What makes it better then vray is a few things : First, we don’t have much sss bleeding like in vray ( with the sss2 or the vray skin mtl ) , so lets say, sss bleeding around the upper eyelid fold, or between the lips etc.
So, of the four modes, here, you’re talking of the Directional one, right? Particularly as it seems to be the only one to possess single-scatter, of the four in question.
Or do all measurably preserve detail as well?
Could i see obvious examples of that behaviour, when compare f.e. to live reference material of the same subject?
Emily and all the accompanying material ought to be fairly easy to convert for an Arnold expert: the scene comes in maya and max versions, and shading and lighting ought to be a straightforward matter.
Plus, it comes with heaps of references, so a match can be done by the numbers, rather than by eye.
Secondly, reflection is better by default.. sss2 reflection are bad ( way too much fresnel effect going on ,no ggx ,etc etc. ) and you have to blend it with some layers with the vray blend to get something better ( but still not physically perfect.. )
The inability to choose the BRDF for the spec inside the SSS2 is indeed an annoyance to me too.
This said, creating a blend with an additive ggx layer is trivial.
As for “not physically perfect”, i’d like to hear more of the specifics, for perhaps there are ways around limitations other than a full rewrite (i.e. matching an ior, adding a map, bumping both together. i’d really like to hear what it is you’re referring to.).
In third, in Alsurface surface, the diffuse color is the only thing that affect the overall color.. ( like in sss2 )unlike the vray skin shader that you’r overall color is splitted into the 3 different sss layers.. wich make it impossible to get exactly what you want for the overall color and make it painfull to just try to get there..
From the link you provide, it looks to me like the SSS part is identical to the vray skin shader (minus the scatter modes): one diffuse, three sss, two spec colours and layers, all additive, all weighted.
Could you show me a comparison, or a sample of where the skin shader fails?
I found that most of its defaults tended to be a bit over-saturated, and that it behaved much better with slightly more muted tones.
I personally prefer to layer my shader by myself, using a blend, as there is no performance penalty, and greater control.
But i agree, more work could be done for some of the UI parts.
plus, alsurface has a better feel to it, it feel more depth.. less opaque then vray sss2 or skin mtl . It feel’s more like skin. the way they builded the alsurface is just better.. and also, I like that you have the choice between 4 types of sss, like cubic,directionnal,diffusion,etc. It’s really cool shader.
I’d really really love to see a comparison, with scenes, ideally, to see which part comes up short.
Besides diffusion profiles choice, which can indeed make a marked difference in the result, the setup is really similar to the vrayskin, and any other layered, packaged shader of the type.
BTW I don’t like you render of emily, she look’s plastic, I don’t feel any front scattering here and the reflection are flat.. I don’t think it’s a good exemple of vray skin capability’s.
You’re right, Emily isn’t the best use of any of VRay’s SSS capabilities (by design from the authors: the single-scatter is map-approximated, not calculated, and the spec map isn’t normalised in the distribution.).
However, there are huge contacts sheets i posted from the article i wrote about shading it, which show how it’s the subject’s skin being oily, and the match is seriously high between reference photos and renders, under the same lighting (a dreadful frontal flash setup, which however serves the purpose very well.).
Not perfect, no, due to a few approximations along the way in rebuilding the reference scene exactly, but quite indistinguishable by features.
As i was referring to in my post, usability is also about the time to get there, the rest can be refined somehow else when there’s time left over.
My Emily scene is available in max form, ready to render as per the reference material provided with the original data.
I remain curious to see measurable differences, and most of all some performance data independently verifiable in the form of a usable scene (again, Emily’s three quarters of the way there.).
This non-withstanding, personal preference is allowed (i’m being a d*ck 
), and there are lessons to be drawn from some of the arnold’s UI design choices, i feel.
We’ll soon enough see where this will go…
) in Vray ? You need to play with it guy’s at chaos group, you’ll see what I mean.. just compare it to the sss2 and you’ll see..
