I recently read a book on CGI lighting techniques where the author discussed
dome arrays.  I wrote a short macro that creates the same thing by creating
a hemisphere of point lights.  The intensity of the lights varies with
latitude, creating a hotspot at the top of the hemisphere.  I also had it
add in a few negative lights to give some subtle variance in the lighting.
Anyhow, the results were pretty nice... very soft shadows with a radiosity
lit look.  Rendered fairly fast too.

--
Doug Eichenberg
www.getinfo.net/douge
dou### [at] nlsnet

Post a reply to this message

Attachments:
Download 'domearraytest.jpg' (59 KB)

Preview of image 'domearraytest.jpg'

Nice! I've tried something similar. Maybe you'll agree that actually the
light sources can be very far apart without noticeable problems. Diffuse and
brilliance are the only ones that benefit from using "real" area lights, but
since they are quite diffuse (brilliance too) the effect doesn't need a
dense grid of lights. The shadows are more demanding, so for smooth results,
one better mix in some area_lights.

I would be interested in knowing more about the strengths of light on the
sky. Did the book have anything on that? You know.. the sun is very strong,
but the sky is also strong near the sun. I wonder what's the ratio from the
sun position and 180 degrees away. No doubt it varies with fog and time of
the day.

Negative lights..hmm.. are those really a good idea?

Regards,
Hugo

Post a reply to this message

The negative lights only look good if they are low intensity and the array
has a lot of lights in it... then they work kind of like a shadow projector,
breaking things up a bit.

The way he described in the book, the key light is usually separate from the
dome array, and the dome array is used to simulate radiosity (if you want to
simulate skylight, anyway).  The intensity of the light is strongest in the
area opposite to the sun's position, and decreases as you get closer to the
sun.  The light at the top of the dome should be closest to white light,
with just a hint of blue.  Lights in the middle levels of the dome will be
blue or gray light, dependent on cloud cover.  Also, the sunnier the sky,
the paler the blue. The bottom of the array is used for bounced light
(radiosity), so the lights there will all be different colors, but the
intensities will be similar to the other lights.

He didn't go into much detail on using dome arrays for indoor lighting, as I
did in my test though.  I'll post the code for my macro in pbsf if you wanna
mess around with it.

--
Doug Eichenberg
www.getinfo.net/douge
dou### [at] nlsnet

Post a reply to this message

> The intensity of the light is strongest in the area opposite to
> the sun's position, and decreases as you get closer to the sun.
> The light at the top of the dome should be closest to white light,
> with just a hint of blue.

This is new to me. I believe you, but none of it seems logical. I would
think, the light looses intensity at the opposite sun direction. I would
also think the light at the top of the dome should be deep blue, unless it's
cloudy. And, the sunnier the sky, the deeper the blue. You're saying all the
opposite is true.

It's what I see in real life... Can't trust my eyes?

Thanks for the input...

Regards,
Hugo

Post a reply to this message

Yeah I was a bit suprised when I read it, seems counter-intuitive.  Maybe
I'll try a test image both ways and see how they look side by side.

--
Doug Eichenberg
www.getinfo.net/douge
dou### [at] nlsnet

Post a reply to this message

Nice pic, looking quite photorealistic.
I downloaded your source to give it a try, it takes a long time to 
render. Can't you get a very similar lighting using less lights ? There 
is almost 200 in the example, is all that really necessary ?

JC

Doug Eichenberg wrote:
> I recently read a book on CGI lighting techniques where the author discussed
> dome arrays.  I wrote a short macro that creates the same thing by creating
> a hemisphere of point lights.  The intensity of the lights varies with
> latitude, creating a hotspot at the top of the hemisphere.  I also had it
> add in a few negative lights to give some subtle variance in the lighting.
> Anyhow, the results were pretty nice... very soft shadows with a radiosity
> lit look.  Rendered fairly fast too.
> 
> --
> Doug Eichenberg
> www.getinfo.net/douge
> dou### [at] nlsnet
> 
> 
>

Post a reply to this message

Should work fine with fewer lights too... I've seen some examples with maybe
17 lights.  The trick is to turn shadows off for the array and use a
separate light for the shadow casting, but I liked the results with a shadow
casting dome array better.

--
Doug Eichenberg
www.getinfo.net/douge
dou### [at] nlsnet
JC (Exether) <no### [at] spamfr> wrote in message news:3f4a11e5@news.povray.org...
> Nice pic, looking quite photorealistic.
> I downloaded your source to give it a try, it takes a long time to
> render. Can't you get a very similar lighting using less lights ? There
> is almost 200 in the example, is all that really necessary ?
>
> JC
>
> Doug Eichenberg wrote:
> > I recently read a book on CGI lighting techniques where the author
discussed
> > dome arrays.  I wrote a short macro that creates the same thing by
creating
> > a hemisphere of point lights.  The intensity of the lights varies with
> > latitude, creating a hotspot at the top of the hemisphere.  I also had
it
> > add in a few negative lights to give some subtle variance in the
lighting.
> > Anyhow, the results were pretty nice... very soft shadows with a
radiosity
> > lit look.  Rendered fairly fast too.
> >
> > --
> > Doug Eichenberg
> > www.getinfo.net/douge
> > dou### [at] nlsnet
> >
> >
> >
>

Post a reply to this message

> The trick is to turn shadows off for the array and use a
> separate light for the shadow casting

You don't need to do that. The book is probably not specifically for
POV-Ray. If you use just, lets say 16 lights in the dome it should be enough
for smooth shadows. They just need to be area_lights, using the minimum of
samples possible (that is a grid of 2*2).

Regards,
Hugo

Post a reply to this message

Hmmm, hadn't thought of trying it with area lights.  Actually, I was trying
to avoid them because of higher render times.  I'll have to give it a try
and see what happens.

BTW, tried one last night with 32 lights and it was almost identical, just
needed a little parameter tweaking.

--
Doug Eichenberg
www.getinfo.net/douge
dou### [at] nlsnet
Hugo Asm <hua### [at] post3teledk> wrote in message
news:3f4b0a25$1@news.povray.org...
> > The trick is to turn shadows off for the array and use a
> > separate light for the shadow casting
>
> You don't need to do that. The book is probably not specifically for
> POV-Ray. If you use just, lets say 16 lights in the dome it should be
enough
> for smooth shadows. They just need to be area_lights, using the minimum of
> samples possible (that is a grid of 2*2).
>
> Regards,
> Hugo
>
>

Post a reply to this message