I believe that the biggest problem is the built in diffuse color lighting model. Its seems wrong for human skin. It's also wrong for an ordinary painted wall. I was moving a camera around in a scene and noticed that as I changed my viewing angle of a wall behind the figures, it got enormously lighter or darker. I know that diffuse lighting is not that strongly affected by viewing angle. The flaw this produces in skin is that surfaces at the edge of the figure are too dark. Skin seen edge-on is nearly as bright as skin thats 45 degrees off the viewing angle.

So I started to play with the clay node. It basically does a diffuse model but this node has some additional control that the regular diffuse node doesn't. It has a parameter for roughness. Microscopically rough surfaces have tiny planes pointing in lots of different directions. The cumulative affect of this is that brightness should not fall off strongly as viewing angle changes.

Using the clay node alone gave me some trouble. I eventually settled on mixing 70% diffuse node with 30% clay node. This I plugged into the alternate diffuse channel and turned off the regular diffuse altogether.

Plugging both the diffuse and clay nodes into my existing work on faking skin imperfections and sub-surface scattering, I got the results you see in this picture.

I've placed two Jessi and two V3 figures in the scene. (Forgive the crappy props - I just wanted to get the sense of them being in a room instead of a photo shoot.) One of each uses the stock textures they come with, and one of each uses my shader tree.

Can you guess which is which? :)

The Jessi and V3 shader trees are identical, except for switching out the correct texture for the model. This tree should work pretty well for any texture you have.

I'll post a screen shot of the shader tree next.

(Please note - in order to keep the image under this forums 200K limit, there are some artifacts. The original looks better.)

I'm not on the up to date with all this, but thank you for your explanation, I think I may try playing around with your instructions based on what is already there when I open it, bet I can figure it out easier with this hands on explanation....and you know until now I never realized about the angle of objects, cameras and lights fading....good catch! Let me guess you went to school and now you are a genious!! ;) Your results are awesome.

Very enlightening thanks!

Thank you for the explanation, the shader works very good, though I can barely understand what is going on there :) 4be

Looks good. Will have to review off-line. Thank you.

@bagginsbill: Just a thought... Have you tried "Edge Blend"? "The Edge Blend node blends color nodes based on surface and camera angles" (from the manual...) What you have done looks excellent; it's just possible that edge blend may be able to achieve the same effect at less cost... I don't know, haven't tried it, (can't try it at the moment), just tossing balls in the air and seing where they land..:-)) Cheers, Diolma

Attached Link: http://www.runtimedna.com/mod/forum/messages.php?ShowMessage=86904

Agghhhh, Rendo ate my detailed response to this thread. In summary....looks great. Jessi not right (see her left thigh) - glowing too much. Colorramp looks off to me - adding red at the top-end, which is the opposite to what would happen in reality. Like the idea of plugging the texturemap into the spots - very efficient.

Glad you all find this interesting.

Face_Off - Lost your fabulous detailed response did you? I've done that too. That's why I now prepare what I'm going to say in textpad away from the browser. THEN I refresh the page and paste in my reply. There is some kind of timeout on rendo that prevents replying more than some amount of time after you read the post.

Anyways... I understand that you think the colorramp should not add red to middle - should add red to shadows right? But remember my experiment with "microscopic" skin model a week ago or so? There was MORE red in the middle, not less, according to that model. Is it realistic? Not sure. But lots of ppl are responding to that skin so I think its at least some kind of improvemnt.

Now I noticed a while ago that all your shaders were heavy on the red in the shadows and edges. In particular, in all your pics, the deepest shadows are never "gray", they are clearly red. I added that to my shader a different way. The edge_blend_3 - see that pink on outer_color? That DARKENS the edges to be more red like your shaders. Frankly, I've been carefully studying lots of people in real life and lots of photos - I think the "red shift" of shadows is an artifact of FILM AND LIGHTS, not real life. I have never seen a real person whose "dark side" was red shifted. But film seems to do that and that's what we are used to seeing.

That's my theory - what do you think?

Also, please try my shader with the red color ramp turned off (i.e. set the color_math addere value_2 to black). Do two figures in same light like I've done here. Tell me what you think. I think the red shift up for surfaces TOWARD camera and red shift DOWN for surfaces AWAY from camera is producing more realism that either alone.

.

Diolma - I'm using edge blend and am very familiar with it. The reason I'm talking about clay is that the poser model for diffusion is too simplistic and doesn't work for skin. You need a way to control how much light fades as the surface shifts away from the camera. I understand edge blend is directly related to the surface-camera angle, but it doesn't take into account the suface-light angle. The diffuse and clay nodes use BOTH the surface-camera and surface-light angles. If I had a node that ONLY did surface-light angles, it would be great. In any case, I didn't want to do something altogether different that diffuse lighting - its just that I didn't think the falloff was accurate. BTW: It may be that I could use ONLY the clay node. But the controls were very hard to get what I was looking for.

Oh Face_off - forgot to tell you. In this picture, my back light was getting shadow from one figure to another. So I cheated and turned it to no shadows so I could get the Blinn highlight on all four figures. Of course with her left hand over her thigh, it shouldn't glow like that, nor would it if I had the shadows turned on for it. Also, I really wasn't working closely with Jessi. Her texture sucks, as we've both pointed out before. I just threw her in to show the improvement, while never promising that the result for Jessi was acceptable. However, I think V3 is looking more believable. Still not photo realistic. But the improvement over the original is clear, right? Personally, I don't think that Firefly *** can *** render photorealistically. I've played with POV-Ray a lot in the past and radiosity has IBL and AO beat hands down. I've done a couple renders of poser scenes in pov - the workflow is a pain. I'm trying to find ways to make Firefly generate believable images. After all, it is a lot faster to try out ideas with Poser/Firefly alone. Plus I like the challenge. One of these days I may take these figures into POV-Ray, work out all the tedious conversions of textures, and let the thing chug away for 8 to 10 hours. I bet the results would, indeed, be nearly photographic.

Attached Link: http://www.povray.org/community/hof/mouille.php

.

Impressive!! Now to apply those nodes to all of the Millenium Material Zones...

Whoa!..this is an awesome thread. Got me on the hot seat (interested) bagginsbill. :-) Been fighting for this to.

bagginsbill, that is amazing work. I didn't realize that povRay could do that !

"Now to apply those nodes to all of the Millenium Material Zones... ShaderSpider?

Slightly OT. I'd love to see the code for that POV-Ray image. WOWSA!

This is a good example of how SSS should be represented. It doesn't make shadows red, and I don't think Paul (face_off) is indicating that it should.

Different lighting situations, of course, will yield different results on the skin, but this is basically a standard of how highlights, SSS, and shadows play off the skin.

This isn't a render, it's actually a photo from the Scream 3 movie poster (all rights reserved), but it seemed like the perfect example of where SSS actually occurs, and how it looks in a single-light situation. If Rendo wants this removed due to possible copyright situations, I understand. Hopefully it will be OK. Message edited on: 09/30/2005 23:02

.

Interesting. I'll keep an eye on this thread. Realism is always a tough nut to crack in Poser.

.

.

Care to upload that P6 mat file so the rest of us don't have to manually follow that image above? (Hell yeah! I'm lazy when I can get away with it!)

I wouldn't mind uploading it. But my version calls for David's map (and some other eyes that were far prettier than David's default ones), so if you haven't got that, or it's installed somewhere else, you'll probably get an error

may I suggest making a 2nd version - one that say points to Dork's maps then upload that one? (it would be version neutral then for 5/6 and if you don't have david etc.)

Mmm that's actually a good idea. Only it ought to be to some character that has separate head- and body maps or there's a LOT to edit... Perhaps to Don/Judy?... they're included in both P5 and P6 as well...

Even with someone else's texture it's easy enough to change the texture file than it is to recreate the whole mat file from scratch so thanks!

Folks, I'm not at my Poser PC on weekends so I can't post the mat. But be aware some of the things that look black aren't black and some that look white aren't white. A few are pink and a few are really dark gray. Why don't they show the RGB values in the UI so we can exchange screenshots of nodes easily??????!? The color_math in the file in post # 31 above, for example, seems to be set to black on the second channel. If you leave it like that, you won't get any of the frontal pinking. I'll post an update on Monday with all the colors. Glad you find this interesting. I didn't know the a material file was just ASCII text. I'd have uploaded before instead of using screenshots. Can I just do Attach File for any post? If so I'll put materials up from now on. I thought the forum Attached file was only for pictures. If I can post materials and light sets, that would be great. I find people trying out my materials don't get what I do because the lights they use are very different.

Yeah just save the file as .txt (rename the .MT5 to .txt) and you can post it here :o) All Poser files are plain txt files :) And I did notice the lsight pink :o) AND that the colour ramp was different dark red colours :o)

I looked inside and it mentions the V3 textures twice. Wierd.

In the future, how about I put in a simple color node and label it "PUT YOUR TEXTURE HERE"? Would that be good?

I wish renderosity had something better than the freestuff area. I'd like to see a place where we could post materials+lights+props, etc and you could put the screen shots, explanations, and actual files all together nicely. Then we could compare notes much more easily.

Oh well - here ya go. After downloading just rename and remove the ".txt" part and replace with ".mt5".

Message edited on: 10/01/2005 17:44

This is very helpful. How nice of you to share your findings with the all of us! Thank you for your efforts.

Of course I have no proof but that photo you posted is, in my opinion, NOT an unmodified photograph. It has obvious photoshopping clues. Somebody was messing around to get a more scary effect.

Look carefully at the RIGHT side of her skin, just along the hairline. See how its red there too? That can't be SSS.
The overall tone (blooming white, to yellow, to red) is a pattern I see in lots of badly edited photos. Its a common artifact of letting the software try to enhance contrast without a human judging the result.

As evidence, I am posting another photo image. The one on the left is, I think, realistic. I made the one on the right by doing a single contrast adjustment in photoshop. Does the coloring there look familiar now?

By the way - I see a lot of poser renderings that look like this too - white bloom with yellow fringe. Its because the lights are on too high. As I've said before, after you like your lights - turn them all down 10% one more time. Works wonders.

Bagginsbill : Red shift in film....I agree - there is heaps of redshift (sss) in film - a lot more than in reallife. However that could be explained in that they use very bright studio lights - so lots of light going through peoples skin. That's why I though the "Incidence Effect" was such a great way to simulate it. The colorramp simply applied a redshift to everything, which could give which unrealistic effects. "should add red to shadows right?"....Mmmmm, no - that's not what I think. I agree if you - there should be more redshift in the mid-tones. Shadows and Specular should have no redshift. But you coloramp is adding red to the upper end. The reason a lot of my renders have red shadows is because I'm using th incidence effect from a main light at the camera - so parallel polys will have some red added - which matches most photo's taken with a flash....you'll never see a black shadow on skin, since the bright flashlight creates a lot of sss on those areas. I'm not sure I agree on your logic with the edge-blends. Light penetrates the skin the same way, regardless of the viewers perspective. SSS is red tinted light coming out of the skin from light entering somewhere else. The camera position doesn't impact the sss. It's happening regardless of the camera position. So to add camera position dependant nodes in there just doesn't sound right to me. See http://www.renderosity.com/viewed.ez?galleryid=1007361 - notice the sss is only on parts of her skin that are parallel to the main light (which is off to the right) - very noticeable on her right arm. And the shadows are still black. IMO, the skintones in this render are pretty close to a photo, and I'm just not sure you would get this with edge_blends. If I get time today I'll do a comparison render. Specular on the other hand IS camera dependent - so should have lots of edge_blends (I think there is 3 in the Realism Kit node setup). Max's photo - the right side of her hair is red because it's light coming from just left of the shadow into the shadow. This is excellent research - keep coming with it. Would be good to see some more renders using the clay technique using different lighting - 1 render isn't enough to really get a full assessment.

I originally tried this with an IBL light setup but got white dots all over the place..... Weird.

Settled on the Winter Queen Light setup edited to add : Didn't realise this looks so blotchy :( Maybe I did something wrong on this one....

Message edited on: 10/02/2005 19:31

Bookmarking for future reference.. thanks for the discussion.

marker

bookmarking :-)