3D-Daz Daz3d Art - Show Us Your DazSkill

5.00 star(s) 12 Votes

atheran

Member
Feb 3, 2020
355
2,756
First attempt at a commissioned character. I very much do NOT like it. The skin detail is too strong (tried to go with Auto Face Enhancer and it proved to be a bad idea). To the point I didn't bother to let it render past 200 samples, or do any postwork. I'm just putting it here for posterity.

Lighting is as simple as it can be, just an hdri, but I like the expression, the general shape of the face, and the camera DoF.

Aphasia_1.png
 

Krosos

Engaged Member
Dec 1, 2018
2,169
2,080
Look at for example InsertFreakyName his results to get a idea how todo it a bit better a bit subsurface scattering at the right resolution is not a bad idea you just shouldn't overdo it thoug in case of the DOF you need to be carefully balance it against each other.
Thoug be also aware doing this will increase your Render times significantly, especially the more volume objects are in the scene visible.
But such results make 0 sense at low sample rates and to see the outcome correct you need to invest a bit of time per test render if you use AI Denoise you need another balance step.
 
Last edited:

atheran

Member
Feb 3, 2020
355
2,756
Look at for example InsertFreakyName his results to get a idea how todo it a bit better a bit subsurface scattering at the right resolution is not a bad idea you just shouldn't overdo it thoug in case of the DOF you need to be carefully balance it against each other.
Thoug be also aware doing this will increase your Render times significantly, especially the more volume objects are in the scene visible.
But such results make 0 sense at low sample rates and to see the outcome correct you need to invest a bit of time per test render if you use AI Denoise you need another balance step.
I'm not sure if there's a language barrier here or not, because I just can't understand your comment. But since you went through the effort to provide feedback, I'll answer it the best I can.

1) IFN is an amazing artist, there's no doubt about that. But his style and mine don't match, especially when I'm going for a headshot in a commissioned work. His is more painterly, while I'm trying for realism. He shines on imperfections (both in materials and forms, while on this character I want none of that because of how the character was described to me.

2)
Wikipedia said:
Subsurface scattering (SSS), also known as subsurface light transport (SSLT), is a mechanism of transport in which light that penetrates the surface of a object is by interacting with the and exits the surface at a different point. The light will generally penetrate the surface and be reflected a number of times at irregular angles inside the material before passing back out of the material at a different angle than it would have had if it had been reflected directly off the surface.
That has nothing to do with the render resolution. It's a well understood effect of light transmission through the soft tissue that the PBR/Iray shader is trying to emulate by subtly adding a very soft glow at the areas with thin tissue like the ears. Most people bump it the hell up and it gives nice but definitely not realistic results. Here the value is set correctly and a color picker on her left ear will show a shift of the color towards a deep orange.

That said, it's wrong. If we want to be physically accurate, we need an extra map just for the SSS because in different areas of the face the color shift changes from orange/red to yellow to blue to even green. Below I attach an image of how it should be set up with the different areas of the face. But to my knowledge I can't do that properly from within Daz studio. I'd need to open up Mari, load the UVs and create the texture map from scanned data, which I /will/ do when I get the rest right.

Even then SSS is not the whole thing. For accurate and realistic results, the diffuse map should be off and the color information should be coming through the translucency map. SSS to make the whole thing react to lighting in the scene properly and then have three different maps for Epidermis, Dermis and lower dermis to make it diffuse properly. You can see all those maps below. I'm not sure if I go through all the effort for the nine different maps, or if I stick to just Diffuse/Normal/Micro normal/SSS and roughness, but that's something that I'll decide close to finishing with the character and not yet. I'm far, far from that stage still.

face maps.jpg
Here you can see all the different maps needed for just the diffuse texture on a face.

SSS map.jpg
Here is a properly colored SSS map for a face to get accurate results. The default orange in Iray is just a fast approximation.
face.jpg
And finally here you can see a 3d result simply from using a correctly made SSS and some specular and displacement (or micro normal), without all the diffuse maps trouble.

3) Depth of Field has absolutely nothing to do with SSS and the only balance it need is with the rest of the camera settings. In reality, changing the F-Stop in a camera, changes the DoF AND the aperture. In Daz, we have two F-Stop values, one in the camera settings, the other in Render Settings/Tone Mapping. Here I have it set to 5.8 in both, with the rest of the settings (except the render resolution) to match a Canon D90 at the best of my (Daz's and Iray's) ability.

4) DoF does increase the render times, that's true. Also true is that SSS needs more passes to show properly than what I allowed the render to go for. But the big offender in this render are the glasses. Emulating the glass effects (both reflection and refraction) with real IoR (Index of Refraction) and Abbe (chromatic aberration) for the material they are made out of (in this case polycarbonate) does increase the render time by a whole damn lot. Normally, I'd have about 1500 samples in the time it took me to get 250ish samples. And yes, both of those effects DO need a lot of rendering to be noticeable. Especially the chromatic aberration.

5) I never use a denoiser.

That said, the problem here was trying to use Auto Face Enhancer instead of micro normals. Which is something I plan to fix in the next render. And once I get the basic shapes ready, I'll be using scanned data from TexturingXYZ for the skin and the micro normals/SSS. Then off to Blender for a proper groom for vellus hair/eyelashes/eyebrows/hair and Clo3d for custom clothing and draping. All the hair in this render are just placeholders. But first, I need to get the shapes/proportions of the face and body correct and a pose I am happy with. All the material/grooming/clothing will come after.

Now, I'm not trying to be an ass and I appreciate any feedback of course, I'm asking for it most of the time, but the feedback needs to at least be constructive. And for that to happen, it needs to make sense based on how rendering and PBR materials work. Might have simply been a language barrier and maybe I misunderstood your comment, but it didn't make any sense to me.

In case you didn't need a lecture (which..probably is the case and it was a misunderstanding), I hope this comment will at least help others understand a proper pipeline and material creation for a realistic skin shader.
 

Krosos

Engaged Member
Dec 1, 2018
2,169
2,080
I'm not sure if there's a language barrier here or not, because I just can't understand your comment. But since you went through the effort to provide feedback, I'll answer it the best I can.

1) IFN is an amazing artist, there's no doubt about that. But his style and mine don't match, especially when I'm going for a headshot in a commissioned work. His is more painterly, while I'm trying for realism. He shines on imperfections (both in materials and forms, while on this character I want none of that because of how the character was described to me.

2)

That has nothing to do with the render resolution. It's a well understood effect of light transmission through the soft tissue that the PBR/Iray shader is trying to emulate by subtly adding a very soft glow at the areas with thin tissue like the ears. Most people bump it the hell up and it gives nice but definitely not realistic results. Here the value is set correctly and a color picker on her left ear will show a shift of the color towards a deep orange.

That said, it's wrong. If we want to be physically accurate, we need an extra map just for the SSS because in different areas of the face the color shift changes from orange/red to yellow to blue to even green. Below I attach an image of how it should be set up with the different areas of the face. But to my knowledge I can't do that properly from within Daz studio. I'd need to open up Mari, load the UVs and create the texture map from scanned data, which I /will/ do when I get the rest right.

Even then SSS is not the whole thing. For accurate and realistic results, the diffuse map should be off and the color information should be coming through the translucency map. SSS to make the whole thing react to lighting in the scene properly and then have three different maps for Epidermis, Dermis and lower dermis to make it diffuse properly. You can see all those maps below. I'm not sure if I go through all the effort for the nine different maps, or if I stick to just Diffuse/Normal/Micro normal/SSS and roughness, but that's something that I'll decide close to finishing with the character and not yet. I'm far, far from that stage still.

View attachment 1541767
Here you can see all the different maps needed for just the diffuse texture on a face.

View attachment 1541768
Here is a properly colored SSS map for a face to get accurate results. The default orange in Iray is just a fast approximation.
View attachment 1541769
And finally here you can see a 3d result simply from using a correctly made SSS and some specular and displacement (or micro normal), without all the diffuse maps trouble.

3) Depth of Field has absolutely nothing to do with SSS and the only balance it need is with the rest of the camera settings. In reality, changing the F-Stop in a camera, changes the DoF AND the aperture. In Daz, we have two F-Stop values, one in the camera settings, the other in Render Settings/Tone Mapping. Here I have it set to 5.8 in both, with the rest of the settings (except the render resolution) to match a Canon D90 at the best of my (Daz's and Iray's) ability.

4) DoF does increase the render times, that's true. Also true is that SSS needs more passes to show properly than what I allowed the render to go for. But the big offender in this render are the glasses. Emulating the glass effects (both reflection and refraction) with real IoR (Index of Refraction) and Abbe (chromatic aberration) for the material they are made out of (in this case polycarbonate) does increase the render time by a whole damn lot. Normally, I'd have about 1500 samples in the time it took me to get 250ish samples. And yes, both of those effects DO need a lot of rendering to be noticeable. Especially the chromatic aberration.

5) I never use a denoiser.

That said, the problem here was trying to use Auto Face Enhancer instead of micro normals. Which is something I plan to fix in the next render. And once I get the basic shapes ready, I'll be using scanned data from TexturingXYZ for the skin and the micro normals/SSS. Then off to Blender for a proper groom for vellus hair/eyelashes/eyebrows/hair and Clo3d for custom clothing and draping. All the hair in this render are just placeholders. But first, I need to get the shapes/proportions of the face and body correct and a pose I am happy with. All the material/grooming/clothing will come after.

Now, I'm not trying to be an ass and I appreciate any feedback of course, I'm asking for it most of the time, but the feedback needs to at least be constructive. And for that to happen, it needs to make sense based on how rendering and PBR materials work. Might have simply been a language barrier and maybe I misunderstood your comment, but it didn't make any sense to me.

In case you didn't need a lecture (which..probably is the case and it was a misunderstanding), I hope this comment will at least help others understand a proper pipeline and material creation for a realistic skin shader.
I would rethink about the part that it has nothing todo with the physical resolution it's a layerd light diffusion transfer as you nicely quoted from what you have learned since now over the BRDF, now think how DOF adds onto that as a diffusion itself mathematically and how resolution impacts it's accuracy.
You got high goals and you also gonna find out that I-ray is gonna severely limit you in those soon enough also if not especially in your case performance wise, especially when you not going to adapt in some way to the limited Denoiser.

I hope the commission here is a free learning case one for the commissioner by you ;)

 
Last edited:
5.00 star(s) 12 Votes