[Half Life]

Since I was interested in doing a comparison I simplified the skin MXMs (removed the "velvet" FX) and changed to "real" SSS instead of thinSSS.


There are some nice details in the digits that the thinSSS doesn't capture but it loses overall color purity (due to the lack of scattering mapping with true SSS) and it takes 5-6 times longer to render... I doubt many users will want to go this route. Maybe we'll get some more options for skin type SSS in Maxwell 3...

Best,
Jason.

[Half Life]

Yesterday I got a funny idea -- I'm not sure why it never occurred to me before, but I decided to mix SSS and ThinSSS to get the best from both using one of the DAZ textures as a BSDF weight mask.

The result is a final product that renders nearly as fast as ThinSSS and captures the nice touches of SSS in one render -- I'm not sure every user would be willing to pay the render time price for the additional subtlety, but I think this is a good compromise preset for those who are willing to render a bit longer.



Best,
Jason.

[rusteberg]

I decided to mix SSS and ThinSSS to get the best from both using one of the DAZ textures as a BSDF weight mask.

The result is a final product that renders nearly as fast as ThinSSS and captures the nice touches of SSS in one render -- I'm not sure every user would be willing to pay the render time price for the additional subtlety, but I think this is a good compromise preset for those who are willing to render a bit longer.

what you get when you do that is a hollow glowing shell. (like a Glo Worm - http://en.wikipedia.org/wiki/Glo_Worm)

these are your last two renders. she's like a x-mass light.

[Half Life]

Some more tests of this quirky idea -- I like the faster render times of the mixed approach but the real SSS gives superior results and I still think that if you are conscious of render times (and it seems most users put that as first priority) you are going to go with the pure ThinSSS option just because it gives "good enough" results in a much more tolerable timeframe.

All of these were set to stop at SL 15... render times are not meant to be specific due to differences in the processing power of different computers, I just wanted to get an idea of the ratio of speed gained/lost due to choice in MXM creation approach.

ThinSSS only -- render time 42 minutes... I probably need to increase the scattering coefficient a bit here.


Mixed ThinSSS and SSS -- render time 55 minutes... achieves some of my goals but lets in some subtle undesirable artifacts as well.


SSS only -- render time 90 minutes... IMO currently the best choice, for this but unlikely to be embraced by the users due to render times.


I know there are not many users playing with skin currently here, but I do hope there are some new tools in Maxwell 3 MXMs for these types of problems... because the current toolset will only get you so far.

Best,
Jason.

[rusteberg]

Below are your last three renders in an animated gif:

I would argue you could get your render time for this under 5 minutes and add that frame to this animated gif and you wouldn't even notice the difference......

I think you are making this way more complicated than it needs to be......

edit: All I'm doing is taking your images and reposting them and yet your images still seem overly saturated in red.

[Half Life]

I have been having a discussion with another skin testing Maxwell user via PM and he raised a valid point about the use of K (extinction coefficient) with SSS objects. In reply to this I sent him a link that was once shared with me regarding how Maxwell reads SSS settings and I wanted to share this with anybody else who was curious: http://www.aurora-nephelometer.com/inde ... scattering

The bottom line is that K setting as we see in MXED is not meant for these types of materials -- in fact its purpose is metals, it is a surface only effect.

The "extinction coefficient" in Maxwell SSS is actually arrived at via the combination settings as follows: (Transmittance color + Attenuation) + (Scattering color + Scattering Coefficient)=extinction coefficient

So you can see that there is no need to fiddle with K values in SSS objects.

As an aside, I want to be very clear about something -- for those of you who may not know the situation. A while ago I added Eric Nixon and rusteburg to my ignore list (it's known as "foes" in the forums user control panel). I cannot see their posts (although I see a notification that they have posted something). I notified both of them publicly at the time I did this, therefor they are both aware that I cannot see their posts. So if you see them post something after me, be aware that they are not posting for my benefit or attention, but rather are attempting to get your attention and color your perceptions.

Best,
Jason.

[eric nixon]

Just think of all the innocent users that have listened to this nut-case over the years, He even used to be the number one google result for maxwell-render tutorial! This is because he is a pro spammer, not just a sad case we could feel sorry for.
This moron is surprisingly effective at destroying threads, and tries to make a living by spreading stupidity through retarded tutorials.

NL wtf? is he holding your kittens in a bag above niagra falls?, just ban the nutter..

[Half Life]

After doing some tests I found that SSS noise seems to clear better (meaning at lower SL) with normal maps (vs bump maps) -- so I converted all the skin bump maps to normal maps and used the latest pure SSS to test that out.


This took 48 hours on my i7 920 machine, so I'm sure it could be done in half that (or better) on most modern systems -- still not what I'd call practical, but this project will have to wait because I'll be too busy with other stuff the next few weeks to give this any attention... I'll probably be back playing with skin after the first of the year sometime.

Best,
Jason.

[Rob_Lucy]

I remember an object had to be watertight to work with SSS and DAZ models are not.
Does this mean the watertight rule is not needed anymore?

[Half Life]

The real issue is any dielectric needs to have proper entry and exit normals (to attenuate/refract/etc. correctly), and since SSS is basically a dielectric with "stuff" in it, the same rules apply.

That said the DAZ figures (or at least ones I've used) are "solid enough" to get the job done -- the place where you have a the main holes (the eye sockets/mouth) have another object blocking most of the time... but I'm sure there are cases where you could get odd results.

My preferences at this point are to use ThinSSS, for several reasons. First is has the advantage of rendering much quicker, it tends to handle problem geometry much better, and you can also map scattering.

The other issue is the body is not uniformly translucent. For instance the ears and nose tend to transmit/scatter heavily due to the cartilage (and relatively thin skin) -- but bone is much denser, so the attenuation of the skin itself may be such that is lets alot of light in, but if bone is directly behind it you would get a very shallow SSS effect. The same is true of hands -- the back of the hand often reveals the differing influence of bone/meat/veins under the skin -- whereas the palm has such a high concentration of fat and thick skin that it is hard to make out those "sub-surface" details.

Since SSS is a global property for the whole object it is difficult to get a setting that will look "right". This really requires either real geometry beneath the surface (which makes posing and animation problematic) or much more sophisticated mapping/materials creation techniques than Maxwell currently allows for. The real crux of the thing is getting materials where we can stack multiple layers with each having a "virtual" thickness that is fully calculated before sending the light to the next layer down. What I'm talking about is a material creation system that creates layers in the model of a matryoshka doll... this has applications in not just things like figures, but also things like modern anti-reflective lens coatings (which require coatings of specific thicknesses stacked one on top of another).

So all that said, while ThinSSS is far from perfect, it's really no less "correct" for this application than SSS (at this point)... the reality is we just don't have the tools to make this work correctly in Maxwell -- and this is compounded by problems in the textures for the DAZ models which were never made for an engine like Maxwell.

So that's my long winded way of saying "Yes, SSS works fine (most of the time), but I wouldn't use it".

Best,
Jason.

[Rob_Lucy]

Thanks.

[rusteberg]

The real issue is any dielectric needs to have proper entry and exit normals (to attenuate/refract/etc. correctly), and since SSS is basically a dielectric with "stuff" in it, the same rules apply.

dielectric materials need to be solid to work properly, the same applies to sss materials

That said the DAZ figures (or at least ones I've used) are "solid enough" to get the job done -- the place where you have a the main holes (the eye sockets/mouth) have another object blocking most of the time... but I'm sure there are cases where you could get odd results.

all the models i've used are not solid

My preferences at this point are to use ThinSSS, for several reasons. First is has the advantage of rendering much quicker, it tends to handle problem geometry much better, and you can also map scattering.

i prefer to use thin sss (single sided sub surface scattering) because the models i've used are not suitable for real sss which i have neglected to admit in my previous point

The other issue is the body is not uniformly translucent.

but it becomes so when it is treated as a single sided mesh which i will again neglect to admit.

For instance the ears and nose tend to transmit/scatter heavily due to the cartilage (and relatively thin skin) -- but bone is much denser, so the attenuation of the skin itself may be such that is lets alot of light in, but if bone is directly behind it you would get a very shallow SSS effect. The same is true of hands -- the back of the hand often reveals the differing influence of bone/meat/veins under the skin -- whereas the palm has such a high concentration of fat and thick skin that it is hard to make out those "sub-surface" details.

hmmmm.....

Since SSS is a global property for the whole object it is difficult to get a setting that will look "right". This really requires either real geometry beneath the surface (which makes posing and animation problematic) or much more sophisticated mapping/materials creation techniques than Maxwell currently allows for. The real crux of the thing is getting materials where we can stack multiple layers with each having a "virtual" thickness that is fully calculated before sending the light to the next layer down. What I'm talking about is a material creation system that creates layers in the model of a matryoshka doll... this has applications in not just things like figures, but also things like modern anti-reflective lens coatings (which require coatings of specific thicknesses stacked one on top of another).

maxwell can't do what i want it to do. so i propose a theoretical solution to solving the problem

So all that said, while ThinSSS is far from perfect,

so since thinSSS doesn't really work in this case

it's really no less "correct"

because i'm not using it correctly

for this application than SSS (at this point)...

because SSS really isn't correct either

the reality is we just don't have the tools to make this work correctly in Maxwell

i really just don't know what i'm doing

-- and this is compounded by problems in the textures for the DAZ models which were never made for an engine like Maxwell.

and i'll take this opportunity to blame DAZ models as well for it not working the way i would like

So that's my long winded way of saying "Yes, SSS works fine (most of the time), but I wouldn't use it".

so that's my long winded way of trying to confuse you instead of answering directly

Best,
Jason.

Best,
Mr. Potatoe

xoxo

[Jenny Fermor]

Hi Guys

We will not tolerate provocative, insulting or abusive remarks against other users on our forum, and for this reason I will now lock this thread.

May I please bring everyone's attention to the FORUM RULES: http://www.maxwellrender.com/forum/view ... 45#p225745

And GUIDELINES:
http://www.maxwellrender.com/forum/view ... 12#p127812


Thank you

Jenny Fermor
Maxwell Render Team