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
© 2024 Daz Productions Inc. All Rights Reserved.
Comments
awesome still improving the shader ! thanks a lot
Yes, this is using the default.
Specular strength for skin is 35% with 'Normalize Specular Map' enabled so the texture values remap to chosen IOR reflectivity.
SSS strength is kept at 100%. I did fiddle with the diffuse strength. As I noted on the tips and tricks, skin with SSS will be around 50%. Some variance will exist due to skin tone and texture quality.
There are still some issue with SSS and displacements when lit only with indirect light.
Technically, it could go down more, but I haven't found the best solution to integrate some extra optimizations without causing more noise. Render times with even higher samples should be a lot better than before.
Tks, and I agree with @khorneV2, awesome:)
Started working on a universe:) 15 min at 10x10 pixel samples...
Noticed that the environmental shader doesn't accept grayscale images in the diffuse color slot, has to be in rgb space. Grayscale works for diffuse strength.
Let me guess. It renders the texture in red? Easy to fix.
This might be of interest to you then - https://store.steampowered.com/app/372590/Grand_Designer/
Yup
Converted to RGB and it worked;)
This might be of interest to you then - https://store.steampowered.com/app/372590/Grand_Designer/
Tks for the link, hmm windows only, saved me 60 bucks:D Yea I love the Mac but really could use a windows rig too
They won't if they are attached to the camera. If they are attached to primitives like UberVolume does by default, they will work alright. Those that do raymarching, like UberVolume, will need traditional lights, though.
I stand corrected.
Hmm, is this a ploy to get me working on a volume shader?
Yup
An experiment... a perfectly centered sphere with the environmental shader applied, a metal cube also centered (spec. roughness at 0, IoR 1) and a camera placed at x=0, y=0 , z=150. Was about to test the number of reflective bounces possible with max rendersettings, well all except diffuse bounce depth, since there are no diffuse surfaces enabled. Pixelsamples at 24x24. Changed the spec BRDF for the cube to cook-torrance (don't recall what that is in reality, still on the old build). Wonder how different this would look with the newest build;) Anyway, turned out pretty cool:) There's a massive difference between the various BRDFs, the default one being quite grainy even with those render settings. Oh and no DoF;)
And, yeah I tinted the reflections a bit in an attempt to be artistic:D
That would be GGX.
There are no changes to GGX. Some changes to Ashikhmin Shirley so it now closely resembles GGX. As for the update, you can just copy/replace the files directly. All you need to do is locate the shader folder where DS keeps the .sdl files and supporting .dsa script.
Combined that old subway station with the latest test scene. Dialed down the IOR and spec/reflection on the dress to 1.25. Only rendered in 720p, 8x8 pixel samples and 2048 irradiance samples all around. Roughly 40 minutes.
Tks!
I was thinking more about the reflection roughness;)
Tks, hmm guess I'll have to do something about that sooner or later:)
40 min is pretty impressive I must say! Lot of emissive planes and all! Hairs are still without diffuse textures? I can see some (very little actually) noise on the skins, SSS samples maybe?
Edit: not even sure about the noise, looks good:)
Yes. There are no changes to GGX roughness ramp or other elements related to GGX. The only change was in build 1.1 - to fix the mismatched options between the UI and the internal one.
There might be some bit of noise if that particular spot only receive indirect light. The forearms has some. But you can still raise the samples selectively if you want. There are extra optimizations for samples above 2048. I was thinking about that for awhile, but eventually decided 8x8 pixel samples and 2048 irradiance samples covers most use cases. Plus, AWE Surface already have the upper limit set to 2048.
Edit:
The minimize subsurface noise, you can raise the subsurface samples and subsurface ray weight. I recommend raising the subsurface samples first, since subsurface ray weight have a bigger impact on performance compared to subsurface samples.
Actually, one emissive plane but instanced many times. I probably should use two though, since the square ones are round spotlights (on the original model).
@Wowie
I don’t want to interrupt like I did at 3DL lab topic. : 0 !
I’d be remiss though if didn’t thank you for amazing piece of reverse engineering. Shader is 3DL, yet also seems to be very much its own animal if that makes any sense.
@Sven
Not only amazing renders, but helps us rookies too!
Back to lurking. :)
Feel free to Interrupt me anytime.
I'll take that as a compliment. But yes, like Mustakettu would say, the underlying framework is done by the 3delight developers. It was always right there, just never fully utilized by DAZ Studio.
A rather strange thing happened when testrendering a character with an OOT hair I converted to awe. (Btw the 3DL mats are UberSurface presets and render nicely with scripted pathtracing I noticed.) Spent a lot of time tweaking the hair surfaces, rendertimes were very long, even spotrendering parts of the hair took a fair amount of time. Finally I found out that the only specular BRDF that would render properly was cook-torrance classic (first build), so after changing that, it started rendering normally. I also reduced diffuse bounces and specular bounce depth to 2 and 4, still looked rather ok. Enabled the hidden normals, spotrendered and that worked fine. Let it render and when I checked the final image there was pixelization all over the hair. Thinking it was the good old DoF bug, but no not this time. Closed DS and re opened the scene, spotrendered...black spots everywhere on the hair, enabled the "use face forward" and the spots went away but still pixelated. Turned off normals, turned off use face forward and everything was fine:) Yeah, so I know there are some issues with normals in the first build, but this was rather odd all the same. Don't know if reducing diff and spec bounce depth really made a difference, I set them back to the defaults and the hair still rendered fine...end of story:D
Well another thing about that scene... I used an overcast HDRI and wouldn't get much eye reflections at all, struggled a bit with that, gave up and loaded a second environment sphere with the same HDRI, made it reflect only, and maxed out exposure. Worked! So tks Kettu and wowie, you mentioned that possibility at some stage and now I made good use of it:)
Sven, once the new DIM build is out, try rendering some of the stuff you've had trouble with. I'd like to see if they behave differently or not.
Absolutely, that's the plan:) Any idea about release date?
@Sven -- I checked DIM last night and AWE 1.2 was in the 'updates' section. I installed it and it was 1.2 and the guide had features that Wowie was working on. So it's out in the wild. Or it seems to be to me.
Hey, tks for the heads up, great news:)
I admit the iray uber to AWE surface conversion is rather 'clunky', but I don't want to rely on a huge script that stores every transferable iray uber settings and re-apply them to their AWE equivalents.
Also, I didn't quite work out a scheme to transfer thin film for the coat. You'll have to do that manually.
Another thing that irks me is Genesis 8 sclera/iris. They maybe too dark even at diffuse strength already at 100%. Two things you can do:
Tks wowie, just downloaded...very excited:)
Installed the 1.2 build, opened a scene I made yesterday, hit render...stalled after rendering a little square, nothing happened for 5 min...had a moment of...umh... despair
Figured it had to do with the BRDF confusion, and yup, set everything to AshikhminShirley and it started working:) And I'm fully prepaired to having to rework all surfaces etc. but one thing confuses me, the white balance. Are there some changes to the environmental shader affecting that? Anyway, first impression is that it renders faster;)
Here is the render from yesterday with the old build, pixelsamples at 12x12:
And with the new build, only difference is changed BRDFs and rendered with 8x8 pixelsamples. Could the rendersettings affect whitebalance? There's a bit more noise, which is understandable with the lower samples. No postwork whatsoever.
Right, back to testing...
Blackbody temperature (both in the lights and surface shaders) was a bit off in the first release. The update(s) are more accurate. If you find everything looks too 'warm', either raise the white balance (if you only use a HDRI, it's probably best to just use 6500K) or add more temperature compensation to your lights.
It may have higher noise at low samples. Since higher irradiance samples are much faster than the first release, it's probably best to just straight up use higher samples anyway.
Ok it appears to be the BRDF change. Testrendering now with AshikhminShirley Classic and from what I can see it looks similar to yesterday's render.
Lol crossposting! Ok tks for clarifying!
Hmm is that why I seemed to have trouble getting proper gamma with HDRI lighting?
Temperature isn't used for ambient/HDRI in the environment sphere shader. I did change how ambient exposure is calculated. Exposure/gain/gamma now works on total ambient (color, texture and intensity).
Plus, the added specular/reflection lobe to both the environment sphere shader and AWE AreaPT.
...in the process of rebuilding my library and runtimes. When I have enough to work with, I'll do some preliminary tests.