It is currently 25 May 2013, 01:46

All times are UTC + 1 hour [ DST ]




Post new topic Reply to topic  [ 161 posts ]  Go to page Previous  1 ... 4, 5, 6, 7, 8, 9, 10 ... 17  Next
Author Message
 Post subject: Re: 3delight tests
PostPosted: 19 May 2011, 15:56 
Offline

Joined: 17 Feb 2010, 16:13
Posts: 103
i remebered something about the z-depth-like creation and thought this comes from the camera... nonsense.
you're absolutely right. it comes frome the lights! makes much more sense... \:D/
thanks for your answer. i must admit that my question deserved a RTFM...

_________________
SI 2013 @ WIN7-64


Top
 Profile  
 
 Post subject: Re: 3delight tests
PostPosted: 22 May 2011, 15:18 
Offline
User avatar

Joined: 08 Jun 2009, 21:11
Posts: 688
Location: Zagreb, Croatia
never mind...

btw, in latest 3dl download, 3.08, point-based color bleeding works with IBL too, with hair as well. A good reason for new tests....

now a shameless plug (but somehow I'm not sure how many SI-community visits the Area too :) ) - latest image is on Area front page for a while. Back to topic, here's a few more details about hair shading, that I already posted on Area. Really nothing so smart, some images were already posted here, but anyway:

Lighting is a light rig with deep shadow maps. On some places, there's ray-traced ambient occlusion - but hair is excluded from AO.
I intentionally went with as much uniform and simple setup - the goal was to get the same lighting setup for entire character. Dust, background and smoke are separate passes, but character is a just one.
So that's a sunlight with 4096 DSM, back light with 2048, about 20 lights for dome lighting, with 1024 DSM. All lights had enabled both diffuse and specular lighting. I've tweaked shadows to get as much low bias and low smoothing - except back light which has a bit longer bias, in order to get translucency effect. Translucency is not so pronounced here because of type of hairdo, but not big deal to get with more 'spreading' hairdos, like this one, or this. I've used only two lights for these two.
Hair shader is a standard, out-of-the-box one, I think it's what is usually called 'shave'. Again, I've been able to go with more advanced shading model as a custom Renderman shader, but I wanted to keep everything as much simple. Some recipe for bright hair is to use colored shadows, a bit darker and more saturated than diffuse color.

I've noticed that hair density has a significant influence, too high density can do a 'melted' shading. So there are a lot of tweaks to avoid this. Let's say, 'clump' effect at the end of hairs isn't an interpolation between guides and hairs, it's a proportional scaling along existing hair shape - as hairs are closer to guides, they are longer. There are five or six mixes between different ways of 'hair grouping', I think. At the end, there is about 30k hairs, all at 0.15 pixel size. By following guidelines from docs, size is the same along hair length - but transparency is varying, from about 0.1 at roots to 0.5 at tips, full transparency just at tips.
There are another small enhancements, for example. specular intensity is randomized along hairs.

happy '3d-lighting' :)

Image


Top
 Profile  
 
 Post subject: Re: 3delight tests
PostPosted: 22 May 2011, 16:55 
Offline
Administrator
User avatar

Joined: 04 Jun 2009, 12:15
Posts: 3213
Very nice and informative as usual... \m/
As to your "plug": I don't know if many people who would still admit to visit the Area
("the place where black holes go to die..."),
like myself ( :-$ ), visit it by way of the front page
(I only go there - by bookmark - to enjoy the welcoming warmth of the Softimage forum directly).
So it's good you've pointed it out...

_________________
Broken, not stirred...


Top
 Profile  
 
 Post subject: Re: 3delight tests
PostPosted: 22 May 2011, 17:19 
Offline

Joined: 09 Jun 2009, 11:36
Posts: 372
nice renders and tests. one thing that could be a problem is the lightdome with 20 lights and its very high dsm resolution. from my experience its possible that you get a shimmering effect of the hair shadows in animation (besides rendetime issues with such high dsm's and more complex hair setups). have you tryed to use pc based ao on it? what is the difference in rendertime compared to a dsm lightrig?


Top
 Profile  
 
 Post subject: Re: 3delight tests
PostPosted: 22 May 2011, 18:24 
Offline
User avatar

Joined: 08 Jun 2009, 21:11
Posts: 688
Location: Zagreb, Croatia
I've already posted some point-based hair renders in this thread, but for these, I've hacked appropriate sl files to get it properly - now it's available for everyone. Generally, point-based approach seems to be much slower (even I didn't bothered that much with tweaking of this).

Only for hairs and a white lambert on character, mentioned setup renders in about five minutes with 2 core 3delight, on 1536x2048, together with shadow map generation. BTW that's a lighting setup for entire character. For hair only, I'd would go with much less than 20, having a much more room to avoid the flicker.

What I didn't want to bother people at Area, it's a huge influence of hair density, ways of interpolation, that only ICE can handle effectively. Especially for short hair-fur, not big deal to create some kind of signed distance field, or shading transfer entirely with ICE. Unfortunately, I didn't found a way for packing that 'ICE lighting' in ready-to-use public compound.

afaik, there is much more to come (even for SI users :) ), related to hair/fur in 3delight, but I think there is no sense to talk about in advance.

cheers


Top
 Profile  
 
 Post subject: Re: 3delight tests
PostPosted: 23 May 2011, 09:49 
Offline

Joined: 11 Jun 2009, 09:13
Posts: 93
Hey Mathaeus.,

That looks really awesome. Can you tell me how did you set up SSS? For me it doesn't want to work, there's a quite large difference between MR's SSS and 3delight SSS. I figured out that it's some kind of issue with scaling, but not sure


Top
 Profile  
 
 Post subject: Re: 3delight tests
PostPosted: 23 May 2011, 20:27 
Offline
User avatar

Joined: 08 Jun 2009, 21:11
Posts: 688
Location: Zagreb, Croatia
j3st3r wrote:
Hey Mathaeus.,

That looks really awesome. Can you tell me how did you set up SSS? For me it doesn't want to work, there's a quite large difference between MR's SSS and 3delight SSS. I figured out that it's some kind of issue with scaling, but not sure

Hi Jester,

I know you know this, 3delight just emulates the mi_sss_fast stuff, internally that's a different shader, with ability to do a color variance by itself. If you'll get the same result in every condition, that would be miracle, I think.
How does it match the scaling of subsurface effect, it seems it's defined in "misss_fast_shader.sl" in RSL folder (I know you're coder too :) ).
You'll find the code about scaling somewhere after "uniform float scale = i_scale;"... Before this, there's what shader is doing in subsurface precomputing phase ....., after this, there are subsurface scattering parameters.

What final result (in 3delight) you'll want for for human skin at 1SI Unit= 10cm, it seems to be something around 0.012 - 0.016. I got this value from hillybilly test by xsi base member kissb, it worked nicely for me too.

For this image, I've used something from my kitchen, what you'll already get by applying a 3dl subsurface attributes to object with any standard XSI shader, but with two modifications: shading in precomputing phase is changed to 'all direction' ( Nn instead of default Nf), and ability to multiply the scattering by color/texture. As it is for now, it looks like a good old Diffusion or Md SubScatter - just one layer, back-scattering is not so pronounced. Preset is "Skin2". Even it works nicely, I have no idea does it cause problems somewhere else.

Cheers


Top
 Profile  
 
 Post subject: Re: 3delight tests
PostPosted: 24 May 2011, 11:46 
Offline

Joined: 11 Jun 2009, 09:13
Posts: 93
Thanks Mathaeus,

I still struggle with it, for some reasons, playing with the SSS parameters seems to produce bit "unpredictable" results...Interesting that using the fast skin seems to be OK is, but again, the scale issue...But the hair rendering, is sooo nice in 3dl, so I'll try my best to figure out a proper sss to me...


Top
 Profile  
 
 Post subject: Re: 3delight tests
PostPosted: 25 May 2011, 09:32 
Offline

Joined: 11 Jun 2009, 09:13
Posts: 93
Hey Mathaeus,

How do you render eyes? The usual setup seems to be not working, I used to have a separate object for the cornea, with transparency and reflection, and an object for the white part and iris, etc. Now when I render it, it's black...


Top
 Profile  
 
 Post subject: Re: 3delight tests
PostPosted: 25 May 2011, 18:36 
Offline
User avatar

Joined: 08 Jun 2009, 21:11
Posts: 688
Location: Zagreb, Croatia
j3st3r wrote:
Hey Mathaeus,

How do you render eyes? The usual setup seems to be not working, I used to have a separate object for the cornea, with transparency and reflection, and an object for the white part and iris, etc. Now when I render it, it's black...


Separate objects, no shadow casting for both.
What you posted on 3dl forums, looks like too high ray trace bias. Default is 0.01, I think. You could set this to 0.001. Once shadows will come, you'll probably want to apply the 'hit mode' property, set this to 'use material for shadow ray', plug something into shadow port. For deep shadow maps, appropriate option in global render options.


Top
 Profile  
 
Display posts from previous:  Sort by  
Post new topic Reply to topic  [ 161 posts ]  Go to page Previous  1 ... 4, 5, 6, 7, 8, 9, 10 ... 17  Next

All times are UTC + 1 hour [ DST ]


Who is online

Users browsing this forum: No registered users and 1 guest


You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot post attachments in this forum

Search for:
Jump to:  
Powered by phpBB © 2000, 2002, 2005, 2007 phpBB Group