Daz 3D is part of 
Connect
DAZ Productions, Inc.
7533 S Center View Ct #4664
West Jordan, UT 84084
Licensing Agreement | Terms of Service | Privacy Policy | EULA
© 2025 Daz Productions Inc. All Rights Reserved.
Comments
I only have that article that I can't seem to locate right now (need to finally rename all those PDFs into something meaningful), where there are skin data represented as points on the graph. Many points, but damn inconvenient to use =) Do you happen to know of any tabulated source?
Made two presets. Clear and slightly 'frosted'.
View from different angles. Mainly checking blowout problems. The specular and/or reflection on the 'frosted' glass seems a bit too much on that last shot, but I'm still pretty happy with that.
16 should be enough, I think. I don't think I have the patience to test raytrace depth beyond that.
Yeah, but it's always good to document that. Hell, if omnifreaker bothered documenting his shaders properly, we probably could have accomplish all of this years ago. :D
That is a cool glass, and the higher ray depth dose look better. Odd tho, what is that blue smudge on the back-right side of the glass in the lower ray-depth render (lack of ray-trace diffuse being replaced by view-field background color?)
I vaguely remember someone doing a mirror depth test in 3delight some time ago, and the depth was not that great.
http://www.daz3d.com/forums/discussion/comment/259910/#Comment_259910
I can't imagine what the render time would be pushing it to far, tho then again, I think I have a vague idea with some SSS shader face-plant times (pre compute delay), lol.
Kettu, so it's nothing that would require PAs or users to go out of there way making adjusted 'negatives' of maps to use. That's about all that came to my mind when I first read that. If it's just a mater of internal stuff that has no bearing on stuff in the 'surface tab', then whatever. If it is something in the 'surface tab', then go with whatever it is on the format your basing the rest on (Maya, omnifreaker, Renderman, Crayola, whatever lol).
There is nothing like trying to port something over, and all the settings behave exactly the same except one value that behaves nothing like the source program/project/shader. Especially when multiple parts require the same adjustments that can not be done with an automated script, lol.
I'm going threw the same exact issue with an OT thing. Youtube vids play perfectly in firefox on my win7 PC, and the same exact settings and config in firefox on a Raspberry Pi2 is just not doing it (not even with HTML5), lol. Four days of fussing with settings and getting nowhere. I threw in the towel last night, and left it to the wizards that know that stuff on another forum.
Thanks!
It's probably the boat or the thing behind it. Doesn't matter though, since raising the raytrace depth shows it shouldn't be there.
The major drawback I see with is that the liquid and ice needs to have 0% opacity. Which is probably correct. You can still do some weird, interesting things with opacity color and translucence. The translucence will still honor diffuse roughness. Haven't tried it out yet though. I'm just ecstatic I can render ice/liquid in a proper glass now and still cast shadows. Would love to add caustics, even fake ones, but that means using Kettu's bounceGI script which means no IPR.
The option in the renderer's option is for max trace depth. Some shaders have the ability to cull depth and optimize performance. For glass/water/ice, I was using reflections on both US2 layers, but one of them needed to be raised to raytrace depth of 2 (rather than the default one). That's on the shader, just in case it isn't clear. At that setting, you should be able to get infinite mirror reflection if you set the renderer's ray trace depth option high enough.
But I think it's the refraction depth that's at work here. I found that in general you need twice the values to get all refraction.
Slightly off topic. Wow, those Image Engine guys are masters at compositing
I'd say very creative use of point clouds.
WOW... that info was hella cool, thanks for the links Wowie!
I think you saw those red/blue/green scatter/absorption sliders in AoA's Subsurface shader network? What we're dealing with is replacing those sliders with a colour box and a multiplier value, for ease of use and a more intuitive finetuning. For those who don't want to mess with those, I will be providing familiar presets (skin1, skin2, ketchup, marble... the Jensen values, I call them this way because they all come from that seminal H.Jensen et al's article on the dipole SSS model).
I was going to suggest making a photo reference, but then I thought that Iray should give a realistic enough idea of how the shadows/caustic pattern looks. Then we'll know if it's okay to have liquid and ice at 0% opacity (with caustics, it's wrong, but without, it may be a good workaround).
I keep meaning to try that glass with my glass shader (specular depth per-surface an'all that) and photonmapped caustics, but something always distracts me and I forget.
Basically outside of vastly corrected absorption, Wowie, it's still the same SimAbsGlass that's in the alpha kit. There's fake caustics in there (opacity colour tricks) that can be useful.
Thank you Luc Begin!
Those WIP shots are just so useful to figure out problems and workarounds.
I think my skin/character base is done. Scaled back on the 1st diffuse roughness so dark/light transition areas are smoother. Both eye surface and cornea have refraction (IOR 1.3) enabled now. I've also gotten around to plugging the diffuse map of the sclera in both diffuse channels. Should make veins visible with certain textures. Just love all those skin details in the close up shot. Probably should've raised the shadow samples though.
I'm thinking I'll make two versions - one with only specular highlights and one with raytraced reflection on both channels for the skin. Not everyone will want or have use of raytrced reflection on the skin. Render times with just specular reflection is around half (from 5 minutes to 2-3 minutes at around 800x1040, I think). The included shots ar 10 and 21 minutes.
Still need to recheck hair and color presets. Have to make scripts to copy specular maps to reflection strength slots and bump maps from 1st to 2nd layer, but that should be easy enough.
Maybe a value between 0 and 10% (the one I used for the glass). Above 10%, the refraction ends up too dark.
But photon mapping without your script is just too slow. I do wonder what the max distance value should be, since a strong enough light may still project caustics onto far enough objects.
Oh yeah. Really good tips from Betrand Benoit - http://bertrand-benoit.com/blog/the-photographic-look/
Thanks RAMWolff.
Nice to see you here too. How's that hair shader?
Hey Kettu, read this yet?
Physically Based Shader Design in Arnold
slides can be found here (along with other SIGGRAPH 2014 courses.
http://blog.selfshadow.com/publications/s2014-shading-course/
Side shelved for now.. I'm trying to finish up a product I have in the making for Dawn for Hivewire right now....
Sounds reasonable. The shots are pretty =) The only thing that throws me off is the UV distortion on her breast, but it's not your fault at all...
You could use my map copy scripts from the alpha as basis, the ones that are in the RadiumSolid folder. You'd just need to change the parameter names there.
And yes this is an official permission for you to do anything you please with what you get out of them; we even have witnesses here =)
Oh, right, I forgot in US2, like in DS default, it's the overall opacity, not just shadow opacity...
Even when it's only used for caustics? I confess I haven't used shader mixer photon mapping... for a very long time.
UPD: and the fake caustic thing is just a trick with shadow opacity. It shouldn't be that slower than any transmapped shadows?
Actually, photonmapped caustics doesn't have a max distance control at all. It's done with physically-based falloffs internally, so the effect only depends on the photon-casting light intensity and the photon opacity/reflectivity of the material.
And thank you very much for the links! The one by Mr (Monsieur?) Benoit should be stickied here, especially around the Iray-centered threads =)
Sticked?
How about engraved in stone and dropped on desks everywhere?
So, I went and grab Maya 2016 Trial and 3delight for Maya. Mainly to see the materials. The setup is pretty good. The combined specular/reflection controls allows for more simplified setup.
Base Layer. With low Reflectivity, it's pretty much just using specular. As roughness increases, the highlights spreads and at very high roughness, most of the highlights are only visible at the edges. Roughly starts the transition past the half way point. With a high Reflectivity, low roughness gives a reflective base layer. As you increase roughness, specular highlights becomes visible and the reflection decreases.
Paolo explained why they went for that scheme and not an IOR based fresnel here: http://www.3delight.com/en/modules/forum/viewtopic.php?f=4&t=4098&sid=bdab76e114758134117710f31a28f83b
Diffuse strength have very little visibility with high Reflectivity but are more pronounced with low Reflectivity.
I'd say that's a very nice setup. Most materials can be replicated just by twiddling three dials (diffuse) roughness, (Reflection) reflectivity and (Reflection) roughness. I also like that there's no strength dials. It gets combined into Color.
Reflectivity are also linked to refraction. You get very low refraction with high reflectivity and vice versa.
That's what I can tell by looking at the shader ball preview anyway. Haven't actually tested the shader on an object with a proper scene.
Already have one. I asked Casual for a script awhile back (before I got your kit) and he made one in less than an hour.
As I discovered and wrote about on the old thread, the ray cache parameter speeds up not only BounceGI but also Indirect Light mode of UE2. I forgot if I tested your script with shader mixer photon mapping, but I think it will speed things up too. All photon mapping tricks should benefit. Though they might not honor the bounce limits.
True, but I still like to have an override control. Look at the 3delight for Maya shaders. Lots of override to help you optimize or troubleshoot. That's on top of Maya's additional attributes such as bypassing GI, reflections, and/or refractions for a surface. US2 only have overrides for occlusion and ray trace depth on both reflections. I just wish there's a bypass AO/GI for the SSS precompute pass. Would've saved a lot of time with IDL or BounceGI.
The only thing I'd like to add to the Maya script is (Oren Nayar) roughness for SSS. But hey, I can probably cobble up a Hypershade network and make one. DAZ Shader Mixer doesn't even comes close.
Hehehe. I still don't get why a lot of people still use pure white in the materials, even with iray.
The Arnold one is pretty interesting. I like how they manage layers, but I wouldn't want to be the one tasks to write the code. Generous too, open sourcing the code for all. And of course, he actually describes Arnold correctly - "brute-force, unidirectional path tracer". Not that unbiased renderer stuff :). If you use Russian Roulette or MIS, you're smartly picking samples rather than sampling uniformly. So there's (good) bias there.
This bit I find particularly interesting:
- The energy conservation model is simple and easy to understand.
- Each layer is an infinitely thin, microfacet scattering interface, which can either reflect or transmit light according to the Fresnel function. Any light that’s not reflected or absorbed – according to the Fresnel function – is assumed to be transmitted. -
So we just evaluate each layer in a top-down order, integrating the Fresnel transmission as we go. - At the same time, each layer then gets the chance to reflect a portion of the total transmitted light.
- So the amount of light reaching each layer is solely dependent on the index of refraction (IOR) of the layers above it.
- What’s also important is that you can’t turn this off: the user only has IOR controls for the layers and that’s it. This means that the shader will always conserve energy and users don’t have to worry about whether they’re setting something right or not.
I'm not sure that ray cache can help speed up photon mapping. But of course, it's a question for the dev team because only they know the specifics of their architecture.
As for photon mapping in DS, you see, there is the problem in that shader mixer uses a camera to invoke the actual generation of photon maps. UE2 alone never generated the maps; it may be able to use them, if its IDL/bounce functions are the indirectdiffuse() shadeop, but it won't generate them. It's a RiOption level of things, and only cameras or renderscripts can do that in DS. Not shaders.
And my scripts don't support shader mixer cameras because, well, I got majorly confused about the various camera classes inheritance an'stuff. Maybe I should stalk Casual and see if he is able to figure that out LOL
So I simply added photon map generation calls to the renderscripts (that's why there are those whose names end in "caustics" - you need to use photon-casting lights, a special caustic light - all those lights are included - and those scripts to get reliable photon-mapped caustics).
Photon map generation has its own bounce depth controls, BTW.
And check out the scary complex caustics workflow in Maya:
http://www.3delight.com/en/modules/forum/viewtopic.php?f=5&t=4201&p=21579#p21592
This is where we have a slight advantage =)
The only "override" for photon mapping I can think of is using the "transparent" photon shading model RiAttribute (or, I guess, setting the "light" ray opacity to 0 inside the shader - should be the same). But I don't know if it actually will speed up photon map generation or caustic rendering...
UberSurface "occlusion" switch is actually diffuse ray visibility. It will exclude from GI as well because, well, AO is a crude form of GI =) The same raytype.
As for the SSS exclusion, nothing really that I know of. If there is something like that, it's undocumented.
Part of the possible speed-up for GI and SSS is using the "irradiance" param in the subsurface() shadeop call. It's what I do in my stuff, but you can't add that to older shaders without their sources unfortunately.
I looked at the code, and the subsurface pass uses the same Oren-Nayar value as the general diffuse.
Sorry, on a smartphone, so for the sake of sanity, I`m skipping quoting your post.
I do understand the requirements of photon mapping in DS, which is why I was surprised UE2 got a major bump in speed in IDL. Without looking at the source code, we can only speculate why. As I have noted, with your script, it`s fast. Super fast even. But it stumbles on SSS precompute. While it will not help any of the shaders available like UberSurface family of shaders, I`m curious if a bypass might help. Who knows it might even help raytraced SSS too. If it`s not roo complex to work out, that is.
Looking at Maya with its render layers, it is done simply by using SSS in a separate layer with all the lights except IDL. My guess is with precompute SSS, the saving is significant, more so if you have multiple figures with different SSS group IDs. Less so with raytraced, I think. Haven`t rried it yet so all this remans speculation at this point.
As for the roughness with SSS, I found having separate roughness adds another control that allows you to shape the diffuse.
Yeah, without the source it is impossible to come up with anything meaningful.
And on a side note...another PA just got bit by the control map gamma problem. I still say that ALL the control map slots (both in 3DL and Iray 'modes' be 'locked' to a value of 1 in the gamma slot. That is basically the 'no guess what it should be option, just use the blasted map'. Or force depricate jpg image maps...and then ban their use in any control map slot. Or both...(forced usage of 32 bit depth tifs would be ideal).
Tobor has a thread about this article...
http://http.developer.nvidia.com/GPUGems3/gpugems3_ch14.html
Mostly math heavy pbr stuff.
Hey Zarcon,
Maybe these will be useful. Sure, you still need to edit them in an image editor, but they are very high res.
http://www.cadnav.com/textures/
Hmm, isn't that SSS approach similar to what they use in Mental Ray? Paolo made a remark that it shouldn't be done that way in the 3delight For Maya forums.
http://www.3delight.com/en/modules/forum/viewtopic.php?f=4&t=4315&sid=130bb67d789a983365bc36b7f7f3da6e
Specular is still using Kelemen/Szirmay-Kalos, so not even close to GGX or GTR. I did a small experiment yesterday by creating a geometry shell and put that on top of the skin. Only one of the specular was enabled on the shell. With some tweaking, on the main figure and on the shell, I did come close to what I think GGX provides (wider specular tail while retaining the peak). Pretty good actually, but having real GGX or GTR simplifies the setup tremendously.
Yep...
And for some reason, more complex isn't always better. Granted more true to reality modeling of the process, should give more accurate results...but (there is always a but), for that to be meaningful, your mesh object NEEDS to be physically accurate, too. If you noticed the meshes used were scans.
Hey, why only TIFF, EXR is as good. I convert Mec4D normal maps to displacement for my use and save them out as EXR.
I kinda wish I had software that could make mipmapped EXR. These are what 3DL takes sans tdlmake these days.
That's because UE2 raytraces everything by default. If there are no photon maps or point clouds, all the GI shadeops simply trace.
In 3DL raytracer, the tracing algorithms are specifically optimised for efficiency, and when you add ray cache, it naturally flies.
I silently raise an eyebrow whenever I see people praise Iray for speed... on CPU. They just never saw the real speed =D
Besides... One of the reasons I prefer 3DL is that I can optimise. If I want precisely one specular or diffuse bounce for a surface, or exclude it at all, or if I want eleven specular bounces for that pesky refraction only - it is supported.
If I just wanted to bruteforce everything... but what's the point?
Layers, you mean AOVs or is it a yet another concept? In the code, there´s a hundred if-checks and none look like deliberately excluding SSS from GI on the first glance... but there's an extern call to some "Maya-specific" "irradiance_ray" which is hopefully defined in some other file.
So I can't yet say if it´s anyhow feasible or possible to repeat for DS shaders. You know that DS does have... quirks.
Oh speaking of DS quirks. I have finally conquered the DS analog of 3DfM "stangent", halleluja! So maybe I will even add normal map support. Don't hold your breath, though... but we can haz nice UV-oriented anisotropy now.
In the ShaderMixer, you should be able to have it, too: if you expose the advanced params of a spec brick with anisotropy and feed a vector based on the dPds variable to it (see the docs for dPds, it's the "stangent": http://wiki.daz3d.com/doku.php/public/software/dazstudio/4/referenceguide/interface/panes/shader_mixer/bricks/dzvariablebrick/start )
Yup, it's why I have it in my stuff =)
One of the reasons I plain out refuse to try and "fix" bad normals via faceforward() or even some clever ShadingNormal constructs. This will _always_ create issues where they shouldn't be, in my experience. The only thing sorely missing from DS geometry editor is the ability to _show_ normal direction.I've gotten so use to making my stuf 'fuzzy' in Blender, I keep looking for it in Studio...(Blender's show normals IS very useful).