jhu wrote:
> There's still some problems with the render, such as those strange black
> splotches  on the table adjacent to the clear sphere

Oddly enough, those splotches could pass for part of the wood texture.

I like this image. :-)

-- 
William Tracy
afi### [at] gmailcom -- wtr### [at] calpolyedu

You know you've been raytracing too long when your wife walks into the
room wearing only a smile and you start thinking of Colefax's fur macro.
    -- Ken Tyler

Post a reply to this message

jhu escreveu:
> Thanks! My global settings are listed below. What do you suggest I change?

hmm, no photons for caustics?  It seems radiosity is creating some weird 
kind of caustics by its own, except it's black!

Now, count 1000 and error_bound .05?!  whew!  Took a few days to render, 
huh? ;)

In my radiosity scenes I like more fine-grained radiosity shading so I 
generally go for pretrace_start 0.04 and pretrace_end 0.008 or something 
like that.  But I go with error_bound .2:  it's acceptably slow and 
gives good results for the blending of blotches, specially coupled with 
nearest_count of about 16.  Count is about right, though I first try 
with lower counts, like 200 to 600...

As for the dark spots, the correct way to get caustics in povray is with 
photons.  Radiosity has problems with black spots and glass surfaces. 
Use a first radiosity pass without the reflective or transparent 
surfaces, save that rca file and then render the final image with 
photons and the reflective/transparent surfaces back.  Test without 
photon first, then good luck. :)

Post a reply to this message

OK, thanks. Never played around with photons before. I'll give it a shot. BTW,
it's "only" a 16 hour render at those settings...

nemesis <nam### [at] gmailcom> wrote:
> jhu escreveu:
> > Thanks! My global settings are listed below. What do you suggest I change?
>
> hmm, no photons for caustics?  It seems radiosity is creating some weird
> kind of caustics by its own, except it's black!
>
> Now, count 1000 and error_bound .05?!  whew!  Took a few days to render,
> huh? ;)
>
> In my radiosity scenes I like more fine-grained radiosity shading so I
> generally go for pretrace_start 0.04 and pretrace_end 0.008 or something
> like that.  But I go with error_bound .2:  it's acceptably slow and
> gives good results for the blending of blotches, specially coupled with
> nearest_count of about 16.  Count is about right, though I first try
> with lower counts, like 200 to 600...
>
> As for the dark spots, the correct way to get caustics in povray is with
> photons.  Radiosity has problems with black spots and glass surfaces.
> Use a first radiosity pass without the reflective or transparent
> surfaces, save that rca file and then render the final image with
> photons and the reflective/transparent surfaces back.  Test without
> photon first, then good luck. :)

Post a reply to this message

Did you use sky_sphere? Radiosity doesn't
play well with infinite distance backgrounds,
it can cause dark spots. Use an inverted sphere
for the sky...

Post a reply to this message

On Thu, 24 Jan 2008 19:23:44 EST, "jhu" <nomail@nomail> wrote:

>There's still some problems with the render, such as those strange black
>splotches  on the table adjacent to the clear sphere

I do like this very much. The level of detail is very good, I am sure that I can
see texture on her sweater. Is she the sort of person who holds a mug and puts
her fingers through the handle? 
I' getting lost in this :)

And I love self referral in anything.
Excellent!


Regards
	Stephen

Post a reply to this message

Very nice, I always like a good Escher tribute. Looks very real. Good wood
texture! It's also quite dark but this works well.

If you solve the blotchiness (looks like radiosity artifacts to me) and
re-render it, then you might also consider using a longer camera angle to
prevent the outer spheres being stretched by perspective (if this means putting
it outside the room then the no_image keyword might come in handy). Just a
thought!

keep it up
Bill

Post a reply to this message

Tim Attwood napsal(a):
> Did you use sky_sphere? Radiosity doesn't
> play well with infinite distance backgrounds,
> it can cause dark spots. Use an inverted sphere
> for the sky...
> 
> 
Does it? I'll try it, thanks!

Post a reply to this message

nemesis nous apporta ses lumieres en ce 2008/01/24 22:24:
> jhu escreveu:
>> Thanks! My global settings are listed below. What do you suggest I 
>> change?
> 
> hmm, no photons for caustics?  It seems radiosity is creating some weird 
> kind of caustics by its own, except it's black!
> 
> Now, count 1000 and error_bound .05?!  whew!  Took a few days to render, 
> huh? ;)
> 
> In my radiosity scenes I like more fine-grained radiosity shading so I 
> generally go for pretrace_start 0.04 and pretrace_end 0.008 or something 
> like that.  But I go with error_bound .2:  it's acceptably slow and 
> gives good results for the blending of blotches, specially coupled with 
> nearest_count of about 16.  Count is about right, though I first try 
> with lower counts, like 200 to 600...
> 
> As for the dark spots, the correct way to get caustics in povray is with 
> photons.  Radiosity has problems with black spots and glass surfaces. 
> Use a first radiosity pass without the reflective or transparent 
> surfaces, save that rca file and then render the final image with 
> photons and the reflective/transparent surfaces back.  Test without 
> photon first, then good luck. :)
In this case, replacing the radiosity with photons don't look feasible. The 
caustics are a projection of the sky. You'd need to use a light dome, possibly 
composed of many area_light with photons{area_light}. You'll end up with tons of 
photons, meaning a photons phase that will be extremely long and a memory use 
that may get prohibitive.

-- 
Alain
-------------------------------------------------
You know you've been raytracing too long when you've tried to scan your face for 
a texture.
Quietly Watching

Post a reply to this message

"jhu" <nomail@nomail> wrote:
> There's still some problems with the render, such as those strange black
> splotches  on the table adjacent to the clear sphere

It's beautiful.
I thought the black splotches were intentional to make the table looking more
differentiated.
They really fit. :)


             bluetree

Post a reply to this message

jhu nous apporta ses lumieres en ce 2008/01/24 19:23:
> There's still some problems with the render, such as those strange black
> splotches  on the table adjacent to the clear sphere
> 
> 
> ------------------------------------------------------------------------
> 
Nice, but it's so dark! Even pushing my monitor brightness to the max I still 
loose most of it. Looks good after adjusting gamma by +2.
Maybe increasing the brightness of the sky.

You may try using:
pretrace_end 0.005, 0.0025 or 0.00125 for 1, 2 or 3 more pretraces steps.
nearest_count 20 // ask to average more samples. Maximum value.
count 1600 // Maximum value.
adc_bailout 0.01/5 //increase divider more if needed.
raising error_bound a little, maybe 0.08 to 0.12. This may at least reduce the 
contrast of the dark spots.

Unless doing multipass, comment out always_sample 0


-- 
Alain
-------------------------------------------------
No man or woman is worth your tears, and the one who is, won't make you cry.

Post a reply to this message