Playing around with radiosity and focal blur. Took about 15 hrs on my 
Celeron 2.5GHz laptop. Any suggestions / criticisms?

Phil

Post a reply to this message

Attachments:
Download 'dice.jpg' (107 KB)

Preview of image 'dice.jpg'

Looks nice, though small objects made of marble should show some sub-surface 
scattering, which would mean using media, which then wouldn't work with 
radiosity... Ah what fun :)

My suggestion: switch off radiosity and fake it with multiple light sources, 
once you have it looking about the same as this make all the dice 
transparent and fill them with dense scattering media with the same pattern 
as the pigments they have at the moment. It'll take hours of work and you'll 
get an image that is almost identical, but they will feel more like real 
marble dice!

Anyway, if you can do it it's worth the effort because sub surface 
scattering gives a clue to the size of the object, at the moment they look 
really large. Or you could change the material they're made out of to one 
that doesn't have significant sub-surface effects (I suggest metal or 
glass).

Also if you use a narrower field of view and moved the camera further away 
they'll look smaller... and that's gotta be easier than my first suggestion 
:)

-- 
Tek
http://evilsuperbrain.com

"Phil Mackenzie" <pma### [at] stmarkssaeduau> wrote in message 
news:43e96651@news.povray.org...
> Playing around with radiosity and focal blur. Took about 15 hrs on my
> Celeron 2.5GHz laptop. Any suggestions / criticisms?
>
> Phil
>


--------------------------------------------------------------------------------

Post a reply to this message

The question you should be asking yourself is : was the blur woth the 15
hours?
If the answer is yes, then good. It is a nice image. Were I to try and fix
the size issue, the only thing I would change is the size of the floor
pattern. Those are pretty tiny checkers, if they are not much bigger than
dice.

Very nice.


-s
5TF!

Post a reply to this message

Tek wrote:
> Looks nice, though small objects made of marble should show some sub-surface 
> scattering, which would mean using media, which then wouldn't work with 
> radiosity... Ah what fun :)


Why doesn't media work with radiosity?  Is it just not realistic?  I 
know there is a media on tag in the radiosity block, but I must have 
missed where the compatibility issues arise.

Skip

Post a reply to this message

>> Looks nice, though small objects made of marble should show some 
>> sub-surface scattering, which would mean using media, which then 
>> wouldn't work with radiosity... Ah what fun :)
> 
> 
> 
> Why doesn't media work with radiosity?

Beats me...

Post a reply to this message

Mmm, I like it.

It's simple, yet quite pleasing to the eye.

Post a reply to this message

Orchid XP v2 wrote:
>>> Looks nice, though small objects made of marble should show some 
>>> sub-surface scattering, which would mean using media, which then 
>>> wouldn't work with radiosity... Ah what fun :)
>>
>>
>>
>>
>> Why doesn't media work with radiosity?
> 
> 
> Beats me...
I think the idea was that it works.....eventually.

Post a reply to this message

No, it doesn't work. The media tag in radiosity is just to tell the 
radiosity on the non-media objects to sample from the media. i.e. if you 
have a glowing red media next to a white wall you'll get red radiosity on 
the wall, but if you have a white scattering media next to a glowing red 
wall the media won't pick up any of the red.

It's a matter of the complexity of the problem. With light only coming from 
light sources the media needs to trace a shadow ray for every sample point 
inside the media, then numerically integrate the samples to work out how 
much self-shadowing the media is doing, which in turn requires multiple 
samples along the shadow ray, effectively an n^2 complex algorithm (though I 
suspect the pov implementation's a little more optimal than that). Anyway, 
if you wanted to do that with radiosity you'd need to do these shadow rays 
in random directions at every point, and to get a smooth effect you'd need 
far more samples than for an opaque object, and I'd guess that the tricks 
for smoothing out radiosity on opaque objects don't work nearly as well on 
media, so you could be talking about several orders of magnitude more 
complexity than just switching radiosity off and using a few extra light 
sources.

At least, that's how I think it works.

Still, I'm puzzled that there's no option to do it in POV. Pov's happy to 
let me do any of the hundreds of other things that give me infeasibly long 
render times :)

-- 
Tek
http://evilsuperbrain.com

"Jim Charter" <jrc### [at] msncom> wrote in message 
news:43ea53a5$1@news.povray.org...
> Orchid XP v2 wrote:
>>>> Looks nice, though small objects made of marble should show some 
>>>> sub-surface scattering, which would mean using media, which then 
>>>> wouldn't work with radiosity... Ah what fun :)
>>>
>>>
>>>
>>>
>>> Why doesn't media work with radiosity?
>>
>>
>> Beats me...
> I think the idea was that it works.....eventually.

Post a reply to this message

Tek wrote:
> No, it doesn't work. The media tag in radiosity is just to tell the 
> radiosity on the non-media objects to sample from the media. i.e. if you 
> have a glowing red media next to a white wall you'll get red radiosity on 
> the wall, but if you have a white scattering media next to a glowing red 
> wall the media won't pick up any of the red.
> 
> It's a matter of the complexity of the problem. With light only coming from 
> light sources the media needs to trace a shadow ray for every sample point 
> inside the media, then numerically integrate the samples to work out how 
> much self-shadowing the media is doing, which in turn requires multiple 
> samples along the shadow ray, effectively an n^2 complex algorithm (though I 
> suspect the pov implementation's a little more optimal than that). Anyway, 
> if you wanted to do that with radiosity you'd need to do these shadow rays 
> in random directions at every point, and to get a smooth effect you'd need 
> far more samples than for an opaque object, and I'd guess that the tricks 
> for smoothing out radiosity on opaque objects don't work nearly as well on 
> media, so you could be talking about several orders of magnitude more 
> complexity than just switching radiosity off and using a few extra light 
> sources.
> 
> At least, that's how I think it works.
> 
> Still, I'm puzzled that there's no option to do it in POV. Pov's happy to 
> let me do any of the hundreds of other things that give me infeasibly long 
> render times :)
> 
Thanks for that.  Your explanation does awaken some dim and nearly 
extinguished memories.

Post a reply to this message