Quote - > Quote -

 i mean, most people like a figure against a sunset to be fully and brightly lit from the front.  totally breaks realism for me, but i've been told several times that it's better that way.  same way i've been told burned in highlights in hair are better. 

i

and then i am wondering why they dont make a good update of poser.
the best ratings in the gallery are from renders where they dont even use shadows. flat shader on skin without specular.

and then i ask them for area lights and so on. they must be laughing at me.

OK I built a new VSS shader using sRGB as the final correction.

I'll show you several renders. In each, there are three figures.

Figure #1 is linear, no GC in, no GC out
Figure #2 is GC 2.2 in and out
Figure #3 is GC 2.2 in, but sRGB out

I should probably do a fourth with sRGB in and out, but I'm hungry and want to eat. :)

edited for cross post:  i see a significant difference by 15%.  and by significant i mean, if i were going for column 3, i'd spend days trying to get there from column 2.

Now I leave work and make FOOD!

i know i said one last thing, but another point.  you keep working with infinites.  but studio and indoor lighting, which i think is pretty reasonable to emulate, is all spots and areas.   which makes this issue much more relevant, because the shadowed areas are way too bright.  i know this sounds lewd, but look at the shadows on the one at 15%, especially under her right butt cheek.  huge difference, imho.  even at 50%, i'd much rather have the 3rd.

Anyway, I prepared a graph showing all the issues. Click for full size and read the notes.

Attached Link: http://en.wikipedia.org/wiki/SRGB

Key paragraph: (emphasis added by me)

Quote - Unlike most other RGB color spaces, the sRGB gamma can not be expressed as a single numerical value. The overall gamma is approximately 2.2, consisting of a linear (gamma 1.0) section near black, and a non-linear section elsewhere involving a 2.4 exponent and a gamma (slope of log output versus log input) changing from 1.0 through about 2.3.

edited because i didn't understand. sorry!  yeah, i can see why to test with infinites, i'm just saying you'd see a more significant change with images that duplicated, say, dramatic stage or studio lighting.

and just to say, to find your old test image, i went through the first few pages of VSS.  at least two people in the first few pages posted images that would be strongly affected by this issue.  and you posted about 2 or 3 others.

woot!  i'm really loving your low-light results.  they're just what i've always wanted.  is it me, or is there a big saturation difference, too?

I noticed the saturation too, even in the 50% lighting. Somehow the SSS is working better, richer color and more contrast. I'm not sure why.

 Bagginsbill, you're a maniac!
In a good way :thumbupboth:

Heheh. So Stephan, will sRGB conversion be added to Poser 8? Hehehe. 

BB, you're my god. There is a matmatic script to ease all this mess?

Hypothetically speaking, if there should happen to be a "make my scene sRGB" script at some time in the future, then my money would more or less leap out of my wallet of its own accord. ;)

And thanks for making me look up sRGB. In my line of work, I ought to know this stuff. Maybe it's time to retire. Then I can spend more time playing with Poser. I just have to break it to Mrs. Bob that we have to live in poverty for the rest of our lives, or until the kids can support us, whichever is first. :)

Lastly, my compliments to this thread's participants for their dedication to scientific reasoning above personal axe-grinding. That happens all too seldom in this (or any other) forum.

Quote - BB, you're my god. There is a matmatic script to ease all this mess?

You could use these functions instead of the GC functions.

def IF(test, tv, fv):
    return Blend(fv, tv, test)

def SRGB(x):
    return IF(x <= .0031308, 12.92 * x, 1.055 * (x ** (1/2.4)) - .055) def ASRGB(x):
    return IF(x <= .04045, x / 12.92, ((x + .055) / 1.055) ** 2.4)

 The SRGB function will convert a linear value to SRGB. It works with colors or grayscale.

The ASRGB is the anti-SRGB conversion - to convert an SRGB value to linear.

Example use:

colorMap = ImageMap("whatever.jpg").labelled("Color Map")
color = ASRGB(colorMap)
linearShader = Diffuse(color, .8) + Specular(WHITE, .1, .2)
srgbOutput = SRGB(linearShader)
s = Surface(colorMap, 0, 1, 0)
s.Alternate_Diffuse = srgbOutput
 

Uh... how I apply this? :blink:

pardon, but i'm wondering how one might put an sRGB filter on the Artisitic Lens?  and is that OK to use, or is it better to add it to each material and therefore deal with anti-correcting the map?  does it make a difference?

The script I posted above is wrong. I forgot that the Blending value on a Blender node is single-channel. We need to have a blending factor that is done once per channel, not for the whole RGB triple. So I built the filter with more math nodes because I can't use one blender to choose between the two ranges for all 3 channels.

As to whether it is OK to use, that's questionable. The difference between GC(2.2) and sRGB is very small, and only applies to very dark things. If you don't anti-GC your incoming material, you're getting a much larger deviation from reality - more than is compensated for by choosing sRGB versus GC. If you're looking for that kind of precision in the color space, you really should be working with anti-gamma incoming material. Otherwise, I wouldn't bother with the sRGB filter. If, for artistic reasons, you're finding your darks are too light with GC(2.2), try GC(1.6) or some other similarly lower value.

I wish Poser had implemented the GC as shader nodes - something we could configure as a whole. Then I could drop in the sRGB filter as a built-in pre-processing and post-processing step automatically for all materials. That's how I would have done it. Then it would be possible to configure the renderer to produce any color space, such as the Adobe color space, and we wouldn't have to muck about with shader nodes. Even better would be if it were possible to set up, using nodes, various pre and post processing filters and then select them for each material or image map. Or insert a node that calls upon these as subroutines. Then you could mess with them in one place.

Hmmm. I have so many interesting ideas for nifty plug-ins. if only the renderer had a more modular plug-in architecture for image pre-processing, materials, lights, and post-filters, as well as user-defined nodes. Sigh.

The only place you'll find a difference is on the dark gray prop, and a few places on the lower and left edges of his shorts.

thing is, i noticed what i considered pretty big differences in all your previous renders.  including the one with the light at 50%.  for me, it's the difference between all the pretty (imho) unrealistic mid-tone renders people have been doing with VSS PR3 and reality.  is it  better to implement it in a material?  because, imho, it's killing my work in anything but really, really bright light.  the difference between my white or bright room renders and anything shadowed is... well... night and day in terms of realism.

it's not just dark materials, or a dark scene.  it's mid to low toned shadows.  it's why that person said implementing VSS made their work flat.  to my eyes, all your previous tests showed a very significant difference in just about anything with meaningful shadows (on a self-shading object, i think).

enough so that i'd rather work an sRGB shift into my materials than GC, if i can figure it out.  i'm pretty sure that it's what had me fighting even after 6 iterations of IBL creation and using only GC materials.   i'll take the time up front over several days chasing my tail, so to speak.

i've been focused on modeling for a while, and not rendering (so definitely not materials and light).  i was hoping to be able to deal with a new version of Poser, VSS PR3, Matmatic and a new workflow all at once.  it seems like that might not work out.  then again, i'm still learning so much in Blender.  i'm (sort of) applying Adorana's technique for making clothes, and adding some of my own findings.  Blender is awesome, but it isn't Clothilde.   

so probably the best thing is to test PR3, my adjustment of PR3, and then a scene with sRGB materials on my last render.  would i be able to ask you questions about using VSS, PR3, or the sRGB component?  i know the information is all out there, but i'm pretty sure i'll hit snags.

Would you like a VSS sRGB skin shader? I could set publish one pretty easily.

well, definitely yes!  but since i'd also need to shift the materials of everything else, i'd still need to learn how to incorporate it into other materials (i say admonishingly to myself).  i'm very sure the hair shader i've been messing with for ages would be better in sRGB, since the most important aspect of shiny hair is the contrast between its light and dark portions.  usually, when people say hair doesn't have bright enough shine, what actually needs to happen is the dark areas get dark enough.

actually, that looks like has exactly the issue i was talking about, but maybe more so.  though it still looks phenomenal to me.  i can certainly say that when i post images with hair shinier than that, all i get is feedback that painted highlights would be more realistic (eww) because it's not shiny enough.  sometimes there's no accounting for taste.

Quote - Would you like a VSS sRGB skin shader? I could set publish one pretty easily.

please!

oh and I'm quite sure someone said they needed the hair shader.

I think we will be held in suspense for the hair shader.

Quote - I think we will be held in suspense for the hair shader.

I think too !

For anyone interested, I've only recently been properly catching up on a lot of bb's excellent shader work. I was playing with the VSS PR3 skin shader in single light (non IBL) setups and noticing what seemed to me like quite harsh "transitions at the terminator" (terminology I'm just learning) - basically where the light transitions into dark on a model (see above image to get an idea what I mean).

The sRGB VSS skin shader would be very cool if it goes some way to solving this.

Now I have to read this thread from the beginning to see what this artistic lens is all about.

bagginsbill, could you post the setup that you used to get that first render? The newspaper comic-y one?

I can't seem to duplicate that one.

Thank you so much!

Quote - bagginsbill, could you post the setup that you used to get that first render? The newspaper comic-y one?

I can't seem to duplicate that one.

Thank you so much!

BB posted the setup for me some time ago, look the post #8 in the page #5

www.renderosity.com/mod/forumpro/showthread.php

oh! thankyou!

Book Marked!

Quote - The script I posted above is wrong. I forgot that the Blending value on a Blender node is single-channel. We need to have a blending factor that is done once per channel, not for the whole RGB triple. So I built the filter with more math nodes because I can't use one blender to choose between the two ranges for all 3 channels.

You wouldn't happen to have the asRGB/sRGB script lying around handy anywhere, Bill? I mean, the revised one... the one you wrote for the artistic lens. I'd be really keen to see it.

I'm not the sharpest tool in the shed, and have been using the above script (which you said was wrong) in my renders as an alternative to GC, just to see. Might just be me, but I kinda like the results better.

So, how would you separate out the colour channel? Component?

BTW, finally converting my skin shader to mix.mm1. There were some massive holes in my understanding of classes and that sort of thing which cobaltdream quite laboriously (well, I'm thick so it took a bit of doing!) helped me with, so I'll be bugging you soon with more questions about that... :laugh:

wow, i'm dim.  this is why i had that complicated IF function. you need to do it with Color_Math not Blender.  ok, looking for it in thre forums.

just to say, i'm trying just the simple Python if condition: .... else: .....  is that OK?

edited to add:

oops, it's definitely not.  because when i do it that way, it automatically uses color math.  which seems to mean (glancing through the nodes of the resultant material), that certain over 1 nodes don't get made.  gotta figure this out before going further with my experiments.

ok, just a note to say i've forced numbers using Add nodes, but the if structure just didn't translate.

well, i seem to have something that might work

def asRGB(x):
    ifMask = Color_Step(x, 12.92 * 0.0031308)
    xFalse = 1 - ifMask) * x
    xTrue = ifMask* x
    return xFalse / Add(12.92, 0) + ((xTrue + .055) / Add(1.055, 0)) ** Add(2.4, 0)

def sRGB(x):
    ifMask = Color_Step (x, 0.0031308)
    xFalse = (1 - ifMask) * x
    xTrue = ifMask* x     
    return Add(12.92, 0) * xFalse + (Add(1.055, 0) * (xTrue ** (1/2.4))) - .055

edited to reverse the mask because i'm really not swift at this and easily turned around

Really cool idea. BB I'd like to see what you could come up with in FilterForge.

Mathman, why are you using blue tinting in Alt_Diffuse?  Is this for the skin?

Quote - I have just discovered this fantastic thread. I started playing with BB's most basic shader, it made a difference but left M4 looking "demon possessed".

Can anyone suggest a solution to my problem here ? .... rather can call in an exorcist, that is !

Did you see my post above?

Does the eye cover involve a refract node? If so, do you have enough raytrace bounces (2+) to get through the lens and the eye cover?