Attached Link: http://www.renderosity.com/viewed.ez?galleryid=757841

I have seen other images where the skin looks increadably lifelike. I know the original texture is important but it seems that lighting makes all the difference.

In my recently uploaded image I used the Anne Marie Goddard texture map in order to get a good skin texture. But it somehow just doesn't strike me as real as it could be.

I guess I'm asking if anyone knows of a good tutorial or can give me any help on making skin look more life-like. I'd also be interested in knowing if there is a better looking texture out there. Please feel free to see the image I have referenced and post any help you can provide. I'd even be interested in ways to make this image more attractive.

Thanks for your input!

Nebula

Attached Link: http://www.renderosity.com/viewed.ez?galleryid=756153&Start=1&Artist=cake1&ByArtist=Yes

Well

On my opinion, what makes a skin look realistic is :
-texture
-bump
-light and shadows (otherwise, it looks all flat)
-and maybe highlight, depends on the effect you want to give.

a lot of possibilities, but what i would suggest on this image would be first to get rid of all the lights you have that make that global illumination...

Then add one spot, and maybe another one, and see how it goes. when you like it, post it here, we'll see if there is any difference ;o)

Use many dark lights combined in an orb formation to create a suffuse light. Dont be afraid of color. Be like a photographer, meaning dare to be daring and always experiment. K73

Attached Link: http://www.runtimedna.com/messages.ez?forum_id=28&Form.ShowMessage=89342

I have posted a neat trick about skin in the P5 forum at RDNA. Also, don't forget about the skin node - I created the attached image using P5's skin node and a light set created from a HDRI image.

Well I've spent a huge amount of effort trying to achieve realistic skin in P5, and had some success and some difficulties. Some points....

1. Textures are of very little importance. The image above was done using the standard V3 hi-res maps. Most of the texture packs I've seen give worse results than V3 standard maps.

2. Lighting is also relatively unimportant for skin realism (but critical for the image as a whole). If you use 1 spotlight, the image will render like a photo that was taken with a flash. If you use one of the many GI light sets, the image will look like a photo taken outside. As a rule, I use 1 main light (like in the image above) and a filler light. There are good websites describing 3 point lighting which is useful.

3. I played around with the skin node in P5 (stewer's suggestion), but couldn't get any decent results with it.

4. The secret to realistic skin from 3d apps (IMHO) lies in the tinting of colour of the skin depending on the angle of the camera. There are huge discussions on the topic at www.cgtalk.com (see Steve Stahlberg's theories) - but implementing them in P5 is challenging. The other facet is getting the specular light right. There are all sorts of tricks for this, but I find the neatest is to use Blinn specular, with the bump map plugged into the input colour. Another neat trick is to set ambient value to 1, and set the ambient colour to (r=20, G=0, B=0). This makes the skin glow a touch in the darker areas - adding to the realism.

Message edited on: 09/08/2004 09:09

.

You can implement Stahlberg's tricks in the material room - see my RDNA link.

WOW!! And I mean WOW!! You all have done some incredable work!! If I could do half as good..... I'd be thrilled!! This is great! Ok, I'll calm down now. Thank you all for all your input. I am going to give it a try. Please feel free to continue adding to this thread. Every little bit helps. Stewer, I must apologize. I almost didn't believe you had uploaded a digital image. My first thought was that you had uploaded a photograph and was trying to pass it off as something you had created. I do believe you now. Not only is her skin gorgous, the out of focus background really made a huge difference in it's authenticity. It's an awesome piece of artwork! Face Off, I am totally impressed at your efforts to figure out the science behind the output. Those are killer legs man! I have gone to the website that you mentioned as well as Stewer's on RDNA. I'm going to have to do some serious study to figure out all the pieces that have to go together to get the output you all have made. If you ever come up with that Python script, I would be very happy to test it for you. If I can help in any way, please feel free to write me! Cake1, Thank you for your input as well. I've been to your site often and am always excited about what I will find new next. You do wonderful work! I have a tutorial on keylighting but I am having trouble "seeing" the subtle effects that make the difference. Going thru the tutorial, I entered the numbers it stated and got basically the same output the tutorial showed. But when I tried it on my own, not as much luck. Sorry to rattle on so. Please, keep your opinions coming! Nebula

"Most of the texture packs I've seen give worse results than V3 standard maps."

I agree with this. In fact, Vicki & Mike's standard Hi-res are much more effective under various lighting conditions than most of the so-called "photo realistic" textures you can purchase. The reason is that most of the photo-based textures out there have a specular effect "baked" into the skin as a result of using photographs with uneven lighting for the texture. This becomes problematic when you want to use a global lighting setup, or animated lights, where the position of the light on the character will change, and may not correspond to the "highlighted" areas of the texture at all times, and this results in less than realistic effects.

Most of DAZ's hi-res mats for their characters do not suffer from this ailment (they are more continuous tone), allowing the artist to achieve the specular and dark areas of the skin texture with the use of lighting and materials - the way it SHOULD be done.

I'd post a side-by-side comparison render of Daz's standard hi res mats vs. one of the top selling "photo" mats that can be found here under true global illumination conditions to illustrate clearly what I mean, but I wouldn't want to single out any one merchant, or indicate their products are not good... because they ARE good for some uses, but just not as usable under various lighting conditions. Message edited on: 09/08/2004 14:53

Maxxxmodelz, I agree totally with you. Nebula, the node configuration looks something like the image above. I haven't expanded the nodes, because a) they wouldn't fit on the picture, and b) it's very WIP, and I'd hate to say this is the solution, when in reality every render I currently do experiments with something new. The node setup from top to bottom is.... First 2 colour maths subtracts from the texturemap the combined spots and the fresnel effect. The next colour maths combines the following two spots - one for larger black spots, and the other for almost undetectable small spots. The next maths function multiplies to edge_blends to form the fresnel effect. The colour_ramp add various reds and blue to the ambient. The final 4 nodes make up the specular. This is a simplification on Stahlbergs method but achieves a similar effect, although a lot more work is need here. The bump map plugs into the colour of the blinn. The edge_blend effect the eccentricity. The maths function just add a little to the bump and probably should be ignored. And yes, I did write a python script to create this. The setup screen is on the right. It is very WIP at the moment. At one stage I thought it would be worth releasing, however the more I delve into this area, the more it seems we don't know! So I'm not really sure if it will ever be ready. And I don't know if it is a "one solution fit all" issue.

Well, I've posted a modified version of my original. I would like your opinions. I've played around with the lighting as Cake1 had suggested. I've also darked the drape upon which she's kneeling in order to avoid drawing the eye away from the subject. I did this based on Steve Stahlberg's composition tutorial that Stewer & FaceOff mentioned. I think it was definatly an improvement, (although you wouldn't think it would be necessary considering the subject - hehe).

Thanks everyone for your help and input. Please feel free to let me know what you think. This has been a very informative thread. Keep it going.

Nebula

What I found recently works, is setting ambiant color for skin at red/black 30 or so on the setting, looks realistic, at least on caucasian Vickys, Native American Vickys glow at that wrong though setting darker red helps.

Attached Link: http://www.3drender.com/light/index.html

You may slap me for this, but your image is underexposed now. It's dark and lacks contrast, and overall looks murky. If you take a look at the histogram, you can see that you're only using half of the tones available. A really great resource for lighting advice is the book "Digital Lighting and Rendering" by Jeremy Birn. It's available at Amazon.com, and I saw the other day that Poser Pros is selling it as well. (Personally, I thought the section on key-to-fill ratios was worth the price of the book alone, because that's the first time I'd had it explained to me.) The author has the first chapter as well as some other useful information on his website, including a tutorial on 3-point lighting.

I wouldn't think of slapping you FyreSpiryt. I am inviting all the input I can get! I appricate your viewpoint and I agree with you that it's dark. I used StefyZZ's Asia lights and didn't modify them at all. I think it made a dramatic difference by itself. I need to really dig into it. It really helps when I see improvement! And it helps to be encouraged by others as well. Thanks! I've passed by the book you mentioned several times mainly from cost. But I do hear and see it mentioned alot. Gonna have to breakdown and get it I think. I've also been playing around with rendering it in black and white. Sometimes that gives it more realism than the color does. I did do one with the bump map attached to the bump node of the skin. She looked like one of the body husks found in Pompei. Pretty wicked! Anyway, thanks again for your input. It's always welcome! Nebula

.

wha? How did you do that?? I'm completely baffled by your settings but would like to know more. Is there any way your rig can be shared so that newbs to this area like myself can examine/dissect them? -WTB

Nice looking skin face_off! I've been tweeking these settings from time to time for the past few months... (3-point lighting, Poser 5 render, texture based on Morris' kit in the mp, figure/model by me with uv-mapping and texturing by Fatale)

...note, enabling the translucence channel will increase your render time, so you might want to disable it under final render.

...another note... yes, I'm 'cheating' on all the secondary maps (including the bump map) - they all just use the primary texture, but hey, it works for me ;).

sorry, here's the node setup again (I had to hide something ;).

Now I gotta ask about this figure! Looks great. Where? When? :) -WTB

ok, one more note.. note that I have the Diffuse_Value set to 0. The primary texture comes from the Alternate_Diffuse channel in renders. I still have the texture plugged into the Diffuse channel though, so it shows up in the editor window (but not used from there in the render).

Thanks WtB... you can see more images of her in my gallery... we're busily trying to get her finished - hopefully before Christmas.

Spanki, the skin and the model look great! Thanks for supplying the node info! :-)

I am particularly curious about how face off added those spots...lots of nodes there. (crossing fingers for built-in realistic high-poly genitalia on new model)

(crossing fingers for built-in realistic high-poly genitalia on new model) ...uhm... she's being re-uv-mapped now, due to some recent surgery ;).

Awesome :D finally someone makes sense

I cant belief I missed this tread until now. Some great stuff here. Face_off: The best skin I have ever seen from Poser. Any changes to get a closer look on this shader? I would love to play around with it. I will use your explanations to build my own version, but I would be nice to see the settings you used. Spanky: You got very good results with this shader, if gives a softer look. A bit less realistic, but very pleasing effect. What is the reason to connect the skin shader to the specular value?

Just pointing up there at Fyrespirit's comment on lighting, that sort of result seems to be a very common fault. I've come into Poser work by way of photography, and I've been known to wave an exposure meter around. Go look at some of the classic black-and-white movies to see that can be done with light. "Casablance" for instance. Even in a night scene they use the full brightness range. Something in the shot will catch the light. Yes, if you're scanning a photo you can do something to stretch the tones. But you need better than 8 bits per colour. Poser doesn't give you the extra bits. Three things to think about: 1: Is the monitor set right? You might have full-feature colour-matching software, but just getting brightness and contrast right makes a big difference. Try QuickGamma from www.quickgamma.de 2: OK, you want a dark scene, with simple lighting. Consider adding a spotlight to give that highlight Remember that Poser gives you more adjustments than real lights can have. Put it in-scene, close to the target, an all you'll see is the light it produces. 3: When you're setting things up, check a render in greyscale. The colour differences can mislead you.

Maybe not easy in Poser, but I recall they did some pretty fancy stuff to get the skin right for Gollum in "The Lord of the Rings" -- things like multiple layers. I think there was something about it on the Extended Version DVD of "The Two Towers"

I've just posted this to the RDNA forum thread meantioned above, but I'll repeat it here. You can make stripes with the Tile shader, using very different U and V scales. That might create a usable transmap for eyelashes (I used if fot a barber-pole effect).

What is the reason to connect the skin shader to the specular value? I missed? Just bad aim, I think. ;)

Mmmm, for some reason I became unsubscribed to this thread and missed out on a bunch of stuff.... Spanki - that render looks very very good. Looks like the left light is blue, and the right is red - by the spec colours...would be good to see the render using plain white lights. Also, IMHO there needs to be more detail on her skin. As for you node set-up....interesting. Firstly, from all the testing I have done, translucent appears to be IDENTICAL in every way to ambient. If anyone can demonstrate a rendered difference, let me know. Secondly, that is the first reasonable render I've seen using the skin node. Which makes me think how good it could be if you didn't use it! Thirdly, have you tried blue in the spec colour? And plug the bump map into it the blinn colour input. I also wonder if you could simply it a heap by just adding a touch of red into the ambient plug (since that is effectly what you are doing by pluging the skin coloured texturemap in). Quinlor....My current node setup is now way more complex than that posted above. I suspect it's got to a point of being too complex to be any use to anyone else. The highlights are: 1) A fresnel effect (adds spots to poly's that are at about 45 degrees to the camera) 2) Large and smalls spots are added to the texturemaps 3) Poly's that are 90 degrees from the main light have red added to the ambient and diffuse channels to simulate sub-subface scattering. This is probably the most critical, yet most frustraiting bit. It /almost/ works, but in certain light and with certain poses it gives odd results. 4) I've simplified the specular a bit - not it's just the bump map pluged into the blinn colour, into the alt-spec. 5) I plug a little Granite into the displacement channel. I've written a neat python script to set all this up for each head and body material. You select the "main" light from the python screen and it does the rest. There are 25 nodes per material. And no, I can't/won't release the python script. And Quinlor, I loved your work on ZB2 and displacement mapping, but I was damned if I could get it to work for V3's torso materials.

face__off: This things get complicated fast. I think the one I am currently experimenting with has around 30 nodes. And some effects still to do. Some comments to your ideas: 1) A fresnel makes a big difference. But I don't understand the purpose of spots at 45 degree. What I do is to reduce diffuse and add secularity near 90 degree. 3) That is definitive one of the most difficult parts. I am adding a little red to the darkest parts, a bit more intense orange to the in-between shades and a pale blue to the brightest parts. But I am still looking for a way to soften up the borders between the bright and dark parts without adding to much ambient. 4) A fresnel effect to secularity makes a big difference for me A other things I try out at the moment: Adding a little bit of highlights of a velvet shader. The trick seems to be to add the velvet only on the bright parts. About the displacement mapping: I have never tried it with V3. What's going wrong, the poser or the Zbrush part?

Well damn! I did a big response to Quinlor, and it must have timed-out. The general thrust was... 1) The spots at 45 degrees is to just add noise. I noticed that at 45 degrees, skin pigments seems to be spottier. You fresnel is similar to Riev's on www.cgtalk.com. I experimented with that technique but couldn't get the outcome I was looking for. 3) I assume you are using a ramp to achieve this? I used the ramp for a month or two (it's consistent with the Stahlberg method), but it required too much tweeking for different lighting conditions. I have have a set of nodes which get the dot product between the light angle and the poly angle and add ambient and diffuse red as the angle decreases. 4) This is the area I need to work on most. I was pluging an edgeblend into the blinn node, and got good results in some light conditions, but others looked poor, so I stopped using it. Tell me more about the velvet shader trick..... Displacement mapping: Played around with it a bit more today, and have worked out that the problem is in the concave areas. I thing displacement mapping in general does not work well in this situation. Example attached.

I am in a bit busy right now, so just a fast response to the displacement issue: It looks like your displacement map is not neutral on average, and therefore makes the mesh swell overall. The Poser displacement uses 0 (black) as the neutral value, and accept negative values for depressions. If you use a displacement map with gray as neutral, you have to set up math nodes to transform gray to 0 and darker tones to negative values: Just subtract 0.5 from the displacement map. How do you generate the map? If you use ZB2, you should be able to get the same look in poser as in ZB2. More comments to the shaders later.

Yep - done the subtract 0.5 thing. After a bit more playing around I removed all morphs and that improved it, however it still has small artifacts which are not pleasing to the eye. Generated the displacement map using DPix (or whatever the term is) of 3, and Subdivided 3 times, with a 4000x4000 map. I got a slightly better result in Maya, so I suspect its something going on inside Firefly. The fact that I haven't seen anyone successfully render V3 with a displacement map makes me suspicious.

Attached Link: http://www.renderosity.com/messages.ez?ForumID=12356&Form.ShowMessage=1996309

Quinlor, more on the Displacement mapping thing at the link above.

Arrr! I just noticed that renderosity ate a long replay I posted a here some days ago. I hate it ...