Here is an image I have worked on during rainy days in the summer (although
most days were sunny). It started as an experiment with radiosity only
lighting using two "umbrellas" as light sources. Then I put the pov logo in
and the sphere and cone so it looks like the letters POV. The render time
was about 1 hour then before I added focal blur and blurred reflections. It
wasn't until I got home from the summerhouse to my apartment (and my faster
computer) that I hade time to render with all effects turned on at the same
time. It took just over 19h on my Athlon XP 2400+ 512Mb.
I am very happy with the results and have learned a lot when making it.
Enjoy!

--
Daniel Nilsson

Post a reply to this message

Attachments:
Download 'pov.jpg' (23 KB)

Preview of image 'pov.jpg'

> It took just over 19h on my Athlon XP 2400+ 512Mb.

Well, the image is nice. But I have come to the point where I am fighting
against the impression that such renderings needs to be so slow. They can be
speed up! If it takes 1 hour for the radiosity alone, then dof+blurred
reflections should not even double that. I have learned that it's a matter
of how we do things.

You didn't ask for criticism or suggestions, though... so I will leave it
here.  :o)

Regards,
Hugo

Post a reply to this message

I'm all for hearing about it, however.  Lately, it seems like if my scene
doesn't take at least 12 hours to render, then I forgot to turn something
on.  There are probably lots of things that I could easily speed up, but I'm
not even sure where to look, other than "turning down" certain settings.

I'm sure some of those ideas have been covered here before, but I wouldn't
mind a new discussion.

-- 
Jeremy
"Hugo Asm" <hua### [at] post3teledk> wrote in message
news:3f547d23$1@news.povray.org...
> > It took just over 19h on my Athlon XP 2400+ 512Mb.
>
> Well, the image is nice. But I have come to the point where I am fighting
> against the impression that such renderings needs to be so slow. They can
be
> speed up! If it takes 1 hour for the radiosity alone, then dof+blurred
> reflections should not even double that. I have learned that it's a matter
> of how we do things.
>
> You didn't ask for criticism or suggestions, though... so I will leave it
> here.  :o)
>
> Regards,
> Hugo
>
>

Post a reply to this message

"Hugo Asm" <hua### [at] post3teledk> wrote in message
news:3f547d23$1@news.povray.org...
> Well, the image is nice. But I have come to the point where I am fighting
> against the impression that such renderings needs to be so slow. They can
be
> speed up! If it takes 1 hour for the radiosity alone, then dof+blurred
> reflections should not even double that. I have learned that it's a matter
> of how we do things.

The radiosity settings were much higher in the final rendering and I think I
used more samples for the blurred reflections than what is really needed. I
didn't think much about optimizing, I mostly guessed the settings and did
small test renders with just one effect.
I think most of the render time was spent on the blurred reflections. I took
the code from the povQandT (http://tag.povray.org/povQandT) and increased
the sample count a bit. Is it possible to optimize blurred reflections when
you also use DoF and radiosity?

--
Daniel Nilsson

Post a reply to this message

> Is it possible to optimize blurred reflections when
> you also use DoF and radiosity?

Absolutely yes. When you use DOF or good AA, it's not necessary to average
normals on top of that. In these cases, using only 1 normal is sufficient
for blurred reflections:

normal {
    bumps .3 // Strength of blur
    scale .000001  // Simulating a microscopic bumpy surface
}

As you see, this code is also easier to handle.

You can speed things up further by calculating radiosity separate of the
final render. Try to render your scenes without DOF or AA, just to get the
radiosity values and save them to disk. By doing this, these relative slow
calculations will happen only "once and for all" and not XX or XXX times,
like they do when you activate DOF or AA.. The resulting image will be
nearly identical.


#declare Render_Pass = 1;  // set to 1 or 2

#include "rad_def.inc" // Great setups; easy to use

global_settings {
    radiosity { Rad_Settings(Radiosity_Normal, off, off)

       #if (Render_Pass=1)
            save_file "Illum2-c.rad"
        #else
            load_file "Illum2-c.rad"
            always_sample off pretrace_start 1 pretrace_end 1
        #end
    }
}


Now, lets see... Render-time has dropped a lot by now! Now it's time for
fine-tuning to balance speed vs quality. Here are some quick suggestions:

Try 40 blur_samples for DOF. That's usually enough for me.

In global_settings, set adc_bailout to 1/64 to increase speed in certain
cases + never a problem.

If you want AA instead of DOF and you're using my approch to blurred
reflections, you may need this: +AM2 +A0.0 +R2 -J    This is my way of
setting AA (in the command line bar) and gives superb AA with reasonable
speed. The quality can go higher/lower, of course. If you want it higher,
try  +R3  but if you think, it's all a waste of time (and you don't mind
some noise here and there)  try   +A0.05 +R2    It depends on the scene.

I hope, your rendering will be a lot more fun now!  :o)

Regards,
Hugo

Post a reply to this message

"Hugo Asm" <hua### [at] post3teledk> wrote in message
news:3f55b5a2@news.povray.org...
> > Is it possible to optimize blurred reflections when
> > you also use DoF and radiosity?
>
> Absolutely yes.
...
> Regards,
> Hugo

I tried your suggestions and the render time dropped to 45 minutes (38+7).
Incredible!
I didn't expect such a huge speedup.
The only noticeable difference is the blurred reflections that is more
grainy using your code. I guess that can be fixed by fine-tuning the
parameters.

Thank you very much for your help. I have learned much about optimizing.

--
Daniel Nilsson

Post a reply to this message

Attachments:
Download 'pov2.jpg' (23 KB)

Preview of image 'pov2.jpg'

Congratulations. It's nice when sometimes I'm able to help.

Yes, it's a matter of tuning, to get rid of the bad noise.

Regards,
Hugo

Post a reply to this message