> With high-precision geometry, you are free to model entire solar systems 
> to scale. With arbitrary-precision geometry, you could model entire 
> galaxies to scale! :-D

And the render time would be similar to flying to another galaxy with 
current technology? :)

Post a reply to this message

Nicolas Alvarez wrote:

>> True. But in the world of brute force renderers that is not _that_
>> impressive
> 
> You still haven't shown me your renderer do it >:)

Here you are :) A bit grainy still and a different setup but still shows
indirect caustics - as it should.

Post a reply to this message

Attachments:
Download 'kuva13.jpg' (19 KB)

Preview of image 'kuva13.jpg'

Severi Salminen wrote:
> W000t! Where can I get my prize? :-)

I haven't rendered it yet :P


> 2. What was the actual color value you specified for the green glass?
> Maybe the glass is so dark that multiple layers simply don't let enough
> light through?

RGB <59, 255, 96> refraction with White 100% reflection also, an IOR of 1.56 and
fresnel falloff.

When I did the MLT/PT render that I posted initially the light passed through no
problems.. since then some setting must have changed. I don't know :)


> 3. My image seems to have less noise on the table and more noise on the
> floor.

That is correct for PT because directly lit surfaces will receive more light and
so smooth out much faster than indirectly lit or occluded (the floor) surfaces.


> So this is not very scientific result. At some point I'll make a better
> test scene and post the dimensions. Or anyone else can do it. Maybe in
> Pov-SDL so we can compare results from it too. And we can keep this even
> remotely on-topic :-)

I wait eagerly both for the chance to do some comparisons with the same data but
also for the time when we can try out your renderer ourselves!

Cheers

Post a reply to this message

"sooperFoX" wrote:
> When I did the MLT/PT render that I posted initially the light passed through no
> problems.. since then some setting must have changed. I don't know :)

Just had a look in the settings.. MLT goes up to 100 bounces with a stay-alive
probability of 0.7 (the same termination method yours does, but also with a
hard limit), but PT has a default max of 6!!

Taking it to 20 is enough to get the glass to render transparent. :blush:

OK, now you can get back on topic. Sorry.

Post a reply to this message

fidos wrote:
> Here a rendering of 1h on one core of a Q6600 2.4 GHz with my MegaPov Montecarlo
> Path tracing patch. Noise is important but there is no spikes.

  I find the almost complete lack of color bleeding from the colored
walls to the white walls/ceiling rather suspicious.

  Don't tell me your patch only reflects brightness, not color?

Post a reply to this message

Nicolas Alvarez wrote:
> And the render time would be similar to flying to another galaxy with 
> current technology? :)

Quite possibly.

I haven't been able to dredge up any material of the performance of 
arbitrary-precision numbers versus simple primitives (not that I've 
looked very hard yet). I don't know if I'm dealing with something twice 
as slow, or a full order of magnitude speed difference, or something 
that gets exponentially worse.

But, yeah, it's definitely going to be slow. :-)

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

    It occurred to him that now he was going to be not only the first 
man on Mars, but the first detective. He grinned at the thought, and the 
last action of the omegendorph set his nerves aglow.
     -- Kim Stanley Robinson, _Red Mars_

Post a reply to this message

> I haven't been able to dredge up any material of the performance of 
> arbitrary-precision numbers versus simple primitives (not that I've 
> looked very hard yet). I don't know if I'm dealing with something twice 
> as slow, or a full order of magnitude speed difference, or something 
> that gets exponentially worse.

GMP is a very fast arbitrary-size number library. But it's integer only 
so I don't think that can help you much :)

Post a reply to this message

Warp <war### [at] tagpovrayorg> wrote:
> fidos wrote:
> > Here a rendering of 1h on one core of a Q6600 2.4 GHz with my MegaPov Montecarlo
> > Path tracing patch. Noise is important but there is no spikes.
>
>   I find the almost complete lack of color bleeding from the colored
> walls to the white walls/ceiling rather suspicious.
>
>   Don't tell me your patch only reflects brightness, not color?

Strange idea !

I think that the walls aren't bright enough.

Some color blending of the green wall is visible on the small diffuse sphere or
on the wall in the penumbrea.

Fidos

Post a reply to this message

Warp wrote:

>   I find the almost complete lack of color bleeding from the colored
> walls to the white walls/ceiling rather suspicious.
> 
>   Don't tell me your patch only reflects brightness, not color?

"almost complete lack of color bleeding" == "some color bleeding", right?

It is just a little faint. Take your favorite photo editing software and
tune saturation up. You'll see some bleeding in the right places :)

Post a reply to this message

Severi Salminen wrote:
> It is just a little faint. Take your favorite photo editing software and
> tune saturation up. You'll see some bleeding in the right places :)

  I would expect it to be a bit more visible. Compare eg. to:

http://www.winosi.onlinehome.de/Gallery_t14_10.htm
http://www.winosi.onlinehome.de/Gallery_t14_11.htm
http://www.winosi.onlinehome.de/Gallery_t14_17.htm

Post a reply to this message