Daz 3D is part of 
Connect
DAZ Productions, Inc.
224 S 200 W, Suite #250
Salt Lake City, UT 84101
Licensing Agreement | Terms of Service | Privacy Policy | EULA
© 2024 Daz Productions Inc. All Rights Reserved.Daz 3D is part of
Connect
DAZ Productions, Inc.
224 S 200 W, Suite #250
Salt Lake City, UT 84101
Licensing Agreement | Terms of Service | Privacy Policy | EULA
© 2024 Daz Productions Inc. All Rights Reserved.
Comments
I think the only differene is the D - in capital rather than lowercase in the new version.
So, if I understand correctly, SSS will be using MIS? That's a very nice approach.
Thanks =) Two more watermarks, then =)
And Happy New Year =D
I did what I described, but it didn't help much, only a little =(
They changed quite a lot about RT SSS between 3DL 11 (4.8) and 12 (4.9), and apparently some of these changes meant 3DL 12 now requires much higher SSS samples irregardless whichever shadeop samples the environment. Maybe this was the change that triggered this (DS has 12.0.27, the standalone is 12.0.19, and they both have the coloured noise at 128 samples, which 3DL 11 doesn't have):
12.0.8 - 2015-06-29
https://3delight.atlassian.net/wiki/display/3DSP/Changelog
On the other hand, those samples seem to add less render time than they used to in 11. From 128 to 512 is a bit over a minute on my laptop, so should be way faster on a bigger machine.
Check out the comparisons... this is just the standard shader. And of course the more diffuse you mix in, the less noticeable any SSS noise becomes. These are all 8 pixel samples.
Sooo technically I guess it all means that every "non-old-style" shadeop is tied up within MIS now, but the details of implementation may vary with builds/versions. And the user has to do some guesswork as for the best params when writing their own shaders =D
There are always surprises =) I was looking at the Maya metal source and I found a new undocumented function called thinfilm()... Very interesting. Hopefully I'll get some time soon to build a dedicated metal-only shader where this can be tested.
Still some noise on the shadow areas, almost like too high shading rate. But yes, if you use more diffuse and less SSS, it won't be as noticeable. How does it affect backscatter?
The 512 samples in 3DL12 one looks acceptable to me (on my computer screen, that is; I don't know what it is that mobile devices do to images too often). But I don't mind a bit of monochrome grain in the shadows - film gets it, too. What irks me is that coloured noise - like phone cameras or high ISO settings on cheap CCD cameras (or like when I take my glasses off in the evening LOL).
I'll post a 1024 sample example when I'm on the computer - it took 6+ minutes, but not much quality increase.
There is still room for further noise reduction - pixel samples of 10 and higher, more GI samples... and either way, using a map alone for all the lighting won't be the most common scenario I think - not that many good maps.
As for back scatter, it is there with the map alone =P (I have a render to be posted later). Lit areas are okay.
But! What I noticed playing with that Hyperfocal HDR. The HDR represents the model real-world sun intensity, correct? So with it and the proper SSS scale, we should get the perfect 'glow'... but it is not as bright as we expect.
Why?
The answer is - the ears of our figures are too thick! Check out this medical article: http://www.ncbi.nlm.nih.gov/pubmed/8867746
A healthy ear should be around 1.5 mm max if I did the math right. M4 ears are much thicker. So translucency maps do have a place after all... or correction morphs.
Lobes can be a bit thicker than that...but then again, the lobes are not the part of the ear we want to 'glow'...
I broke out my various measuring devices and started measuring my ear...my lobes are maybe 3mm at their thickest. Other parts are in the 1.5 mm range.
Creating a 5mm diameter cylinder and positioning it in G3M's ear showed that most of the ear is over 5mm thick...even what should be the thinnest parts were between 4mm and 5mm.
Happy new year y'all. I've been a tad busy of late.
If I was to venture a guess, you, me, and probably most people that visit the forum don't normally take micrometers to our ears, lol. Tho that sounds about correct for ear thicknesses. With the utmost respect to daz3d and all the PA's, I also don't think Genesis (1, 2, or 3) is completely 'human' at that level of dimensions (closer then some others I've seen years past).
If I was to guess, I think ear shapes and sizes are probably one of the more dynamic parts of the human body. Probably with more variations then there are fingernail polish styles, lol.
wowie and Kettu. That grainy shadow thing is interesting, and I agree with the color-noise, especially when the colors don't match the lighting and surfaces.
I normally wouldn't either...but I've got the flu and am bored...especially of staying in bed.
Happy New Year Zarcon!
I read somewhere that actual ear _shapes_ are like fingerprints - unique (you could tell a person from their ears even after extensive cosmetic surgery). But from the medical articles and Mjc's measurements it seems that the thickness is more or less the same around.
And yeah, noise doesn't match the colours of the rest of the image... this is why it's "noise" not something good =)
Get well soon!
What's the English name for this? http://ujutnee.net/sites/default/files/u65/2010/11/shtangencircul1.jpg
It's not a device well-suited for measuring ears, though. =D Thanks Mr Brin (and what's the other guy's name, Larry something) for Google.
M4's ears are also around 5 mm. That just about explains it all.
Here are the renders. Note M4's ears. "Close, but no cigar". Hopefully I'll manage to roll out a working shareable morph just for the heck of it - but for what figure first? G2?
Got some new ideas and valuable input plus feedback. Revamped Lumina skin preset.
Looks great, wowie. They both look good, but if you toggle between the 2 instantly, you can see the harsh (relatively) shadows on her cheek/nose disappear and the highlights pop more overall. The detail surrounding the eyes is also vastly improved. Excellent work!
- Greg
Thanks, just need to test the settings on other textures.
Beautiful =) Would be awesome to see the effect on a paler skin colour.
Already done testing with some of my favorite texture sets. Seems good so far. Done some tweaks to the SSS settings, mostly to get the ears backscatter right with really strong sunlight. Haven't tried any pale skin textures yet. The only pale skin set I have is probably V6 Anna.
Need to rework the hair presets too. I'm thinking of focusing the preset to use raytraced displacement rather than bump. Seems like that's the only way to get really good strand/fiber details.
Btw, cool breakdown though I think the movie was not very good.
That was cool - thanks for posting the link.
- Greg
I should check my library to see if there were any good freebie pale textures for Gen4.
Yeah, I notice it too that the best results with transmapped hair or painted fur (short fur, like most horses and some cats) are with RT displacement. Did I post my cat and horse tests here?
Interesting video, thanks =) I refuse to watch the movie, though... it's personal =)
Oh, forgot about these:
http://www.daz3d.com/mortella-for-v4
http://www.daz3d.com/valeria-for-v4
Pale enough for testing?
Oh, totally great! The vampire skin on Valeria, especially. My "pale pretties" all turned out to be inexpensive but commercial sets from other websites, and moreover, very few are still available to this day. I guess goth isn't fashionable anymore...
I come in peace. You do not know me, but I am not a troll. I am a person who doesn't use a hammer to fix everything. I use the correct tool for the job. Sometime one tool isn't better or worse than another, they are just different. On my dual-Nvidia PC I will probably use Iray. On my 8-core AMD CPU dual-AMD GPU PC I will probably use 3Delight.
I wish to learn how to use the 3Delight renderer to its full potential. But I am a novice. I have read a lot of theory, but I have yet to put it to practice.
I have seen the promo shots on 3Delight's website and they look good. I have downloaded the latest full version to my PC. I have used DS to render to a RIB and I have rendered the file on my PC using the standalone version of 3Delight.
I now wish to begin my studies under the kind tutelage of the participants of this thread.
What are the top three strengths of 3Delight? What are the bottom three weaknesses?
What are the three general things I can do in (almost) any render to improve the quality when using 3Delight?
I shall start with that and ask questions as I progress.
Thank you in advance for your help.
In respect of 3delight and iray in DAZ Studio?
I'll actually start with the weaknesses:
Caustics and GI would be the top of my grievances with 3delight and DAZ Studio. Third will be the lack of modern, properly physical plausible shader. Well, scratch that - lack of proper shader, real physical lights, sun, sky and basially pretty much everything you need to do proper photorealistic render.
Kettu's kit have helped somewhat, but I'd still wish we didn't have to jump through holes, with one hand and one leg tied together, running an obstacle course blindfolded.
The strength? Robust - you can do both a PBR render and an NPR (non-photorealistic render) in the same frame. It's up to you whether or not to follow the physical rules or break them. Performance - once you use all things like motion blur, DOF, displacement, especially for animation work where you most likely have to cheat a lot and don't mind not having the most physically accurate render. Third would likely be customization, though to be completely honest, that can be a double edge sword if you don't know what you're doing.
I've not looked into caustics much yet, but I keep seeing references to it and products in passing and I'm unclear on it. Does 3DL do caustics, or not, or only with a 3rd party product, or only faked with a 3rd party product, or does them but not in a desired way, or what's the deal?
It does them...but in Studio it is a very awkward process so it is easier to 'fake' them. It requires a light, camera and shader all designed specifically to work together and is rather slow. True caustics in Studio is probably the slowest thing you can render...one render that I did was over 24 hrs, with caustics and volumetric fog and about 20 mins without them (caustics...still kept the fog). In the wider 3DL world it's done in passes and even with scripted rendering in Studio getting access to those functions is difficult. 3DL for Maya does some wonderful caustics, quickly because it's much easier to access those functions.
Offhand, do you know any speciifc light/camera/shader combinations that will work?
The Shader Mixer example for caustics...but I could never get that one to work correctly with transmitted caustics, just reflective caustics. Others have had the opposite experience. The best is to build the needed components from scratch in Shader Mixer...
Test results with the Hyperfocal HDRI.
With the HDRI, these UE2 settings produces almost the same results.
100% Contrast - 95% Intensity at 192,192,192
0% Contrast - 1100% Intensity at 192,192,192
Distant light intensity needed to be raised to something like 1050% at 192,192,192 (560% at 255,255,255). So very close numbers (1050% and 1100%). That's quite a bit of range in the image. Not all HDRI have that range though.
I also did tests with UE2 without the HDRI and a single distant light to emit diffuse/specular, trying to (roughly) match the luminance.
Distant light has to be 900.0% and UE2 100% at 192,192,192. The total sum of light intensity (1000%) is almost the same as both the UE2 with HDRI or just using one single distant light. I'd say that's pretty good approximation of energy conservation. If you prefer using 255,255,255, distant light is 480.0 and the UE2 is 60%. Sums up to roughly half, which is about correct since in gamma space 50% of 255,255,255 is 187,187,187.
I probably going to include this in the readme (or the guide), regarding using HDRI and lighting in general. Of course, this is for an outdoor scene. Indoor scene will be different.
Oh, btw. Very nice freebies from Evermotion:
http://www.evermotion.org/shop/show_product/archmodels-vol-152/11779
There's textures, OBJ, FBX and some IES files included.
The primary weakness of using 3Delight _in_DAZ_Studio_ is that the shaders available for DAZ Studio, along with the non-scripted renderer integration interface, are sorely out of date.
Official 3DL plugins for Maya or Max come with all the integration scripts and pre-developed shaders that make use of these new features; but the DS integration is not maintained by 3Delight developers.
When you are limited to just what there is in DS already, the No 1 quality thing is to enable gamma correction in the render settings, and then learn to push the 'oldschool' DS shaders into making sense. Wowie's the resident expert on this subject.
To really use the 3DL+DS combo to its fullest potential, you need new shaders and custom 'scripted renderer' controls. I'm the geeky kid here who's building an ever-growing system of these in the vein of the official 'big name software' plugins, and if you aren't afraid of unfinished documentation, you are invited into helping test it (no restrictions on what you do with the renders).
If you don't want to depend on anyone else, you need to know your way around the following things:
1) the 3DL-specific extension of the Renderman Shading Language (RSL): to write shaders;
2) DAZ Studio "Shader Builder" tab: to help create the UI for your shaders;
3) what RiOptions and RiAttributes 3DL supports, plus a number of 'scripted rendering'-related functions from the DAZ Script to actually set those options/attributes from the DS interface.
This is a strength and a downside at once, as Wowie says - 3Delight can do just about anything, but you have to know what you want it to do and how to tell it to do it.
There is also the fact that the renderer itself is being constantly updated, while its documentation is not. So some info hunting is involved to find out what's the latest and greatest addition to the toolbox.
Outside that - when you know which tools from the toolbox to use, the first impression that professionals get from 3Delight, judging by various forums, is that it is __fast__. And that it handles displacement with more detail than many other renderers.
The biggest "weakness", in my opinion, is caustics - since the pathtracer module is unidirectional, raytraced caustics are rarely feasible (too much noise). So photon mapping is the official way to get caustics, and it requires certain complications to the pipeline.
GI, to me, is not a problem when it's done via the pathtracer.
Thanks for the comments re: my post. Keep them coming. Thanks.
It is obvious I need to learn more of the technical aspects of rendering. I think I will actually learn elsewhere. I have Unity and Shader Forge. I have not yet fired up Shader Forge yet, but believe it might be the best way to learn shaders by visually programming them. Then I can move to hand coding.
I also need to stick a cube on a plane and just turn stuff on and off and see how it reacts.
I sounds like 3Delight and DS are moving away from each other "officially", now that they have Iray. I'd fully support a community effort to keep it alive.
What makes you say that? Daz3D also renewed its license to include 3DL in DS and has been updating the included version.
Mustakettu85's comments make it sound like there were no updates and that DS was falling behind 3DL. I assumed he was correct. Do you have news from Daz that they are actively getting DS to work with the updates to 3DL?
Updating a license doesn't necessarily mean actively developing new tools and keeping up with the updates. But you know what, I'd settle for fixing already known and reported bugs first. I have submitted some bug reports that are now something like 8 to 9 months old, with no updates or even answers.