Loading...
> >

Fiddling with Iray skin settings...

1…7071737576…91

Comments

  • Oso3DOso3D Posts: 15,047

    I was interested to read that one reason sculptors like marble is that it has translucency and SSS like human skin. Huh.

    (Now let's see about using human skin shaders and adapting it to marble... ;)

     

  • AndyGrimmAndyGrimm Posts: 910
    edited March 2016

    @timmins.william

    they made AMAZING sculpts using marble in the past-> not just skin.. even veiled skin (womens)... check Antonio Corradini's veiled womens..  i did not believe it as i saw that the first time.

    Edit here is one...

    Post edited by AndyGrimm on
  • AndyGrimmAndyGrimm Posts: 910
    edited March 2016
    Arnold C. said:
     

    Aww, I just got my wishlist down to 100 items...frown

     That's why i delete my wishlist from time to timelaugh. unshaven 2 was in my wishlist allready 3 times or more smiley

    I dont want to buy a shader product just to figure out that it has the same weakness as my setups do....   i never saw a long fibermesh hairstyle in iray with flat hairs yet - that's where anisotropy is extrem important to make it look right...and that's where i am testing   on wilder or curly dark hairs a basic shader setup (following pbr principles) does look good...

    Post edited by AndyGrimm on
  • nonesuch00nonesuch00 Posts: 18,333
    edited March 2016

    I had to search that Veil of Humility (I forget the name already) because I thought it wasn't really marble...

    Post edited by nonesuch00 on
  • Arnold CArnold C Posts: 740
    edited March 2016
    throttlekitty said:

    Some types of hair product use metals/minerals for tinting as part of their chemistry, particularly the "fun" colors, so you also end up with tinted specular highlights in real life. For renders, anything goes, it's just a matter of what look you want...

    True, but with a big exception: metals used in hair products are mostly oxides, like ferric oxide. And when a metal chose to become an oxide it ceased to be a conductor, it becomes an insulator/dielectric instead. And in the realms of physics and physically based shading the color of specular reflection of a dielectric material is and always will be... White.

    The problems start if you use any colored lighting, let's say red, and someone's useing a blueish Glossy Color tint, your highlights on your hair or whatever that material should be will be of a Purple color, although they should appear Red.

    So Andy is completely right on this matter. Hair isn't a conductor, and if a shader creator sets it's "Glossy Color" to anything colorful that's just plain nonsense. "For renders, anything goes" and "artistic freedom" is just one's poor excuse for either not willing to do one's homework or don't have any clue what exactly he or she does.

    Post edited by Arnold C on
  • KhoryKhory Posts: 3,854

    I think that depends entierly on what they are trying to mimic. You see each individiual strand of hair has a wide variety of highlights on it and then there is an overall gloss that you get. Some layer somewhere needs to be used for those tiny bright sparks on hair. In theory metalic flakes could be used but to be honest that is pretty touch and go.

  • Arnold CArnold C Posts: 740
    AndyGrimm said:

     That's why i delete my wishlist from time to timelaugh. unshaven 2 was in my wishlist allready 3 times or more smiley

    I dont want to buy a shader product just to figure out that it has the same weakness as my setups do....   i never saw a long fibermesh hairstyle in iray with flat hairs yet - that's where anisotropy is extrem important to make it look right...and that's where i am testing   on wilder or curly dark hairs a basic shader setup (following pbr principles) does look good...

    Me too. Just scrapped a bunch of product for the older Millennium figures, chances are very low I'd put them to use anyways. smiley

    Well, at least there aren't any fancy Glossy Color tints in them. smiley

    Didn't try any fibermesh hairs in Iray yet, in 3Delight they're kinda horrible on render durations.

  • Arnold CArnold C Posts: 740
    edited March 2016
    Khory said:

    I think that depends entierly on what they are trying to mimic. You see each individiual strand of hair has a wide variety of highlights on it and then there is an overall gloss that you get. Some layer somewhere needs to be used for those tiny bright sparks on hair. In theory metalic flakes could be used but to be honest that is pretty touch and go.

    That might be counted as an exception then to simulate any remaining metallic particles in a hair product. But setting the entire Glossy Color of a layer or all layers to anything metallic-like will still be plain wrong and hasn't anything to do with PBS. If highlights turn out to be too strong, in most cases Glossy Reflectivity/Specular Color value and/or Glossy Roughness/Glossiness are set wrong. Turning on the Glossy Color screw than is IMO the laziest attempt, and there really is no need to if one has set those correctly. Colored highlights on hair (and any other insulator) are always the result of the light's color and never the result of the material itself, except it's a conductor.

    Post edited by Arnold C on
  • AndyGrimmAndyGrimm Posts: 910
    edited March 2016

    yes - the shaders from unsheaven 2 looking good on hairs )promorenders)...  but the real magic is done using different colored wisps (as Khory mentioned above) ... darker roots.. for trendy hairstyles artifical (colored) highlights... i did never expect me having a interesst for this kind of things.. but at the moment i study hairstyle & trend websites cool

    So a lot is in the geometry and surface groups using dfferent gradients on base color and translucency for different wisps...   


    Geometry... 5 profile single hairs seems to be the best compromise yet in my tests... this adds up - a basic hairstyle will have about 2 mio polygons.. and unbelievable.. it fits on my 1GB gpu card... render time is definitly ok... but slow because of SSS and refraction...

    here a funny test .. (single hair thickness and current basic hair shader)

    hair thickness test.png
    1080 x 720 - 439K
    Post edited by AndyGrimm on
  • KhoryKhory Posts: 3,854

    I agree that the top coat, which is being used as the actual gloss should be white, but I am still not sold that having an undercoat that will have multiple levels of color is a bad thing. I keep meaning to tell this story and putting it off.. When I first started I read an article about the guys who did dynamic clothing and how few there were that were considered fully qualified to do it. At the time there were about 6 in the world so pretty much every movie you saw that had any clothing in it with movement was done by one or more of those 6 guys. Keep in mind that at that time even poser had dynamics so it was not as if only 6 people had ever used dynamics or anything. No, the reason that there were so very few was because they had to have multiple majors. They needed a computer science degree with a strong math science background and they needed a degree in art as well. The thing was no matter how exact the math and science were for the movement of the dynamic cloth they still needed an artist eye to tweak those settings so that the fabric behaved in a way that humans would percieve as realistic. Sometimes you actually need artistic elements rather than just science and math to produce realistic results.

  • Arnold CArnold C Posts: 740
    edited March 2016
    AndyGrimm said:

    here a funny test .. (single hair thickness and current basic hair shader)

    Oh, you... you just uploaded an old photo of a The Beatles member. laugh

    Looks odd, but nevertheless interesting. smiley

    I wonder if such a fibermesh hairstyle could be transformed into something dynamic with that interesting DynCreator for DazStudio script... cool

     

    Post edited by Arnold C on
  • AndyGrimmAndyGrimm Posts: 910
    edited March 2016

    ....  as for me..  i appreciate input from people with different point of views .... take what looks ok to me.. and cook my own soup with it smiley... at the end .. i have my goals. and want some fun and interessting new things to learn... and i make my own conclusions what works and what not...

    if i would want to render one impressive single image.. i would have zero doubts to cheat physics with creative solutions all the time smiley But my goals are solutions for animations which work in different light situations..  and that's why i stick to real world scales and datas  IF they work.

    Blond hairs seems NOT to work in IRAY.. the difference between specular/highlights and base are to small.. and they look very flat all the time... the dynamic range is to low... now i could try and render RAW- full linear and do the usally photographer postwork stuff.. but using colored specular seems to be the more easy solution at that moment smiley... but who knows.. maybe there is a way to render blonde hairs whitout cheating physics.. not sure yet.

    Edit: the uploaded images show the problem with blonde hairs....   the light is sunsky 2PM... (more light is not natural - glossines is full there... glossicolor white.. and specular IOR 1.55... No way.¨to render ash blonde hairs with reflections following PBR principles. well that's my current state...   any advice or ideas are appreciated.

    hair thickness test - dark.png
    1080 x 720 - 485K
    blond.png
    1080 x 720 - 407K
    Post edited by AndyGrimm on
  • Arnold CArnold C Posts: 740
    edited March 2016

    Depends on what parameters you currently have in use. You could try repeat the base color in "SSS Reflectance tint" and use a RGB 220-210-200 for Translucency Color. Glossiness between 0.8 and 0.9, after all, the surface of a single hair is not completely smooth on a microscopic level.

    You could also use different IOR's depending on the color of your hair. Regarding to the Biomedical Photonics Handbook, red hair would have an IOR of 1.56, and black 1.59, suggesting that the IOR decreases the lighter the color of hair is. 

    Post edited by Arnold C on
  • nonesuch00nonesuch00 Posts: 18,333
    AndyGrimm said:

    ....  as for me..  i appreciate input from people with different point of views .... take what looks ok to me.. and cock my own soup with it smiley... at the end .. i have my goals. and want some fun and interessting new things to learn... and i make my own conclusions what works and what not...

    if i would want to render one impressive single image.. i would have zero doubts to cheat physics with creative solutions all the time smiley But my goals are solutions for animations which work in different light situations..  and that's why i stick to real world scales and datas  IF they work.

    Blond hairs seems NOT to work in IRAY.. the difference between specular/highlights and base are to small.. and they look very flat all the time... the dynamic range is to low... now i could try and render RAW- full linear and do the usally photographer postwork stuff.. but using colored specular seems to be the more easy solution at that moment smiley... but who knows.. maybe there is a way to render blonde hairs whitout cheating physics.. not sure yet.

    Edit: the uploaded images show the problem with blonde hairs....   the light is sunsky 2PM... (more light is not natural - glossines is full there... glossicolor white.. and specular IOR 1.55... No way.¨to render ash blonde hairs with reflections following PBR principles. well that's my current state...   any advice or ideas are appreciated.

    Well technically I don't know a thing about the PBR you are talking about but actually the color for your blond hair is pretty spot on for medium ash blond and those are results in your picture that you often see with dyed blond hair. There are so many shades of natural blond, shades of brown, shades of red and most of the natural ones tend to be a mix of colors which is why most dye jobs look dyed because they give only one color and not a mix of colors that we are used to seeing. You should research some of the professional dye jobs that Hollywood stars pay for to get a better ideal of the mix of shades of hair colors that will make the blond hair look more natural. Some people call this highlighting but in reality the colors are much more mixed than a few bundles of streaked bleached hair of a highlight job. 

  • AndyGrimmAndyGrimm Posts: 910
    edited March 2016

    ok.. i see was not clear enough... it is not about the colors - for a single color it looks actually really good and i know that using different colored strands will make even look the blonde one good  - but while i think this is a very good shader setting..(the only change for the color is in base).   i know that blond,ash or grey hairs.. would clip to white on the highlights in this light situation....   Iray seems to LIMIT highlights...the highlights are the same on all 3 beatles smiley...   they are not bright enough.. and disapear on blond hairs complete...the dynamic range is extremly limited somehow..

    i set also burn highlights to zero and burn highlights on components to off in tone mapping. This should give me full dynamic range and clipping ..   but nothing changed on fibers.. but if somebody missing the redish skin from 4.8 that's the way how to do it .. lol.

     

    Post edited by AndyGrimm on
  • AndyGrimmAndyGrimm Posts: 910
    edited March 2016
    Arnold C. said:

    Depends on what parameters you currently have in use. You could try repeat the base color in "SSS Reflectance tint" and use a RGB 220-210-200 for Translucency Color. Glossiness between 0.8 and 0.9, after all, the surface of a single hair is not completely smooth on a microscopic level.

    You could also use different IOR's depending on the color of your hair. Regarding to the Biomedical Photonics Handbook, red hair would have an IOR of 1.56, and black 1.59, suggesting that the IOR decreases the lighter the color of hair is. 

    that's more or less what i use right now ...  sss and translucency dont have an impact on glossiness brightness  (tested)... that one is new to me: The IOR decrases the lighter the color of hair is - great info...  but i use a 1.55 right now.. should be ok.

    Post edited by AndyGrimm on
  • evilded777evilded777 Posts: 2,466
    AndyGrimm said:
    evilded777 said:
    Arnold C. said:

    The "SSS" designation of the maps currently used on the Translucency Color texture slots is a bit misleading, since they don't have a direct impact on Subsurface Scattering (although you need a material to be translucent to have Subsurface Scattering happening). All they do is additionally darken the color set in the Translucency Color parameter and/or mask out parts where translucence shouldn't have an impact. A better fitting one could be "TM" (Translucency Mask) or "TCM" (Translucency Color Map)... if it would be one.

    Is it so much to ask to use nameing conventions that are correct/relevant? You wouldn't think so...but someone comes up with something and then everyone else runs with it whether its correct or not. Can be quite frustrating, not to mention down right confusing.

    I do nowhere see a problem with naming conventions in the postings above...

    there are just two point of views... those which split the effects by formulas exactly.. and those which see transparency as the sum of  SSS and TMC...  

    there is no SSS without translucency OR transparency (refraction)....  translucency weight drives the amount of SSS and TMC.  You want more SSS effect ? you have to go up with translucency weight...  as such the only thing which is confusing is that translucency weight and translucency color are in another section of the ubershader...  

    No, Andy, in this discussion people tend to agree on the terms.  Its everyone else that I am talking about

  • evilded777evilded777 Posts: 2,466
    Arnold C. said:
    throttlekitty said:

    Some types of hair product use metals/minerals for tinting as part of their chemistry, particularly the "fun" colors, so you also end up with tinted specular highlights in real life. For renders, anything goes, it's just a matter of what look you want...

    True, but with a big exception: metals used in hair products are mostly oxides, like ferric oxide. And when a metal chose to become an oxide it ceased to be a conductor, it becomes an insulator/dielectric instead. And in the realms of physics and physically based shading the color of specular reflection of a dielectric material is and always will be... White.

    The problems start if you use any colored lighting, let's say red, and someone's useing a blueish Glossy Color tint, your highlights on your hair or whatever that material should be will be of a Purple color, although they should appear Red.

    So Andy is completely right on this matter. Hair isn't a conductor, and if a shader creator sets it's "Glossy Color" to anything colorful that's just plain nonsense. "For renders, anything goes" and "artistic freedom" is just one's poor excuse for either not willing to do one's homework or don't have any clue what exactly he or she does.

    But isn't there a Siggraph paper somewhere that states that hair does have two specular highlights and one of them is tinted? I know I'm not making that up,.. I read it, many times.

  • AndyGrimmAndyGrimm Posts: 910
    edited March 2016

    good input evilded777.. i also saw that more then once... and many renderengines have a special hair shader with TWO primary and secondary specular slots (and offset)....

    we can set just the one in base.... but i had a setup where i used colored specular in base and IOR in top coat... did look ok... but top coat specular (grey) was to strong..

    Post edited by AndyGrimm on
  • KaboomKaboom Posts: 40
    edited March 2016

    _removed by Kaboom_

    reason: typing witout thinking firstindecision

     

     

     

    Post edited by Kaboom on
  • Arnold CArnold C Posts: 740
    evilded777 said:

    But isn't there a Siggraph paper somewhere that states that hair does have two specular highlights and one of them is tinted? I know I'm not making that up,.. I read it, many times.

    You remember the title or at least the year by any chance? Just found this article yesterday, which has been published in Siggraph 2003. Just skimmed through by now, the effect you mention is among others described in section 3.4 on page 5 ff. But they doesn't mention a secondary specular highlight, just a secondary highlight, which from what I get out of the text is rather the result of internal reflection off the back side of the hair fiber (that's subsurface scattering then) than specular reflection.

    I'm not sure if Iray is capable to simulate the effect of internal reflection. It's subsurface scattering solution is very much simplified.

  • mjc1016mjc1016 Posts: 15,001

    Basically, it sounds like the Marschner model...that's the foundation paper, Arnold

  • evilded777evilded777 Posts: 2,466
    Arnold C. said:
    evilded777 said:

    But isn't there a Siggraph paper somewhere that states that hair does have two specular highlights and one of them is tinted? I know I'm not making that up,.. I read it, many times.

    You remember the title or at least the year by any chance? Just found this article yesterday, which has been published in Siggraph 2003. Just skimmed through by now, the effect you mention is among others described in section 3.4 on page 5 ff. But they doesn't mention a secondary specular highlight, just a secondary highlight, which from what I get out of the text is rather the result of internal reflection off the back side of the hair fiber (that's subsurface scattering then) than specular reflection.

    I'm not sure if Iray is capable to simulate the effect of internal reflection. It's subsurface scattering solution is very much simplified.

    That's the one. I'm not as conversant with all the physics as some of you folks, so I can't say as I would have spotted that difference.

  • Arnold CArnold C Posts: 740
    edited March 2016
    AndyGrimm said:

    that's more or less what i use right now ...  sss and translucency dont have an impact on glossiness brightness  (tested)... that one is new to me: The IOR decrases the lighter the color of hair is - great info...  but i use a 1.55 right now.. should be ok.

    In the above linked paper it's mentioned that "blond, brown, red, or other light colored hair is very translucent", indicating that the Translucency Weight value on lighter hair should maybe higher than on darker hair, just like skin.  It also has a list of typical parameter values, on the top left of page 8. Among others a typical absorption coefficient.

    Post edited by Arnold C on
  • Arnold CArnold C Posts: 740
    evilded777 said:

    That's the one.

    Nice, thanks. Seems like Prof. Marschner transfered from Stanford to Cornell. smiley 

    There are a lot of more recent publications availabe on his Cornell website, among others one about hair, cloth and other things. If only theese guys would use a tongue that one can clearly understand. Reading Nerd is soo exhausting. laugh 

  • nonesuch00nonesuch00 Posts: 18,333
    Arnold C. said:
    AndyGrimm said:

    that's more or less what i use right now ...  sss and translucency dont have an impact on glossiness brightness  (tested)... that one is new to me: The IOR decrases the lighter the color of hair is - great info...  but i use a 1.55 right now.. should be ok.

    In the above linked paper it's mentioned that "blond, brown, red, or other light colored hair is very translucent", indicating that the Translucency Weight value on lighter hair should maybe higher than on darker hair, just like skin.  It also has a list of typical parameter values, on the top left of page 8. Among others a typical absorption coefficient.

    That makes sense. Vitamin D and sunburn and all.

  • mjc1016mjc1016 Posts: 15,001
    edited March 2016
    Arnold C. said:

     Reading Nerd is soo exhausting. laugh 

    Ummm...yeah.

    BTW...the most recent versions of 3DL include a Marschner implementation...just nothing in Studio to actually use it.   So it would surprise me that if buried somewhere in Iray, it wasn't already implemented.

    The only thing is, the Marschner model only works on curve or fiber type hair.  It won't do much for a transmapped hair.

    Post edited by mjc1016 on
  • Arnold CArnold C Posts: 740
    mjc1016 said:

    BTW...the most recent versions of 3DL include a Marschner implementation...just nothing in Studio to actually use it.   So it would surprise me that if buried somewhere in Iray, it wasn't already implemented.

    The only thing is, the Marschner model only works on curve or fiber type hair.  It won't do much for a transmapped hair.

    Thanks for the info, mjc. Fibermesh hair is the idea that Andy works on.

    That just gave me the thought that maybe Iray couldn't be the real restriction but DAZ's Uber Iray Base material. Might be possible to build a hair material (or convert an existing fiber material) in MDL and than import this as a specialized shader setup via Shader Mixer.

  • AndyGrimmAndyGrimm Posts: 910
    edited March 2016

    thx for the links... reading.

    i worked on testing geometry (thickness, profiles).... and glossiness improved a lot....

    Sorry for the "bold" not finsihed hairstyle... but i want to keep the polycount low as long as i work on my hair shader settings....

    blond1.png
    1080 x 720 - 654K
    brown.png
    1080 x 720 - 673K
    reddish.png
    1080 x 720 - 930K
    Post edited by AndyGrimm on
  • KaboomKaboom Posts: 40

    Hello physically based friends cool I see things are starting to get hairy in here. angel

    I´m still on skin and I have been reading up on stuff and I have a question:

    What unit is used in "distance" tabs? Scatter measurement distance etc...mm, cm, m , inches?

     

     

Sign In or Register to comment.