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
Na, just texture. It's got the metal, and the shape. there are no machining marks, and looks more like a smelted featureless ingot then a manmade object.
Then again, even that has texture. (shrugs shoulders) I don't know.
The color thing, it was just my impression of what alloy it should have been. Each metal dose have a variety of color ranges depending on how much and what impurities are in it (not to say oxidation or other coatings).
Sorry if I'm wrong, but I'm not completely sure you understand the physics here. I haven't found any useful and concise explanations, so let me try. Simplified as much as possible.
The same Fresnel laws apply to all types of materials, but the IOR of dielectrics is merely a real number. Like 1.3. For a metal, the IOR is a complex number: a real part (n) and an imaginary part (k). The imaginary part is called "extinction coefficient".
IOR = n + jk
Using this approach, you can model the correct look of any metal.
"Custom Fresnel curves" are just one of the ways of getting to the same physics.
Take a look at the DB: http://refractiveindex.info/?shelf=main&book=Au&page=Rakic
The general caveat is that IRL, these IOR values are different for different wavelengths (play with the wavelength in the DB and see how n and k change). This is the only reason why you'd want "custom curves" over built-in 3DL metal IOR support: the built-in support does the "three-tuple approximation"; i.e. the "wavelengths" are red, green and blue. It's perfect for greyer metals, but for something with more colour like gold, you may want better spectral discretisation than just the three basics.
The paper I keep linking to provides a way to forego looking up the n and k in DBs and just use photo-sampled colours to match the look you need: jcgt.org/published/0003/04/03/paper.pdf
Here's a dragon with the colours for gold taken from that paper and an HDR from the sIBL collection.
Me for half the posts in this thread:
Glass is not the same as opacity-mapped cheats like lace. Good glass is refraction-based. Refraction is a raytraced effect = needs the raytrace hider for best speed.
Now, opacity maps. I have no idea how Iray handles this (and if it has any optimisations at all), but for 3Delight, user interference is crucial. Excluding these surfaces from GI or limiting diffuse bounces on them will drastically decrease render times. And of course, for something simple like soft RT shadows, this is also solved by using the raytrace hider.
These will always be slower in a given "conventional" (non-voxel) renderer than its surface shaders - because of the amount of computations. SSS is technically a volumetric effect, BTW.
Use your mathematician hat: mesh density will affect the normal angle per each face. Shouldn't be that way for Sub-D surfaces, but for poly meshes, it's a fact.
We're talking antialiasing here. It's a complex mathematical problem. And one that doesn't have a definitive one-size-fits-all solution, unfortunately.
There is a chance that I may be able one day (DAZ Soon (tm)) to reproduce this for DS:
https://3delight.atlassian.net/wiki/display/3DFM/Rendering+Displacements
Kinda pwns just scaling those normal maps, huh?
Cuz it's da lab here ;)
Nothing too complicated, though. Mostly geometrical optics.
I've been greatly informed by the bits I understand. ;)
lol. Don't feel to bad, I'm looking up a lot of words, and filling lots of sticky notes. And losing some of the sticky notes as well (where did that note go), lol. For example, Some where I have a note of what IRL and IOR is, and I can't find that sticky note, lol.
Thankfully, I have a relative familiarity with scientific and logical stuff, it's just jargon that throws me (and some of the more convoluted math).
I'm with you, I did learn some things since I started reading this thread but really following ... no. Just hoping by picking up bits and peaces i can slowly fill in the bigger holes I still have
It excites me because I'm like 'go go team figurethingsout! I can't wait to see what you end up with!'
No, I understand the formula.
What I'm saying is that real world materials have much more 'noise', which generally couldn't be accounted for. Here's one example - you're matching a rendered chrome ball with a chrome ball probe used in your HDRI. Even if your workflow is 'correct' (camera settings match, linear workflow etc), there will always be deviations. You just need to figure out where and how much are there and workaround them. The common practice is to match the backplate as close as possible, so if you need to alter things in the shader/material settings to match it, you do it. Hence the need for custom curves.
Look at the Disney's principled BRDF paper again. Even they mentioned that the MERL database have some glaring issues with the data. They simply discounted those as errors during capture.
Ah those. Yes, those are generally done as textures, via diffuse or specular/roughness maps. What's the word? Grunge? My render didn't use any, since the textures that came with those props are very low res. If you want those, go with the painted metal settings, since they allow diffuse maps to be used.
For finding shader settings, just type the word into the search field. It will filter out settings that do not match. Best workaround I've found. Of course, you need to familiarize yourself with the terms used.
Changing settings via a graph is way easier.
Qt can do it, hell, even DS can do it (just look at the graphical editor of d-form strength).
As for US2, it's a compiled, encrypted shader. Making presets (with and without maps) is easy enough. DS even allows you to customize what sections you want to save (and override).
This is a scenario I frankly didn't even think of in the contest of DS. I just can't see DS being used for serious VFX work that would require precise plate matching. It would need so much more than just new shaders.
So what you need is a widget for the surfaces tab that would be able to take points and translate them into an array for the shader to pick up. Ideally, this widget would also be supported by the shader builder. If it can be done via a plugin, then there is hope. If not... I'm not sure it's the sort of a feature request which would take priority for the DS developers.
Hey, but if you know "lol" you should know "IRL" too ;D
Jargon and holes... if you folks could list the parts you find difficult, it could help with expanding and improving the "mini-glossary" in the first thread post.
And maybe we need a microcontest about the best portrait of A.J. Fresnel =D http://micro.magnet.fsu.edu/optics/timeline/people/antiqueimages/fresnel.jpg
Yes, it should be doable...
Now on to using the N-K data for metals. There is one thing about most published N-K data...TMI. It usually spans from the very far edges of the infrared through ultraviolet and sometimes even beyond. For artistic work, that data needs to be trimmed to just the visible portion.
One of the nice things about Luxrender is that it can directly use that data (and I've hand crafted several data files, mostly transcribed from pdfs). It would be nice if there was an easy way to 'crunch' that data so that 3DL could use it. But there isn't because there is no 'standard' for the number of points that can used for the visible porition and no 'set' frequency of the measurements. So just creating an array of 10 values wouldn't really work (10 IOR/frequncies would be much better than the 3). But a plugin that can create a custom curve should also have the ability to accept a data file that already defines that curve (the N-K data...).
And once that data is in a curve, you could pick a number of locations to feed the shader and that number would be constant (like a dozen points fed to the shader or something). No need to worry about the variations in the data (as few as 7 to over 100 in the N-K files I've worked with). No, it won't be 'exact' with few points, but it sure would be a hell of a lot closer than 3.
So it would also be an interpolator essentially. Awesome idea! Damn, it's so frustrating that I could write this (and much more) in theory, but in practice, I'm just looking at the DS SDK thinking "now what?" - if only there were tutorials for writing DS plugins.
Maybe we could convince mCasual that it's the bee's knees.
I'm in the same boat...I know what needs to go into it, but look at the SDk and my brain starts melting. (I've thought about this for quite a while, because being able to use N-K files would be fantasitic.)
Yea, that's about right. I'll copy a side-by-side from my thread for this.
for 7, yes, antialiasing. It's all them swirls and such.
And there is a 'Pixel Samples' X and Y thing that helps to an extent, tho it mostly deals with star-casing edges in a render rather then the Moire Interference Patterns (posibly because of what 'Sample' combination method is used). This is one of them not so well explained set of render dials, and setting them to high will make doing test renders for setting up scenes, impossible. If this actually is a form of antialiasing, I think My old "3Dfx graphics card" had more options then this. Settings, 8 is ok (for setting up stuff), 12 is painful (tho looks better on some stuff), 16 is impossible to get anything done (today), lol.
9, I'm sure is a slightly different issue. It has to do with how stuff losses the surface appearance, as more surface texture pixels are crammed into smaller numbers render pixels. for example, up close the texture makes the red triangles brighter to the left and right of the render...
Yet when backing up some, it loses all hint of the texture, not just the directionality of the "texture gloss thing".
It is almost as if the voxel/whatever thing is averaging out the value of all the points on the maps, and ignoring things like grain direction in the process. It's the very reason for having one set of "Maps and Stuff" for close up renders, and another set of "Maps and Stuff" for distant shots.
Yes pixel samples are part of the whole AA suite, for lack of a better word. In layman terms, higher pixel samples = better quality overall (less aliasing, less noise). There is sadly no simple way of explaining how they work. Actually the whole interaction of pixel sampling, pixel filtering and shader-specific AA is a very complicated mathematical topic lying at the very heart of digital image synthesis. There is a legally free e-book by Andrew Glassner... it's actually an amazing and super useful book, but these sections are not for the faint of heart:
http://www.glassner.com/portfolio/principles-of-digital-image-synthesis/
Ouch, no voxels here, voxels are "3D pixels" and they are currently used only in very specific applications.
But yes there is a sort of "averaging", and it is twofold: 1 - antialiasing via the filters in the texture() shadeop call, and 2 - the textures being mipmapped ( https://en.wikipedia.org/wiki/Mipmap ).
And then actual image reconstruction comes into place - pixel samples and pixel filtering - and it's all so very very complex.
But it's a fact that without texture()-based AA and mips, there would be a lot more scary moire and random noise everywhere.
There is another problem with rendering fabrics. They are NOT flat or uniform. They are volumetric by themselves! They consist of fibers weaved in 3D space.
A few renderers support experimental weave generation models (like Mitsuba and probably later versions of Lux - Mjc?), but it's still relatively rare. I have seen successful fabrics "built" out of particles. 3Delight has supported particles for ages, but DS cannot generate them. There is a chance we will get a plugin eventually, but who knows.
What I do for fabrics is make sure every channel uses strength maps (and the maps are crisp enough): specular, velvet, translucency. This helps maintain a more bellievable look (strength maps "simulate" that volumetric inter-fiber shading).
PS Zarcon, what are your pixel filter settings? If you are on "sinc" at 4x4 or wider, it's a filter so sharp that it will cause "ringing". Maybe it's what makes your moire problems worse.
Progresive render mode, the setting further down are 'inactive' or whatever in that mode. It's one of them that I have no control over, I don't want to wait days for each test render kind of thing. I'm looking at setting up stuff, not doing final production art, lol.
(EDIT) I think it's still 1x1 Box filter, I think. DS hasn't popped up that warning after 4.6 I think it was.
Some times, you can see just how bad something will be, just with the OpenGL view field in DS, lol.
No wonder your renders take nigh on to forever...a shading rate of 0.01 is absolutely INSANE! And for Sinc 8 is too high...for filter width.
I'll start looking threw that, looks good.
I think some cloth patrons are just Prone to loosing detail or getting Moire Interference Patterns. This is a simple satin weave with 4 pixel wide threads.
And shrink in half by Irfanview (Lanczos filter, whatever that jargon is, lol).
And at 25% size.
Even at 25% size, there is still texture there. Yet with renders that would be completely featureless after that mipmap-thing.
And I can keep going to at least 10 percent size, and the texture is still there when shrunk with that Lanczos filter doohickey thing.
Shading rate below 0.2, well it's for the floor (the polish reflections).
http://www.daz3d.com/forums/discussion/comment/586709/#Comment_586709
And as said, in Progresive mode, thoes settings do nothing, they are completely shut down, disabled, turned off, bypassed, etc. There not there for Progresive render mode, lol.
It's OK, there are lots of dials and buttons everywhere and it is easy to overlook something important, like the altimiter if you can find it, lol.
I know the feeling, lol. Buttons and switches and levers everywhere, including buttons and switches and levers On the buttons and switches and levers. It takes three people to fly this thing, there is just one of me
.
The pixel filter is 'box' for Progressive..and 1 x 1, but ALL the other settings are still active, especially the Shading Rate.
And Max Ray Depth is going to do more for the reflections than a Shading Rate at such an insane setting.
Properly set up lighting, shaders and gamma correction, a shading rate of 1 will be fine for all but the most 'critical' final renders (and a shading rate of bellow 0.2 shouldn't be needed for anything other than super fine...like trying to do baby hair by displacement...displacements).
Here's a Luxrender cloth material...no normal, bump or displacement...
It's using the Denim preset at 100 x 100 thread count...
Rendered to about 250 s/p (in about 7.5 minutes).
This one is the same denim, but at 200 x 200 thread count...
And this is the silk Shantung at 600 x 600...
Theoretically you can have any type of cloth that the measured data exists for, but in practice Luxrender only has six included.
Going way back to April of 2014, lol. Getting rid of grainy reflections. 'Max Ray Depth' only controls the number of bounces, it dose nothing for the quality of the reflection. The times were on the old 2.0GHz 2-core CPU from the dawn of time.
This image; Time 3 min 47 sec, Shading Rate = 2.00
I can live with the time if I have no choice and something else away from my computer I must do, however the reflections are a 'Fail'.
This image; Time 5 min 25 sec, Shading Rate = 1.00
I don't want to wait this long for adjusting lights, and the reflections are a 'Fail'.
This image; Time 35 min 43 sec, Shading Rate = 0.10
notes;
This was painfully slow to render, yet I am impressed with how good the reflections did come out, even the third reflection of the galaxy is impressive.
Now only if the G1 Bare Foot Dancer base didn't look inherently plastic like.
O.K. Painfully slow is an understatement, I need a dedicated National Laboratory CRAY to do this on, so I can actually use my workstation as I wait for the render to complete. I have no intention of ever waiting over half an hour for a render to run to completion again, I don't care how good it looks, I will settle for "good enough" for now on.
< Whozzzap! Back to today.> I have eight CPU cores now, cruising along at 4GHz, and thus it no longer takes so long to render something.
Assuming the figure's SSS settings are not set for simulating the magnetic field threw the center of the earth rather then something a Photographer would care about, lol.
Very nice looking. And the ability to mix up the cloth types I will assume is either beyond the capability of the 'material property language', or limited only by your imagination.
Because when it comes to cloth, there are many ways it can be aranged in different patrons.
Original Photo from, http://en.wikipedia.org/wiki/Indigenous_peoples_of_the_Americas#Bolivia
Tho getting that into DS for 3delight, Hmmm. I still think Maps would be easier for making some of those patrons then crafting complex mathematical formulas "to represent" them.
It dose after all come down to me just wanting to have fun, rather then sifting threw "Strings, Conformal Fields and M-Theory" looking for the proper formula components to represent threads in spacetime, lol.
Whoop, I didn't even kow there was a person behind the fresnel thingy, well portrait I can do...
Ok I started collecting terms that are not clear. As I'm trying to get thought the old posts of the thread I will keep collecting and I guess some will be explained somewhere, so I gather what I can get and post that here and you geniusses here maybe fill the open ones. for the mini glossary (won't be so mini anymore I fear)