Page 1 of 12 123 11 ... LastLast
Results 1 to 20 of 234
  1. #1
    Moderator m9105826's Avatar
    Join Date
    Dec 2007
    Location
    Fairfax, VA
    Posts
    3,067

    Arnold skin shader ported to Cycles

    Click image for larger version. 

Name:	strip.jpg 
Views:	20552 
Size:	45.8 KB 
ID:	226131

    My project for the day has been porting the Arnold skin shader to Cycles now that we have SSS. I've made good progress, but there are still speed up options that I need to figure out how to properly implement. Mostly some math relating to light paths that's holding me up, but I should be able to work through it soon.

    UPDATE: Version 2 (link below) - Adjusted some weighting after Brecht fixed difference between prog/non-prog rendering. Reflections are now clamped at 1 provided that you don't enter a weight higher than that. Should help with noise a bit as well.

    2.1 - Correct Oren-Nayar values

    http://www.file-upload.net/download-...ked.blend.html - Obsolete skin shader in here. Grab the one below and append if you'd like to render the head.

    Just download, unpack, and render!

    http://pasteall.org/blend/20547 - Version 2.1

    A smaller file for those only interested in the material without the head/textures/displacements/etc.

    Let's jump right in!

    Diffuse Color:

    The color of your diffuse material. For skin you will usually be plugging in your skin color map here.

    Diffuse Bump:

    For plugging in a bump/normal map for your diffuse component.

    Diffuse Weight:

    Click image for larger version. 

Name:	diffuseWeight.png 
Views:	607 
Size:	214.4 KB 
ID:	226134
    Diffuse weights: .01, .25, .5, .75, 1.0

    Change to affect the overall amount your diffuse component is expressed in the final material. Used to balance diffuse/SSS contribution.

    SSS Overall Weight:


    SSS Weights: 0, .25, .5, .75, 1 (Diffuse Weight 1)

    Overall weight of all three SSS layers. Can plug in black and white map here to control which areas are affected most by scattering.

    Shallow Scatter Color:

    Controls the color of the tissue just under the skin surface. Default is #FFE8C4.

    Shallow Scatter Weight:

    Controls contribution of shallow SSS to final material.

    Shallow Scatter Radius:

    Controls scatter distance of RGB for scattering events. Recommended to keep all 3 the same in this setup.

    Mid Scatter setting and Deep Scatter settings follow the same parameters as Shallow.

    Primary Reflection Color:

    Controls the reflected color of your primary reflection/specular component.

    Primary Reflection Weight:

    Controls contribution of primary reflection to final material.

    Primary Reflection Roughness:

    Controls the roughness of your primary reflections.

    Primary Reflection Bump:

    Allows you to connect a bump/normal map to your primary bump component.

    Primary Reflection Fresnel:

    Control the facing falloff of your reflection component.

    Secondary Reflection settings are the same as primary. It's recommended that you use primary for tighter reflections and secondary for broader reflections.

    Global SSS Radius Multiplier:

    Overall scales: 0.1, .5, 1, 1.5, 2.0

    Allows you to control the global scale of your SSS components. Use to change the overall depth of your SSS.

    To Do:

    Options to only sample only the SSS material in GI and Glossy rays. Should save a good amount of time and help clear up noise without changing the look any appreciable amount.

    Bugs:

    Not sure if it's my material or SSS in general, but I get very different results between progressive and non-progressive rendering. I've been using non-prog for all of my tests.

    Special thanks to Lukas Tonne for helping me with figuring out how to clamp SSS/Diffuse contributions. Special thanks to Thomas Dinges for helping me get my build environments up and running to actually get a copy of Blender working with SSS!

    Keep an eye out for updates to the shader as I go.

    Please leave me any comments or questions you might have, I'll answer them ASAP.

    Enjoy!

    Name:  untitled6.png
Views: 18758
Size:  151.4 KB

    Last edited by m9105826; 03-Apr-13 at 19:00.
    Long time 3D artist and member of the official Cycles Artists Module
    https://www.youtube.com/user/m9105826 - Training, other stuff. Like and subscribe for more!
    Follow me on Twitter: @mattheimlich or on my blog



  2. #2
    Member
    Join Date
    Apr 2013
    Location
    Northern Germany
    Posts
    113
    Wow! Nice work. Thank you very much for sharing this ;-)
    can't wait to give it a try!



  3. #3
    Member PA3D's Avatar
    Join Date
    Oct 2012
    Location
    Montreal, Canada
    Posts
    195
    Thanks a bunch, having a look at it right away!
    Paul-André

    If You got nothing good to say about someone, say nothing at all!



  4. #4
    Oh shit... I'm in the bed with an iphone, i will want see the nodes... T_T




    i go to sleep



  5. #5
    Excellent!



  6. #6
    This is AWESOME!

    yeah the non-prog give a completely different result. its a little blown out for starters



  7. #7
    Moderator m9105826's Avatar
    Join Date
    Dec 2007
    Location
    Fairfax, VA
    Posts
    3,067
    Comparing to Arnold, Non-Prog would appear to be the "correct" setting.
    Long time 3D artist and member of the official Cycles Artists Module
    https://www.youtube.com/user/m9105826 - Training, other stuff. Like and subscribe for more!
    Follow me on Twitter: @mattheimlich or on my blog



  8. #8
    Member cekuhnen's Avatar
    Join Date
    Mar 2003
    Location
    Grosse Pointe Woods, Michigan, USA
    Posts
    10,536
    very impressive - how fast or slow does SSS work in Cycles with GPU?



  9. #9
    Moderator m9105826's Avatar
    Join Date
    Dec 2007
    Location
    Fairfax, VA
    Posts
    3,067
    Doesn't run on GPU yet. Brecht is working on it per the comment in the commit logs.
    Long time 3D artist and member of the official Cycles Artists Module
    https://www.youtube.com/user/m9105826 - Training, other stuff. Like and subscribe for more!
    Follow me on Twitter: @mattheimlich or on my blog



  10. #10
    Originally Posted by m9105826 View Post
    Comparing to Arnold, Non-Prog would appear to be the "correct" setting.
    Im getting a weird checkered pattern with non-prog. I'll keep testing



  11. #11
    Moderator m9105826's Avatar
    Join Date
    Dec 2007
    Location
    Fairfax, VA
    Posts
    3,067
    If you can, find a newer build. That issue was fixed earlier today.
    Long time 3D artist and member of the official Cycles Artists Module
    https://www.youtube.com/user/m9105826 - Training, other stuff. Like and subscribe for more!
    Follow me on Twitter: @mattheimlich or on my blog



  12. #12
    Originally Posted by m9105826 View Post
    If you can, find a newer build. That issue was fixed earlier today.
    Ah, i see. The 32 bit version is newer. It works fine.

    Thanks Bro!

    looking forward to updates



  13. #13
    Member DingTo's Avatar
    Join Date
    Mar 2008
    Location
    Southern Germany
    Posts
    1,469
    Amazing work, thanks a lot for your effort and the detailed explanation! Just checked the simple blend file quickly, looks great!
    Will test more with it this evening.
    Blender developer (Cycles / User Interface)
    Blog -- Blender Podcast -- OpenShading -- BlenderDay -- Blender Network profile



  14. #14
    Member rasper's Avatar
    Join Date
    Aug 2008
    Location
    Utrecht, The Netherlands
    Posts
    17
    Thank you very much indeed! I really appreciate the detailed info you give along with the blends and the hard work. Enough talk for now, time to check it out!
    CG Artist and animation lover
    linesofjasper.com
    Blender Network Profile



  15. #15
    Member
    Join Date
    Jun 2010
    Location
    Warsaw
    Posts
    786
    thank you man this looks awesome!!

    About the non-prog/prog integrator differencies, as you said there is a really big difference. But I am not sure the non progressive is looking better. Isn't the SSS way too deep in the non progressive (i am talking about the big file with the scanned face)? On the other hand the progressive with the default file settings looks too sharp, but increasing the radius a bit in my opinion give the best result (and by far also the fastest render time).
    My website - www.mccinfo.it/bernardo.html
    Blender Cycles: Lighting and Rendering Cookbook - http://www.packtpub.com/blender-cycl...-cookbook/book



  16. #16
    Member blazraidr's Avatar
    Join Date
    May 2009
    Location
    Australia
    Posts
    550
    Oh well yours destroys my test
    (don't mind the grain)
    YouTube Channel : www.youtube.com/user/blazraidr
    Jaguar XKR-S WIP
    Any requests? Send me a PM



  17. #17
    Member manitwo's Avatar
    Join Date
    Feb 2007
    Location
    Austria
    Posts
    102
    why is there a difference between progressive and non-progressive to begin with?
    shouldn't the final images be identical? that's the beauty of unbiased renderers right?
    i'm glad cycles offers us a way to introduce bias to cut rendertimes but i wouldn't expect
    that each integrator produce a completly different image. or is this something solely sss-specific?
    if yes, what integrator should we use to get the most correct result?
    i get that this sss implementation is not the most physical correct there is – but it is fast – and i'm loving
    this but the idea of twisting the knobs till il looks right – then changing the integrator and do all over
    again is not something i'm really fond of



  18. #18
    Member
    Join Date
    Apr 2013
    Location
    Northern Germany
    Posts
    113
    Originally Posted by manitwo View Post
    why is there a difference between progressive and non-progressive to begin with?
    shouldn't the final images be identical? that's the beauty of unbiased renderers right?
    i'm glad cycles offers us a way to introduce bias to cut rendertimes but i wouldn't expect
    that each integrator produce a completly different image. or is this something solely sss-specific?
    if yes, what integrator should we use to get the most correct result?
    i get that this sss implementation is not the most physical correct there is – but it is fast – and i'm loving
    this but the idea of twisting the knobs till il looks right – then changing the integrator and do all over
    again is not something i'm really fond of
    I think this is due to the fact that SSS was just added yesterday and still has some bugs...

    Patience my young Padawan



  19. #19
    Member manitwo's Avatar
    Join Date
    Feb 2007
    Location
    Austria
    Posts
    102
    right
    if this is a bug and not intended behavior i'll take it back and shut up



  20. #20
    Member elbrujodelatribu's Avatar
    Join Date
    Dec 2009
    Location
    Sevilla - Spain
    Posts
    445
    Wow, fantastic node group. Awesome result!
    Yo fuí a Blendiberia 2013

    elbrujodelatribu - blog / deviantart / blendswap / sketchbook



Page 1 of 12 123 11 ... LastLast

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •