Adding to Cart…

Licensing Agreement | Terms of Service | Privacy Policy | EULA
© 2025 Daz Productions Inc. All Rights Reserved.You currently have no notifications.
Licensing Agreement | Terms of Service | Privacy Policy | EULA
© 2025 Daz Productions Inc. All Rights Reserved.
Comments
Flame is actually two area lights, I had to copy it because apparently area lights aren't two-sided?
And the cat DOES have LAMH fur. I think the settings could use work, though. Maybe more translucency? Hrm.
Yeah I do...sorry.
Ok...
Shading Rate: the 'multiplier' that is applied overall (modified by individual surfaces own shading rates). Since it is really a divisor that means a fractional Shading Rate increases the quality/number of overall samples. It's really a bit more complex than that, though...
Pixel Samples: The number of 'samples' per pixel...
Filter width: the filter control...
I've got them straight in my head, but unless I'm staring at them when typing, I'm likely to mix them up...don't know why (I've done it before...).
Increasing Samples isn't needed in most cases 4x4 to 8x8 should cover everything but DoF/motion blur. Nor are 'extreme' Shading Rates, simple surface adjustments have as much or more impact on the look, are often easier to control and have little impact on overall time (sometimes reducing it).
For the filter width...box, gaussian,triangle will blur like crazy going much higher than the default values. Sinc and Catmull-rom will 'ring' going higher.
here there be dragonz
The joys of thin film! It's designed to be multiplied over a surface. It takes metallic complex IOR as input, the "simple" IOR of the "oxide", and thickness. You can see how it looks over a metallic reflection or over a basic diffuse (examples with no thin film are also attached; I used descriptive filenames... I think).
For realistic tempered metal, you should use the same metallic IOR as your underlying metal and the oxide IOR of about 2.5 for steel. Many oxide IORs can be found here: http://www.nist.gov/data/PDFfiles/jpcrd623.pdf
Here is a small jerky (4 fps, hey) animation of the MilDragon with the "right" ferrous oxide param (looks heaps better after you click "play"!!):
Or you can use different metallic IORs for your metal and film. Or, well, whatever.
It renders very fast, BTW. I didn't notice any gain at all.
I'd also add curves (doesn't matter RiCurves or thin enough fibermesh) to the high pixel samples scenario.
Maybe it's not long enough to be noticeable, then? I used the "long fur" preset from here once, and it was very glowy:
http://www.furrythings.com/presets-animals/
But as you can see, for a healthy cat it needs more actual hairs than the default preset gives.
EDIT: You are right, mesh lights are one-sided.
SWEET!!!!!!!!
This changes everything. There's another 'it can't be done in 3DL' shot down.
Yup =) The only two things really missing are a) bidir light transport for those RT caustics, b) a built-in way to add dispersion without manually splitting "wavelengths".
I think I used short hair because the model is at about 35% scale (the default cat is as big as a person).
I also used, I THINK, 1 million hairs. Ultimately, the problem is probably the texture.
It does work as advertised. Pretty much similar experience to IPR in Maya. I'm still in 4.7 btw. I can watch vids in a player or within a browser, listen to music, have browsers with lots of tabs open, with downloads running in the background. And that's with an 8 GB machine. If you're getting BSODs, then there's a serious problem with your setup.
I'd bet most of your problem stems from using DS default surface shaders. Solution is simple - don't use them. Ever. Use at least UberSurface.
Don't forget the model's distance from the camera plays a part too. I did some test awhile back, mostly debating Kettu's choice of using SSS scale of 0.1. For a render where you can see the entire figure, SSS shading rate need to be adjusted (more accurately, lowered) to avoid artifacts. For head and torso shots, 2 or 4 is OK. Further will be 1 or 2 and faraway will be lower.
Hmm.,
Indirect lighting with directional shadows (with HDRI) is way faster than with soft shadows. There's no directional shadows without a HDRI (UE2 will fallback to soft shadows). Didn't try with BounceGI though, if I remember correctly. You don't need to use two ambient light, you just need to properly balance ambient and indirect light. My mistake was thinking I can just turn on indirect light without changing ambient light. If you use indirect light (even at 100%), you need to lower the ambient light strength a bit (usually one third of the strength is sufficient).
Don't try doing indirect light or bounceGI without Kettu's script though. Best combo so far is DS 4.7, UE2 IDL and Kettu's script to enable raycaching.
Yeah, working with better lights and coming back to 3DL I've quickly realized 'convert everything to omUberSurface' should be step 1. (Or PWToon or whatever else I'm using)
I'm also realizing how cool some of the options are; Velvet is AWESOME.
Which reminds me... AoA Subsurface or omUberSurface? Subsurface seems to have a lot of fiddly stuff Uber doesn't, but I think I've noticed Uber renders WAY faster... am I right?
I've encountered something interesting, sort of stumbled on it really...the default Catmark subdivision algorithm really does make a negative impact on render times. Especially with the AoA shader...not sure why, But switching to the 'legacy' Catmull-Clark sped it up significatntly. I did a simple test with extreme subd levels on a simple 6 poly cube. The Catmull-Clark in both Studio and as a RIB rendered in around 12 secs at a subd level of 9...Catmark about 10x slower in Studio (no change in RIB, because it is passed as Catmull-Clark to the RIB).
For something using the AoA shader, I used a naked G3F default load...3 min at Catmark...under 1.5 min Catmull-Clark, so about half.
Yes, Catmark is an OpenSubDiv algo...does 3DL actually support OpenSubDiv (I know it didn't when the change to Catmark was made the default)?
Does Iray support OpenSubDiv? If so, that could explain why the change was made, but if it does then why is it so memory impacting to use?
Isn't OpenSubDiv supposed to be 'better', more efficient, etc?
In your opinion, does the legacy subd look noticeably worse?
UberSurface2. =)
And you could be right about speed, shader mixer networks like AoA's tend to be slower than precompiled ones like those of Omnifreaker's. Or it's the case of a larger SSS shading rate being used by UberSurface by default.
The texture on my cat was either from the standard MilCat package or some freebie... And we're using the same cat, so mine is scaled down too.
Well, I don't have UberSurface2 YET... hopefully in next purchase wave in a few days.
Trying to make my final buy list of really useful things -- my PC membership ends 3/11, I set aside money for purchases every 6th, so that's two more purchase waves until I'm done here.
Which reminds me, any broadly useful 3DL content available elsewhere? Stuff on the order of great lights/shaders.
Not noticeably different...there may be something, somewhere, but I haven't done extreme closeups and run any image analysis, so just going with eyeballing it...not that I noticed.
The results of the two algorithms are so similar, there is a reason for the switch (could be the HD morphs need OpenSubDiv to be importable........)
I actually noticed this, too. But I thought I was hallucinating or something. Even that sinc example sphere I rendered today with nothing but DS default stuff, it slowed down enough for me to notice when I subdivided it.
Google returns nothing re:OpenSubdiv on the 3Delight website. Their features page doesn't seem to suggest support either: http://www.3delight.com/en/index.php/3DSP_features#fourThat's what I thought...so there has to be some other reason (my last post, maybe?).
The cube in Studio on legacy was just a hair over 13 seconds and in RIB just over 12, the first run, but averaging several runs in both it was just a bit over 12. But it was consistently over 2 mins on Catmark.
In other stores? Nope, not that I know of.
Pleas forgive the short reply, tad bit busy here. It is odd, in that it maters not what shader is in the scene, Daz Default, Omni, or AoA figures. Regular renders are fine, spot renders in the view field are fine. Yet as soon as I try to move dials with IPR going, it breaks. If it was a mater of shader in a render vs another program, I think it would happen regardless of where the render is happening in Studio, not just when IPR is going and I'm trying to adjust something.
You could be onto something with the HD morph idea.
But basically, what you're saying is that whenever we export to RIB, we get "standard" Catmull-Clark, correct? It's interesting because I also noticed that LAMH-to-RIB renders differently than just in DS. Not necessarily related to the subdivision thing, but just shows that the RIB generation and the renderer API can receive different data from DS.Yes, it was easy to find with just a cube...then I looked in a couple of others using that at search string.
And I think Kendall made a specific RIB script for LAMH (at least I think he was the one that did it).
Picked up a bunch of stuff... UberSurface2, AoA lights, and a few other things.
Really enjoying them... though I'm seeing a number of cases where I won't want to bother with UberSurface2, particularly with items having other specialized shaders. For example, a lot of later skins I have use AoA_Subsurface. I tried converting to UberSurface2... and, frankly, it's more trouble than it's worth. (Velvet and Subsurface stuff doesn't translate, so I have to reset each).
I hear you. I have some difficulty with the newer AoA HD figures on my 'Pathetic' computer. When some parts of the eyes don't keep the maps, and some things spontaneously get 100% reflection strength (teeth), it is a lot of work to convert them over to a shader that dose not take bloody ages to get threw the Sub Surface Precompute delay.
Simply turning Sub Surface off doesn't help much at all on AoA, especially with some shaders using Sub Surface heavy skin tinting (Diffuse strength under 50%). Thankfully, some PAs appear to have stuck with a consistent shader setting envelope of sorts, and do most of the variation between figures on the maps for that figure. So I can make a single "Alt Shader" for the newer FWSA figures (for example), and use that same "Alt Shader" on almost all the FWSA figures. I just click the Alt Shader preset while holding the 'Ctrl' key, and next to 'Images' select "Ignore" in the window that comes up.
Here I'm using a shader preset that is still a WIP for Paloma, on FWSA Finley. The result is in my Art thread. Not exactly like the promos, yet it still looks very good.
It is not a one click solution, as some times the eye maps need to be put back in for some figures (hence that extra 'eyes' sub folder).
Playing with AoA lights, grass shader, ubershader2 (for the dress), and a few other things.
http://willbear.deviantart.com/art/Against-the-wall-589125046
The grass seems to have a lot of ambient? The dress looks really nice
There's no ambient, but there's a UE2 ambient thing going on that might be a little too high.
But manually reworking stuff is the only way to learn and especially develop your own consistent visual style, independent on whatever floats the vendor's boats =) Besides, with US2 you don't really need "velvet" on skin materials because it's just a hack for faking correct Fresnel attenuation of specular, and US2 can do true Fresnel natively.
Hmm, interesting. Do you remember where it can be found?
And how I wish for a Garibaldi-friendly render script.
I did some testing. Not with G3 which had the OpenSubDiv thing in their adverts, but with the original Genesis. Don't have the renders handy, but the interesting thing is, everything works the same with the "legacy" catmull-clark but the HD morphs. They just don't show up much, only a hint. For these to work, the Catmark is mandatory. And exporting to a RIB works fine afterwards.
So I guess it validates your point.
And you don't really need velvet on EVERYTHING, like I've seen some presets load with...yes, there's velevet on the eyeballs on a couple of them. And the teeth....
Velvetty teeth. Ha!
Having a bunch of different shaders gets you into some interesting trade-off debates. Like I have Shades of Life projection shaders, Dirt shaders, and a great tiling shader, but they lack qualities US2 has, but they do cool things, so...