Radiosity in interiors

[tica]

I've rendered a few interiors with radiosity (monte carlo), using a distant light for the sunlight and some area lights to add illumination inside the rooms.
This has produced some nice results in some scenes, but there are others where the colors sort of "spill" all over the ceiling. In one scene, for example, a blue carpet produces a strong blue effect on the ceiling. In another situation, a terracotta flooring produces a brown "reflection" on the ceiling, as well.
Has anyone had the same problem? Is there a solution for this?
What really bothers me is that some rooms produce good results, while others - with the same flooring, for example - get really weird looks.
I've tried to change a lot of settings, but nothing seems to make a difference... I get those blotches no matter what I change.
Thanks in advance for your suggestions.

[tektonik]

did you try lowering the specularity of the surfaces especially your blue carpet?

[tica]

Tektonik, thanks for your suggestion. The carpet doesn't have any specularity though, as carpets usually are matte. So it has no reflection, luminosity, etc. It's just a plain blue carpet, really... I've tried to change diffusion settings, to no avail. I've also tried to change the settings within the lights panel, but nothing changes dramatically.

[ingo]

Is your raydiosity intensity above 100 % ?

[tica]

No, Ingo, it's not. Should it be? I'm going to try it out, anyway... I've tried so many things by now, I guess it can't hurt :o

[robk]

Lightscape had a setting for the material where you could control how much color bleed/reflected light you would get off a material. Does anyone know if there is such a setting in Lightwave. If so that should solve your problem.

[otacon]

How many bounces are you using? Ive seen too much color bleeding when i do tests with over 2 bounces...especially if the blue carpet area takes up a lot of space.

[tica]

otacon, just 1 bounce...
robk, thanx for your input. that is precisely the problem - i don't think there is such a setting in lw. or is there?
it seems obvious that lowering the radiosity to around 50% (instead of 100%, which is the default setting) almost solves the problem. but then the light becomes very "hard" on objects, and the result is much less interesting than with radiosity.

[ddho1981]

I think it would be a lot easier to help you if you posted some images to reference. As it is, myself and others only have what we imagine you to be seeing making it more difficult to pinpoint what the issue may be. Post an image with the problem you are describing and maybe some of your radiosity settings.

[robk]

I think this would be an example of what tica means. The green felt from the snooker table casting a green tint to the ceiling.

[Lude]

Lower the diffuse level of the table green from 100% or what ever it is to say 50% and see what it looks like.

[tica]

Lower the diffuse level of the table green from 100% or what ever it is to say 50% and see what it looks like.

This example with the green tint is exactly what happens in my project (my client does not allow any images to be shown, which is why I couldn't post any examples).
I had initially tried to change diffusion levels, but then the surfaces became too dark - though it helped with the color bleeding problem.
Thing is, there seems to be no real solution for this situation.
There seem to be partial solutions, but one always has to sacrifice something in order to get rid of the color blotches - you have to set the radiosity level lower, or you have to lower diffusion, or whatever.
A setting like the one described from Lightscape would be excellent! But I'm increasingly convinced that Newtek is not very interested in the architectural visualization market - I guess it is a lot less glamourous than making movies in Hollywood.
Anyway, I went for the lower radiosity setting. 80% was the compromise level between a not-so-soft shading, and minimizing the color bleeding.
BUT this is not over. I'm still looking for better ways out of this.
Thanks everyone so far for your input!

[otacon]

Have you tried maybe using interpolated? It might not be as strong of a color bleed....just takes some more practice to get the settings right.

[Haven1000]

Why not do a spot of post-production editing? = Photoshop

[gerardstrada]

You can separate the problematic surface in another layer and exclude it of the radiosity calculation. So that the surface maintains the same look, (before) you might to bake the illumination in that surface or to use some occlusion shader and bounce lights (or gradients) to simulate the same look that it had with radiosity.



Gerardo

[tica]

Gerardo, thank you for your suggestion. Sounds like a good idea!

[gerardstrada]

I hope that it works, Tica :) however it is a weird phenomenon that we should try to understand yet.



Gerardo

[tica]

Yes, Gerardo, I usually prefer to understand why things happen, rather than open forum threads in order to find workarounds and quick fix-its... But in my experience (I'm in this business for almost 12 years now, which is a long time...) I'm finding that it is becoming increasingly difficult to find straightforward solutions for most of the problems, maybe because software is becoming more sophisticated. This doesn't make sense, as one might think that evolution means more control for the user over his/her tools - but everyday I notice that software is getting further out of control. This must be what they call growing old... :( Anyway, I'm always glad there are so many Lightwavers out there who are ready to help with their experience whenever someone needs advice!
And if someone at Newtek is watching, please don't forget that there are lots of users who do arch viz! This is not an area you should neglect... Not everyone can be in Hollywood making movies

[gerardstrada]

Oh! don't misunderstand me, my scientific curiosity is just for fun :) Maybe because I only have 11 years using LightWave. :D
Btw, I've noticed that the phenomenon can be caused by the ambient intensity (this adds certain luminosity feature in surfaces) for that reason the effect.
if this is the problem, to solve it you can:
1. not to use ambient intensity (0%)
2. If you want to use it you can reduce the diffuse of your surfaces (50% for example) and to increase the intensity of radiosity (200% for example); maintaining the same value in ambient intensity, this will almost make to disappear the effect that you try to avoid and will maintain the same initial illumination look.



Gerardo

[Lude]

This has me thinking, is there a quick way to link the diffuse level of everything in a scene so you could reduce or increase them all by a percentage of there current values at the same time?

[Exception]

Um, that color bleeding... that is what radiosity is... it is physically correct that it should be doing this... or am I missing something here?

That ceiling is green because light bounces off the pool table, and becomes green. It will be like that in reality.

[gerardstrada]

I think that you are right Exception, (is what I like of Radiosity) however when you use Ambient_Intensity is as if you added luminosity to the surfaces and a rebound of light that it should be very subtle, is too notorious and it doesn't look realistic.
Lude, about adjusting the diffuse percentage of all surfaces at the same time, we can add a Master_Channel and to apply it to a Slider, then for all diffuse channels, we tie each percentage (Follower, SetDrivenKey/in GE) to the slider. This way when we diminish or increase the percentage in the slider, will increase or will diminish in the same proportion all the diffuse values in all surfaces.



Gerardo

[Muad'dib]

It appears there is a simple way to deal with these problematic bleeding surfaces (no pun intended) ... all in LW without post and on a surface/layer basis with a final look that is almost correct :)


http://paulatreides.homestead.com/files/t1.jpg
Monte Carlo 2 bounces / No tweeking - Bleeding on ceiling


http://paulatreides.homestead.com/files/t2.jpg
Monte Carlo 2 bounces / With tweeking - Less Bleeding on ceiling but notice shadow density under yellow ball and reflection colour


http://paulatreides.homestead.com/files/t3.jpg
Monte Carlo 2 bounces / With different tweeking - Less Bleeding on ceiling / correct shadow density under yellow ball and almost correct reflection colour
The green now also bleeds less onto the balls as well.


Apologies on the crappy renders. I am at work atm but thought I'd lend a hand.
Will post tweek as soon as I get home ;)

[Muad'dib]

... apologies for it being a day or two later than expected. Here it is ... enjoy;)

http://vbulletin.newtek.com/showthread.php?t=32819

[lwnewbie]

I know this is a stupid question, but what is bounce?

[robk]

Bounce is how many times the light from the light source bounces from one surface to another. i.e. light from ceiling shines on floor and bounces to wall which bounces back to ceiling and so on and so on. The more bounces the better (more accurate?) the radiosity solution.

[robk]

Of course some of the light energy is lost with each bounce. Also the more bounces the longer the rendering time.

[lwnewbie]

Thanx for the reply. Is the bounce set by RPE?