Radiosity without radiosity II

[robcat2075]

As the light bounces off the walls it takes on the wall's color.

 

OmniTestEH.mov

 

A typical Cornell box image takes about an hour to render with radiosity, but this takes about 15 seconds per frame to render.

 

This scene also has something in it that technically doesn't exist in A:M. Can anyone guess what it is? There are clues in the render.

[JavierP]

As the light bounces off the walls it takes on the wall's color.

 

OmniTestEH.mov

 

A typical Cornell box image takes about an hour to render with radiosity, but this takes about 15 seconds per frame to render.

 

This scene also has something in it that technically doesn't exist in A:M. Can anyone guess what it is? There are clues in the render.

 

Nice! It looks like you're using a z-buffered bulb light ( or omni light as it is called elsewhere). The only thing that I can't see is some yellow light reflected off Thom. Still great work thought. This is definitely proof that with a little work AM can certainly do just about anything.

[itsjustme]

That is a fantastic test, Robert! I have no idea what the answer to your question is...I'm very curious.

[Eric2575]

Towards the end, the light source seems to be going through the back wall and illuminating the scene from the outside, casting shadows and all. It gives me the impression of a translucent wall.

 

Impression vs render times = fantastic!

 

Now you've gone and done it, I'm gonna have to pick up that book.

[jimd]

outstanding tests and research

[robcat2075]

thanks! The appearance of light spill is simulated with lights behind the walls, animated to change intensity as the light in the room moves about. I'm sure a real radiosity render would be more nuanced.

 

 

It looks like you're using a z-buffered bulb light ( or omni light as it is called elsewhere).

Ding, ding, ding ding! Yup, it's a shadow-mapped "bulb light" made out of an array of klieg lights. A square gel on each one keeps them from overlapping. It's only slightly faster than a 1-ray bulb light but much faster than a many-rayed bulb light that would be needed to do soft shadows.

 

 

 

 

Towards the end, the light source seems to be going through the back wall and illuminating the scene from the outside,

I think the unnaturally large hot spot from the light being so close to the wall does that. There's probably a way to not have the hot spot at all.

[Tralfaz]

Really well done Robert! Eric mentioned something about a book. Do you know which book he is referencing?

 

Al

[robcat2075]

Really well done Robert! Eric mentioned something about a book. Do you know which book he is referencing?

 

Al

 

Jeremy Birn's Digital Lighting and Rendering

 

I'm testing out ideas I get as I read it.

[thejobe]

im always testing new way to light stuff, but im a little to focus on realism so i need to broaden my techniques

[robcat2075]

A:M's AO is about occlusion from the sky dome which isn't quite the same as occlusion from indoor surroundings.

 

So here's a couple of attempts to do indoor ambient occlusion. The first is my omni directional light set to be almost as wide as the room and letting A:M jitter that with multipass

 

 

That averages out to the light originating from the center of the room which isn't right since the ambient light is really from the walls

 

This one has a box of splines crawling along the walls of the room. This is better but I need to find a way to make the light not shine on the wall it is closest to at any point. That is what's making the bright spots.

 

[Elm]

....

 

This one has a box of splines crawling along the walls of the room. This is better but I need to find a way to make the light not shine on the wall it is closest to at any point. That is what's making the bright spots.

 

I don't get it. Box of splines crawling along the walls of the room?!

[robcat2075]

....

 

This one has a box of splines crawling along the walls of the room. This is better but I need to find a way to make the light not shine on the wall it is closest to at any point. That is what's making the bright spots.

 

I don't get it. Box of splines crawling along the walls of the room?!

 

I drew a spline path just inside the walls. it made a zig-zaggy path across all six sides to simulate bounce light from the walls

[mouseman]

This one has a box of splines crawling along the walls of the room. This is better but I need to find a way to make the light not shine on the wall it is closest to at any point. That is what's making the bright spots.

It looks like the beginning of a kind of neat disco ball effect, however.

[robcat2075]

You can see the more correct effect in the bottom example of "indoor AO" in this post

[Elm]

...

 

I drew a spline path just inside the walls. it made a zig-zaggy path across all six sides to simulate bounce light from the walls

 

 

Do I get you right: A light travels along those paths in a multipass frame?

 

Greetz,

Elm.

[robcat2075]

...

 

I drew a spline path just inside the walls. it made a zig-zaggy path across all six sides to simulate bounce light from the walls

 

 

Do I get you right: A light travels along those paths in a multipass frame?

 

 

 

Yes. You can see a set up where a spline light simulates the sunlight reflecting off a spot on the floor here.

 

I used the lowest res Kleig light for speed.

 

Paths with straight lines are not really ideal, the best path would be a completely random squiggle that still evenly covered the area.

[Xtaz]

...

 

I drew a spline path just inside the walls. it made a zig-zaggy path across all six sides to simulate bounce light from the walls

 

 

Do I get you right: A light travels along those paths in a multipass frame?

 

 

 

Yes. You can see a set up where a spline light simulates the sunlight reflecting off a spot on the floor here.

 

I used the lowest res Kleig light for speed.

 

Paths with straight lines are not really ideal, the best path would be a completely random squiggle that still evenly covered the area.

It's moofuf's principle there ...

a long time ago I posted s similar image testing "fakeosity"...

 

[robcat2075]

It's moofuf's principle there ...

a long time ago I posted s similar image testing "fakeosity"...

 

 

that looks great. I remember that and at the time it sounded like you were using a rotating light of some sort. Was it a light on a spline?

[Xtaz]

yes ... on the windows ... same way that you post there.

[Elm]

All great, still non-comparable with the real deal, my opinion..

[Elm]

Hi Folks!

 

I fiddled around with soft reflections a bit to yet achieve some kind of another fake GI/radiosity effect:

 

First image contains a closed room, test character and a bulb light. No reflections at all:

 

 

Second Image - same setup, but everything is set to 70% reflectivity and specular size to max (10000). rendered with soft reflections, 1 level.:

 

 

Third image - same with 2 level reflections:

 

 

Fourth image - some cornell disco fun - no light at all, but the coloured walls turned to 100% ambiance intensity, not reflecting, set to glow. the rest keeps reflecting as above.:

 

 

 

You might try that yourself.

Greetings,

Elm.

[robcat2075]

That's quite convincing! When you say "everything" is set to reflective do you mean everything or just everything on the man?

[Elm]

That's quite convincing! When you say "everything" is set to reflective do you mean everything or just everything on the man?

 

 

I mean everything. the walls, the floor, the ceiling, the "man".

 

(as i said - the coloured walls in the last image are NOT reflecting. They are set to 100% ambient intensity, 0% reflection, glow ON. Glow settings in chor: radius 50, intensity 170.)

[NancyGormezano]

I will have to play with this! Excellent!

 

In the last image: I am guessing the shadow of the man on the floor is really a soft reflection? which would imply it would change based on the view angle?

[Elm]

I will have to play with this! Excellent!

 

In the last image: I am guessing the shadow of the man on the floor is really a soft reflection? which would imply it would change based on the view angle?

 

 

Uhmm ... as there is *no* light in the chor, it *has* to be the result of all the reflections that are casted in the scene. So I guess you're right.

 

 

Greetz,

Elm.

[robcat2075]

I can't figure that shadow out if there's no light there. It's not really in the right place for a reflection.

[Elm]

I can't figure that shadow out if there's no light there. It's not really in the right place for a reflection.

 

 

That's why I said "Uhmm"! It's definitely not in the right place.

I'm at home right now, but tomorrow I'll open the chor again and take a look from a different angle.

 

There's definitely no light in there. Maybe one more hidden A:M gem.

[NancyGormezano]

I did a quick test (16b-32, no multipass), 1 level soft reflections, glow - radius 50, %100. Only the colored walls are 100% ambiant, the white walls (back, floor) and the character are 70% reflective (10000 % spec size), 0 ambiant.

 

Yes there is a reflection on the floor in mine.

 

EDIT: in the 2nd image I decaled the colored walls with an image to see what it would look like. I suspect glow is not always needed. (it is still ON in 2nd image)

Edited August 15, 2013 by NancyGormezano

[Elm]

That looks great.

I like his shiny shoes!

 

Greetz,

Elm.

[NancyGormezano]

I like his shiny shoes!

 

Hmmm...I wondered about that and found out that I had accidently set the spec size on his shoes to 0% which made them look shiny. The cape was at 20% (also looked a bit shiny), whereas everything else was at 10000%.

 

Interesting since there weren't any lights in the chor with spec on. I guess soft reflections also uses the spec size. Strange.

[robcat2075]

I guess soft reflections also uses the spec size.

 

It does. Yves weighed in on that in post #23 in this thread...

 

http://www.hash.com/forums/index.php?showt...g,and,Rendering

[Rodney]

That is a great topic, Robert.

Thanks for the reminder that it is there.

Parts of the topic look familiar but other parts (particularly the page by page write up you did on Brin's "Digital Lighting" book) I don't recall reading.

This is odd because I recall thinking it'd be nice to take an A:M-centric look at Brin's projects/examples.

 

I need to break out that book and run through it again using your reference!

 

(I believe I was one of those that was ranting about the enhanced Null shader that you refer to at the end of that thread. As far as I know we still don't have that capability.)