cut those little holes into the mesh, even, when its carefully modelled, the caustics disappear.

So, I guass Adam you're right, but I really don't know why a couple of triangles a promblem for MW, I mean, the geometry itself, along with proper reflection renders perfectly, so, why teher's no caustics?

[tanguy]

So, i ll work on my windows library, and i had caustic trough the glass, and after an action, i don't know wich one, no caustic calculate, or, display in my render......
It seem depend on a shuuter speed, fstop setup, in my case......
So, good luck, it s the best way for the pionneer !!!lol
Test and test and Test.....

[iandavis]

It appears that glass both casts a dark shadow and blurs objects seen through them... I would suggest removing the glass altogether (just repeat 10 times "it's a beta.. it's a beta")

to simulate warm light try using a pure yellow MXI emitter, though this may increase noise and not produce quite as sharp a shadow, though if you keep it small it should be ok.

Short of that you could always just tint the hilights warm in photoshop. (of the glass-less render)

of course this way you should have shorter render times, less noise and LIGHT on the wall.

cheers

ian