[Kabe]

@Kabe: total internal reflection is a case were light can't get out of the material anymore, this is what happens inside optic fibers to transport light.

I know what it is, but the term is a mistake nonetheless - well, it is in all other languages and used to be one in English, too -, though it seems that "internal reflection" has won the in terms of web dominance :-/

It is misleading because this is not about reflection but about the fresnel terms, which is a tad more complicated. I know a tad more than the schoolbook explanation about this, but this would lead much too far here.

So enough with this nitpicking...

My point is:
Test and compare M~R against standard apps with a solid renderer for such stuff before claiming that something is broken. Images of dielectrics depend strongly from the environment, so tests should run against proven software or even better, against real scenes - like it has been daon in this forums some months ago...

Cheers

Kabe

[Frances]

My point is:
Test and compare M~R against standard apps with a solid renderer for such stuff before claiming that something is broken. Images of dielectrics depend strongly from the environment, so tests should run against proven software or even better, against real scenes - like it has been daon in this forums some months ago...

Cheers

Kabe

I agree. This is like looking for a brain tumor in a dental xray.

[max3d]

WinOsi is like maxwell an unbiased renderer, but not a commercial one. We have however done lots of test with this scene on vray, max scanline, viz radiosity etc. Could publish these data if you are interested in doing it in maxwell.

I rendered the reference scene in Maxwell 1.2.2a and this is the result (I also sent it to the WinOSI author); I hope this can help comparison with the 1.0 release...

Thanks Petrucci. Do you recall the time you needed to do this in the beta? BTW youre light seems to leak.
This is f.i. instance what i got out of Viz in 20minutes.


The radiosity was quite extreme and completely unrealistic as compared to maxwell's. And of course no caustics so the 20 minutes was just a waste of time.

Maxwell renders a black front on the ball which is not in the reference and the light reflection on the left wall is much larger.

[Frances]

The radiosity was quite extreme and completely unrealistic as compared to maxwell's. And of course no caustics so the 20 minutes was just a waste of time.

Maxwell renders a black front on the ball which is not in the reference and the light reflection on the left was is much larger.

What were your radiosity settings? It looks like they were very coarse.

Since Maxwell doesn't support camera clipping, that black bit on the ball is the scene background, which is black. Plus, taking away a wall will not give you an accurate GI simulation.

[Mihai]

@Kabe: total internal reflection is a case were light can't get out of the material anymore, this is what happens inside optic fibers to transport light.

I know what it is, but the term is a mistake nonetheless - well, it is in all other languages and used to be one in English, too -, though it seems that "internal reflection" has won the in terms of web dominance :-/

I don't understand Kabe, in that pic, the red ray represents refraction, then what does the green ray represent? If the ray stays within the same medium and is bounced at the border between two mediums, then it is a reflection. The ray is not bent.

[tom]

Sorry, I needed to make a clarification that it is scientifically "reflection" and
this is nothing about popular knowledge. It's simple, if a ray stays in the
same medium after hitting a boundary, it's always "reflection" and if it passes
through a boundary with the index greater than 1.0, it's refraction. If the
index is 1.0 then this means there is no boundary. You just can say that
the internal reflections are made of refracted beams

[Kabe]

My intention was not to fight about words... please let's go back to the render results for dieelectrics...

Kabe

[Micha]

Tom, an other question: do we get independent roughness parameter for brush effects? Nobody speak about this simple feature, but for design stuff it is very needful. Is this Bug fixed or was it a unfinished feature and is not ready now?

[j_petrucci]

[@max3d:] unfortunately I did this scene a bit ago, and I cannot recall the exact rendering time, but it was something about 3~4 hours.
what are you referring to when talking about "light reflection on the left wall"? the room is approximately 16 m^2 and the light is about 200 W, all the walls are diffuse material, so the diffused light cannot be any more than the one you see in Maxwell rendering.

[@frances:] yep, not only camera clipping, but also backfacing polygon exclusion from rendering is unsupported; however, if we think about it, this is acceptable in terms of "reality", just because as we say in Viz's rendering, the reflection of an invisible object is totally impossible!

[max3d]

Frances, I did my test serie with brazil, vray, viz, max etc. all two years ago so i can't recall specific radiosity settings. Brazil and vray did a reasonable GI job without the missing wall.

Hi petrucci, with the reflection on the wall I mean the small bright spot in the 100 hours winosi image in the middle of the left wall. I couldn't reproduce it with any other renderer.

[Frances]

Frances, I did my test serie with brazil, vray, viz, max etc. all two years ago so i can't recall specific radiosity settings. Brazil and vray did a reasonable GI job without the missing wall.

Gee, that's an entire wall of bounced light missing. Something that looks reasonable doesn't necessarily consititute an accurate simulation.

[Thomas An.]

Frances, I did my test serie with brazil, vray, viz, max etc. all two years ago so i can't recall specific radiosity settings. Brazil and vray did a reasonable GI job without the missing wall.

Gee, that's an entire wall of bounced light missing. Something that looks reasonable doesn't necessarily consititute an accurate simulation.

The thing is that in his example we can still see the wall in the reflection on the sphere as well as the green color bleeds in to the scene. Somehow the wall seems clipped only from the camera but not from the render.

[rivoli]

Somehow the wall seems clipped only from the camera but not from the render.

well, yes, that's something you can do with any renderer (but maxwell). i mean to hide an object from camera (or reflections/refractions, gi and so on) but still having it in scene.
i think that's exactly fran's point, with maxwell you can't do that, so you have to remove the wall from your scene, so that the radiosity simulation is totally messed up. how can you have an accurate simulation of a closed room, with light bouncing against four walls, when this room has only three? light would behave in a totally different way.

[Thomas An.]

Somehow the wall seems clipped only from the camera but not from the render.

well, yes, that's something you can do with any renderer (but maxwell). i mean to hide an object from camera (or reflections/refractions, gi and so on) but still having it in scene.
i think that's exactly fran's point, with maxwell you can't do that, so you have to remove the wall from your scene, so that the radiosity simulation is totally messed up. how can you have an accurate simulation of a closed room, with light bouncing against four walls, when this room has only three? light would behave in a totally different way.

You are right. I missed that point.

Sorry Frances

[Gijs]

@Kabe: I do not want to fight about words either. My point is that, although the pictures are small, there are quite obvious differences (beside the noise). Secondly, I did not claim that anything is broken, it was just a question