Ok ... here's my offering, and I've got to say that using the jade 
definition in stones.inc was an epic fail. Render time was almost 13hrs 
and I just didn't like the results. Layered textures doesn't seem to be 
a very good fit, but of course it wouldn't surprise me if I was doing 
something wrong ... I only made a few minor tweaks to the jade def :-(

Also in this version I chucked the reflection attribute, which I'm not 
totally understanding because /everything/ has some reflection right?

Post a reply to this message

Attachments:
Download 'work.png' (302 KB)

Preview of image 'work.png'

On 12/03/2011 3:18 PM, Jim Holsenback wrote:
> Ok ... here's my offering, and I've got to say that using the jade
> definition in stones.inc was an epic fail. Render time was almost 13hrs
> and I just didn't like the results. Layered textures doesn't seem to be
> a very good fit, but of course it wouldn't surprise me if I was doing
> something wrong ... I only made a few minor tweaks to the jade def :-(
>
I'm not really surprised using the standard Jade as a base for your 
subsurface texture. Real, expensive jade is very light in colour. The 
lighter the colour the better quality.
I found these values on the internet.
colour rgbft < 0, 0, 0,0.000,0.000>// Jade	
subsurface { < 0.657, 0.786, 0.9>,< 0.00053, 0.00123, 0.0021> } // Jade




> Also in this version I chucked the reflection attribute, which I'm not
> totally understanding because /everything/ has some reflection right?
>

Mostly yes. I've been using reflection {
rgb <0.000,0.000,0.000>, rgb <0.100,0.100,0.090>

The image does seem to be missing something. That's a problem with SSLT, 
it is back to the days of slooow renders. :-(

-- 
Regards
     Stephen

Post a reply to this message

Stephen <mcavoys_at@aoldotcom> wrote:
> On 12/03/2011 3:18 PM, Jim Holsenback wrote:
> > Ok ... here's my offering, and I've got to say that using the jade
> > definition in stones.inc was an epic fail. Render time was almost 13hrs
> > and I just didn't like the results. Layered textures doesn't seem to be
> > a very good fit, but of course it wouldn't surprise me if I was doing
> > something wrong ... I only made a few minor tweaks to the jade def :-(
> >
> I'm not really surprised using the standard Jade as a base for your
> subsurface texture. Real, expensive jade is very light in colour. The
> lighter the colour the better quality.
> I found these values on the internet.
> colour rgbft < 0, 0, 0,0.000,0.000>// Jade
> subsurface { < 0.657, 0.786, 0.9>,< 0.00053, 0.00123, 0.0021> } // Jade
>

>
>
> > Also in this version I chucked the reflection attribute, which I'm not
> > totally understanding because /everything/ has some reflection right?
> >
>
> Mostly yes. I've been using reflection {
> rgb <0.000,0.000,0.000>, rgb <0.100,0.100,0.090>
>
> The image does seem to be missing something. That's a problem with SSLT,
> it is back to the days of slooow renders. :-(

BTW, I thought povray's SSLT (why not the more common SSS?) was supposed to
optimize away media-based SSS.

have you guys been using the full-resolution model or the highest resolution
decimated version?

Post a reply to this message

On 12/03/2011 7:47 PM, nemesis wrote:
> BTW, I thought povray's SSLT (why not the more common SSS?) was supposed to
> optimize away media-based SSS.
>

Don't ask me ;-)

> have you guys been using the full-resolution model or the highest resolution
> decimated version?

I've been using my own decimated version. I reduced is as far as I could 
and still keep most of the detail.

-- 
Regards
     Stephen

Post a reply to this message

Stephen <mcavoys_at@aoldotcom> wrote:
> The image does seem to be missing something. That's a problem with SSLT,
> it is back to the days of slooow renders. :-(

I agree, that's why I further developed my multipass render tricks this last
weekend. The technique allows you to render the subsurface pass separately,
without radiosity or area lights and it's then pretty fast, even with focal
blur.

Then you can render the regular shot without subsurface, but with radiosity,
area lights, etc. and just screen the subsurface pass on top. It's a HUGE
render-time saver.

-------------------------------------------------
www.McGregorFineArt.com

Post a reply to this message

On 12/03/2011 8:40 PM, Robert McGregor wrote:
> Stephen<mcavoys_at@aoldotcom>  wrote:
>> The image does seem to be missing something. That's a problem with SSLT,
>> it is back to the days of slooow renders. :-(
>
> I agree, that's why I further developed my multipass render tricks this last
> weekend. The technique allows you to render the subsurface pass separately,
> without radiosity or area lights and it's then pretty fast, even with focal
> blur.
>
> Then you can render the regular shot without subsurface, but with radiosity,
> area lights, etc. and just screen the subsurface pass on top. It's a HUGE
> render-time saver.
>

That sounds a great time saver. At the moment I'm rendering a Poser Mesh 
at 14 PPS and that's without area lights or radiosity.
I think I must have the settings wrong. :-(
Well tomorrow will show how it comes out. (I hope)

-- 
Regards
     Stephen

Post a reply to this message

Stephen <mcavoys_at@aoldotcom> wrote:
> On 12/03/2011 7:47 PM, nemesis wrote:
> > BTW, I thought povray's SSLT (why not the more common SSS?) was supposed to
> > optimize away media-based SSS.
> >
>
> Don't ask me ;-)

my bad: as McGregor noticed in another thread, SSS plays bad with radiosity like
radiosity used to play with reflections, but you can still render fast SSS
results and composite them with a radiosity scene. :)

> > have you guys been using the full-resolution model or the highest resolution
> > decimated version?
>
> I've been using my own decimated version. I reduced is as far as I could
> and still keep most of the detail.

I've used the highest resolution decimated version from this:

ftp://graphics.stanford.edu/pub/3Dscanrep/dragon/dragon_recon.tar.gz

to "photograph" this miniature:

http://i54.tinypic.com/s0wj1c.jpg

sadly this tone mapped and defocused composite can't quite show off the
different SSLT in the 3 figures (lot in the front, quite less in second dragon
and none in the one behind), so here's the raw render from Blender:

http://i54.tinypic.com/69m4d4.jpg

took 15 minutes on a Q6600.  Then again, SSS in Blender is just a post-process
involving the shadows and the z-buffer.  Much like the focal blur too...

Post a reply to this message

Am 12.03.2011 20:47, schrieb nemesis:

>> The image does seem to be missing something. That's a problem with SSLT,
>> it is back to the days of slooow renders. :-(
>
> BTW, I thought povray's SSLT (why not the more common SSS?) was supposed to
> optimize away media-based SSS.

Patience, people! Performance improvements are yet to come; the current 
implementation is still a most basic brute-force approach.

As for using more common SSS algorithms, they all require UV-mapped 
meshes to work, and therefore are unsuited for use in POV-Ray in their 
original form, and either produce far inferior results (simple depht-map 
based SSS), wouldn't give any significant performance benefits when 
adapted accordingly while still being physically less accurate 
(oversampled depth-map based SSS), or simply cannot be adapted to 
POV-Ray at all (texture space diffusion).

Post a reply to this message

On 12/03/2011 10:47 PM, nemesis wrote:
> Stephen<mcavoys_at@aoldotcom>  wrote:
>> On 12/03/2011 7:47 PM, nemesis wrote:
>>> BTW, I thought povray's SSLT (why not the more common SSS?) was supposed to
>>> optimize away media-based SSS.
>>>
>>
>> Don't ask me ;-)
>
> my bad: as McGregor noticed in another thread, SSS plays bad with radiosity like
> radiosity used to play with reflections, but you can still render fast SSS
> results and composite them with a radiosity scene. :)
>

Yes, your bad. It is impolite to call someone by their surname without 
their title. It sounds disrespectful.


>
> I've used the highest resolution decimated version from this:
>
> ftp://graphics.stanford.edu/pub/3Dscanrep/dragon/dragon_recon.tar.gz
>

Thanks for the link. I didn't know of it. :-)

> to "photograph" this miniature:
>
> http://i54.tinypic.com/s0wj1c.jpg
>
> sadly this tone mapped and defocused composite can't quite show off the
> different SSLT in the 3 figures (lot in the front, quite less in second dragon
> and none in the one behind), so here's the raw render from Blender:
>

Shame about the focal blur. :-(



-- 
Regards
     Stephen

Post a reply to this message

Stephen <mcavoys_at@aoldotcom> wrote:
> On 12/03/2011 10:47 PM, nemesis wrote:
> > my bad: as McGregor noticed in another thread, SSS plays bad with radiosity like
> > radiosity used to play with reflections, but you can still render fast SSS
> > results and composite them with a radiosity scene. :)
> >
>
> Yes, your bad. It is impolite to call someone by their surname without
> their title. It sounds disrespectful.

Is it so Comrade Stephen?  I wouldn't know -- perhaps call him Chevalier
McGregor?  McGregor-san?

I think on the internet titles are irrelevant, specially as people from all over
the world gather around.  It'd be silly to call me "Mr. Nemesis" or something...

> > to "photograph" this miniature:
> >
> > http://i54.tinypic.com/s0wj1c.jpg
> >
> > sadly this tone mapped and defocused composite can't quite show off the
> > different SSLT in the 3 figures (lot in the front, quite less in second dragon
> > and none in the one behind), so here's the raw render from Blender:
> >
>
> Shame about the focal blur. :-(

but macro photography is all about focal blur! o_O

http://www.wired.com/images/slideshow/2008/04/gallery_faves_macro_photos/macro_faves_1.jpg

Post a reply to this message