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
Made a new attempt of getting aweHair to work properly, downloaded what I think are the latest versions of awe surface and awe hair, installed manually this time and made sure I didn't miss anything. These are the shader versions I have now:
Test scene with default HDRI and a fresh G1 with just a shape- and mat- preset. Loaded a hair and converted to aweHair, applied the black hair preset, removed the diffuse textures and rendered. The opacity maps are there and opacity is enabled. I also still have the glint strength in the shader UI. (and the error message). Still the wrong shader version?
Pretty much. Those links I've posted are for the most up to date AWE Surface and Hair shaders. It's newer than the one included with the AWE Shading Kit install.
If you look at the .sdl files, you'll see the one included with the kit is from Oct 28th and the newer one is from Dec 11th.
If you have the AWE Shading Kit, you can set reflection/refraction/diffuse and hair depth independently of each other via the AWE Environment Light. If you're talking abuot using AWE area lights, they will only work with AWE Surface or mustakettu's Radium shaders. Those light won't be seen by other DAZ studio shaders. On the other hand, AWE Surface and Radium shaders still will work with other lights (default point/spot/distant, AoA's Advanced lights, or Parrish IBL Master).
If you disable reflection, global illumination and support for path traced area lights, AWE Surface will pretty much behave just like dsDefaultMaterial or omnifreaker's HSS/EHSS/UberSurface and AoA's Subsurface shader (limited to direct lighting from point/spot/distant light). In this type of situtation, you can use UberEnvironment to add ambient light with occlusion, or use AWE Area PT lights with global illumination disabled or with diffuse ray depth 0 to get roughly the same effect.
If you want to have shadows with AWE AreaPT lights, what you can do is use AWE Surface on surfaces that 'receives' the shadows, but disable reflection and global illumination. This is done per surface.
The regular shaders (dsDefault/HSS/Uber/AoA's shaders) don't use raytraced reflection and global illumination by default, which is why they render very fast. Especially if you're using reflection maps/IBL light. If you want a similar output with AWE Surface, you can either disable reflection and global illumination per surface or use the AWE Environment controls to limit ray depth. With the new update, there's an additional control for separate metal/dielectric materials, so you can limit specular/reflection depth to 1 for non metals, but keep the ray depth high for metals and refraction. This is simply can not be done via the renderer's max ray depth option available in the renderer's settings tab/menu.
Of all raytracing rays, reflection is the one that takes the most time to render. Global illumination which are diffuse rays don't really take that long to render. But if you prefer an ambient occlusion look, you can either disable global illumination and rely on something like UberEnvironment. An alternative will be to limit diffuse ray bounces and play around with the new Occlusion controls in AWE Environment Light.
I've compared my shaders with 3delight's OSL/NSI based shaders and I think my shaders performs pretty well to ones used by the new 3delight. Especially considering DAZ Studio's included 3delight build is very old. By limiting dielectric reflection ray depth to 1, I can basically render a well lit scene at 1280x720 in 24 min with my aging 4 core/8 thread hardware (Intel Core i7 - 4770K). If you have newer/faster and more capable hardware, it'll undoubtedly be a lot faster.
The banner image used on my Patreon page renders in roughly 30 to 40 min with the default pathtracer draft render script, if I'm not mistaken.
AWE Environment light is not really 'light' in the sense of Uber Environment. It's simply a placeholder for specific settings AWE Surface/Hair/Environment Sphere shaders uses. All the render relevant operations is done in these surface shaders, not in light shaders.
@wowie
Ok so I have the latest awehair, see my previous post, shader versions attached, but it doesn't work, any ideas?
I think that's just a screw up with the AWE Hair Definition/Params.dsa files. The .sdl is the new one, but those Defintions/Parameters files might still be from the old one. Thanks for looking into it. By the way, you don't need to remove the diffuse/color texture. SImply use the 'Override Hair Color' slider all the way up to 1 to disregard diffuse color/texture completely. Plus, typical IOR of hair is 1.55 (based off literature). If you're seeing to strong spec, it's very likely you have too strong light. There's also the Specular Exposure slider if you want to over/under expose the R lobe. Naturally, the GI Exposure controls the TRT and TT lobe.
FYI the 1st specular lobe is the R lobe, which is the monochromatic hair reflection. The second lobe is the TRT, which is the colored hair reflection/absorption of the hair. Translucency is from the TT lobe. The lighter the hair, which is a combination of hair color strength and melanin/red melanin content, the less absorption and color. If you need to mix a light highlight on very dark hair, you'll need to use a map to the hair color strength slot or the specular 2 strength and translucency slot (either will work).
The less color/absorption you have will also mean the less translucency and actual transparency of the hair. As a result, darker hair will look quite solid solid compared to lighter hair.
Roughness determines how spread out the reflections will be, while root to tip shift will move the highlight closer to the root or tip. It's generally closer to the tip, but some hair have inverted UVs. Haven't found a way around that (except for fixing the UV or making a separate surface zone for affected areas). I did put in a switch with mask that you can use to switch between a vertical and horizontal ramp, since I've found several hair props (from older DS/Poser stuff) with those issues.
Since I generally not use any specular maps of hair props, I haven't put too much effort into dynamicaly adjusting bad levels like I did with AWE Surface. While you can still use them, I've found generally every hair looks great without those.
Tks for this info, sounds very promising!
Slightly out of topic, I did some testing today.
Here's a render with UberArea lights and UberEnvironment set to BounceGI. Both set to 512 samples and all surface shaders are using dsDefaultMaterial. Render time : 17 min 29.55 secs
A render of the same scene using AWE Surface and AWE AreaPT lights with reflection and GI disabled, set to 1024 samples. Render time : 6 min 52.46 secs,
With GI enabled, render time was 8 min 39.5 secs.
Used the Raytracer Final option so using 8x8 pixel samples. I didn't match the intensity and falloff of the UberArea lights though. But it should be a close enough comparison to those wanting to do comparisons with using old school surface and light shaders. 3delight's path tracer is very robust and fast, especially when used properly.
I second that:) Still...lately been having wet dreams about the new MacPro and what it could do to my rendertimes;)
It turned out I was not including the proper support scripts for AWE Hair. I've uploaded an updated installer with proper scripts and revised presets. It uses the old link, so please re-download the installer files. Once everything is verified to work properly, I'll upload the shaders to other sites.
Thanks.
So I uninstalled awehair with DIM, downloaded a new ZIP using your DL links on the previous page, installed it with DIM. Still ignores the opacity masks, so no cigar...
The awe hair UI looks identical to the preview version, can't see an "override hair color" slider, and there is still the "glint strength". And the hair presets seem to load with IoR 0.00
Correction: If I load a hair that has the dz default shader with IoR set to 0 (which is pretty much the default), then converts to awe hair, it leaves the IoR untouched, same goes for the awe hair presets.
Delete all the old shader files first and then do a re-install.Nevermind. I figure out what the other problem was. I didn't include the proper references in the .dsx files. Should be fixed now. Please re-download the installer and do a re-install.
wowie
Thanks for the detailed answer! I turned off reflection, global illumination and "Path traced area light" in the shader settings and ran the test again. Rendering is definitely faster, but still noticeably slower than a regular shader. 5 minutes and 20 seconds awe. 1:50 sec Dsdefauld)
I repeated the test in an empty scene, the result is 1:20 dsdefauld and 2:20 awe. This is a good result, but I'm afraid the rendering time will still increase greatly in a filled scene.
Quite recently, along with metal and glass, I began to apply the awe shader to the character's skin, which affected the rendering speed, but the result is worth it. I'm afraid my weak PC won't be able to handle if I start applying the Awe shader to walls and clothes and other rough surfaces. Although I would gladly switch to AWE completely.
I have another question.
I have updated the shader as it seems to me to the latest version 1.3. Although the program itself says that version 1.1. But it's not that.
I rendered with the new shader and noticed a difference with the old one. The surface of the awe has ceased to catch the shadow of the master's IBL, or the surface has simply become too bright, I still cannot understand. This applies to both dielectric and metal. What could have changed in the shader to give such an effect?
Awe1.0
Awe 1.3
Don't forget that AWE Surface is also faster at rendering stuff with opacity maps. I've seen instances where the dsDefaultMaterial or even other shaders grind to a halt when rendering surfaces with opacity maps. Especially when combined with raytracing.
I don't have IBL Master so I couldn't confirm your tests. But if global illumination is enabled, you should expect more light bounces.
Ok seems to be working now, atleast in DS 4.7, uninstalled/re-installed through DIM and got a bunch of new sliders to play with
. AWE Hair presets are working. Have yet to install it on the main rig with 4.9 and do some more serious testing. Looks great so far...
Hmm I'm getting fireflies in the hair on the laptop with DS 4.7 (3Delight 11). Got me a bit worried, but installed AWE Hair on my main computer with 4.9 and thank goodness no problems so far. Here a testrender where I used bump, displacement and normalmaps on the hair. My laptop with 4.7 is so slow so I didn't try to find the culprit, but thought it was worth mentioning.
Used AWE Hair on the garibaldi brows and that seems to work better now with this new build, even without the "AO distance". Or maybe that's a hidden property now? Didn't come to think of it:)
It should look best in 4.9 due to 3delight built into 4.7 have several bugs in relation to the hair raytype. Never did experience fireflies on 4.7 during testing though.
If it looks noisy, you probably need to raise the hair samples to 512. 256 is enough if you rely mostly on AWE Area PT, but if you use a HDRI, you'll need higher samples. Also, most noise will be in the colored (2nd) specular lobe and/or translucency.
Occlusion on AWE Hair and AWE Surface is now controllable only via AWE Environment Light.
Happy Holidays wowie and everybody:) Stay safe & enjoy!
@wowie
So I wanted to make this quick and stoopid scene, but after three days of trying to get a clean render it's not so fun anymore
. Lighting scenario: The environment sphere with a png and 14 instances of a 1 polygon plane with the PT arealight shader for ceiling lights. The light bulbs with the environmental shader with only camera visibility enabled. Here is the most recent render, non progressive 2048 Irradiance samples, 10x10 pixelsamples, rendertime 10h:
Problem: Again got this really ugly noise that looks like specular noise, and especially the metal surfaces were really hard to set up without being totally overblown. The diving thingy they are standing on has no maps at all, i'ts just a light grey diffuse color and a tiny bit of spec2 with about 10% roughness. So, after seeing this render I decided I must find the culprit and actually I did. It's the poly shape on the area light causing the noise. Turn that off and the noise is gone. Just wanted you to know, maybe you could look into it?
PS I'm pretty sure that was the problem also with this scene. Remember I told you the noise went away when I turned off the emitter outside the window? IIRC it had poly shape enabled, and the ceiling lights definitely had it on, so I will revisit that one at some point...
Hmm would it be better to use an opacity mask instead to keep the highlights round?
The fact that it needed that much time to render is a big hint what the problem is. Your surfaces aren't finding the light emitters at first/second bounce. Hence the noise.
What you should is setup your lights with out GI/reflection first to setup the light's intensity/EV. Then enable GI/reflection to see what your final render will look like. The easiest way to switch between preview and final is by adjusting diffuse/specular ray depth via AWE Environment Light. Then add the environment sphere. Relying too heavily on the environment sphere on indoor scenarios will produce lots of specular/diffuse noise.\
Here's a step by step of what I recommend. I've setup planes with Poly shape enabled on the lights and apply AWE Area PT on the ceiling windows.
Initial setup with default values of shaders. Direct light only (both diffuse and reflection depth set to 0).
Raise the intensity scale until you have enough light around the scene. In this example, it's 11 EV.
A preview with GI enabled.
Turn off the ceiliring lights and fine tuned the ceiling lights. With it's bigger area, 5 EV is enough.
Here's what I have with both emitters enabled. Raytracer final with 1024 samples. 11 minutes 9.98 seconds
Once that's done, fine tune your camera exposure and tone mapping values. Plus, fine tune roughness to all relevant materials (i've used 12% for most of the wood panelling). If you want to keep render times down but retain most of the quality, set diffuse depth to 2 and (dielectric) reflection depth to 1.
Reflection enabled. 22 minutes 21.38 seconds
I've enabled the environment sphere and wall closest to the camera but kept visible to camera to 0 (disabled). Then set wall's opacity to 0 so I can see into the room for the viewport. Since opacity is disabled, the wall will still be seen by other surfaces in the final render.
How is that possible when everything except the ceiling gets direct light from the ceiling emitters? The diving board is, as you can see from the characters' shadows, directly under 4 emitters. Environment on or off doesn't matter, still noisy as hell, unless I TURN OFF POLYSHAPE. Why is that?
I appreciate the step by step turorial, some good tips for sure. Have yet to try making the windows emissive.
A simulated overcast light setup, using a big, low EV emitter and a small high EV emitter with a falloff of 1.
25 minutes 21.14 seconds and 22 minutes 29.10 seconds. Anything using SSS will raise render times. The jacuzzi's edges still need a smoothing angle adjustment.
If you have lots of emitters with too low intensity, then you'll end up not having enough light around your scene. Especially important if you have a physical falloff. Plus the emitters you are using are very small.
Look at my renders. The specular noise will be on areas where there's very little light, especially surfaces with glossy (>0.05 roughness). Which is specular noise, so raising your irradiance samples won't change a thing.
You simply need to add additional emitter(s) with low enough EV near those areas. Mind you, not just low EV, but large ones. Look at the sizes I'm using. Having such large sizes helps rays find them.
Raising the scene/camera exposure won't help, pretty much like raising a camera ISO (increase light sensitivity at the expense of noise). Raising pixel samples might help, but I only recommend doing so when you're doing renders with DOF and/or motion blur.
As for polyshape, you do know that using it effectively reduces the size of the emitter, right? For instance, if you have a square inside a square, the inside square will be half the area of the bigger square. Here's an illustration:
While you can compensate the area reduction by increasing the intensity, there's a practical limit to that.
Ok tks a lot for this lesson;) I thought I had made the emitters large enough, with poly shape on I used 13.5 EV so obviously a good bit more than your example, and yes I used physical falloff, so after all they must be too small as you say. With poly shape off I adjusted light intensity by setting intensity scale offset to around -4 (from memory) for roughly the same overall light intensity, meaning the noise is gone with larger emitters. It's slowly starting to make sense
. Will have to rethink the whole thing
Intensity scale offset will basically added/subtracted to your selected Intensity scale. If you set EV to 13.5 and the EV offset to -4, you're going to get a final EV of 9.5.
Testing progressive and non progressive. The non progressive has some slight tweaks to camera exposure, materials and placement.
Non progressive.
Progressive.
37 min 49.47 secs vs 42 min 15.54 secs. So not that much difference it seems.
Yes! What I meant was turning off poly shape will increase intensity since the emitter will get bigger. So I used the offset to make up for the difference.
I've noticed that non progressive is even faster now, in many cases.
Here's a version I had started rendering prior to our discussion about poly shape. Render settings are the same (2048 samples, non progressive 10x10 ps), the size of the emitters are the same, only poly shape is OFF and I instead use an opacity mask, adjusted intensity to match the original. But I increased the environment exposure and the diffuse exposure offset, so there's more light coming in throug the windows.
8 hours 58 minutes 9.39 seconds, so slightly faster. Much of the noise is gone. However now I get occlusion(?) on the ceiling where the emitters are, don't know why, but it looks odd. Testing continues...
Hmm, comparing the two versions you can see from the reflections in the windows that the new version has larger relative size emitters, probably hence the lesser noise?
What you should do is (in steps).
If you do this, you should end up with less noise and faster renders. The time it will take will vary depending on the machine, shader settings (SSS, opacity). On a personal note, I've hardly seen any scene that renders more than 1 hour, with proper preparation (at least limit diffuse to ray depth 3 and (dielectric) reflection to ray depth 1). It could be just your insistence on using 10x10 pixel samples and 2048 irradiance samples.
The awe environment light settings have been at default all the time, except for temperature that I set to neutral 6500K and saturation that I changed to 0. In the environment section I raised the dome exposure to 2 + diffuse offset slightly raised, 0.5 IIRC. In override section I just set Irradiance samples to 2048. And I always render through a camera with DoF ON.
Anyway, I zeroed everything and started over, following your tips. Before doing anything else I scaled up the ceiling emitters 8 times, so now they are much larger than the sixe of the actual light bulbs. I ditched the opacity mask and use poly shape again. Reset rendersettings to 8x8 ps and 1024 Irradiance samples. Turned off diffuse and reflection depth with dielectic and metal override ON. The progressive testrender I never finished would have taken about an hour, I let it go to about 80%. Then I made the windows emissive, but it turned out the polygon count was around 1500, so ditched that idea. Created a cube, chopped off the bottom poly and inverted normals, wrapped it around the house and applied the PT area light. Ended up using physical falloff with 4 EV on that, along with the ceiling emitters at 12.5 EV. The dome with reflection/refraction and camera visibility. Have a test render going and it is faster than before by the looks of it. Now the only problem is that the emitters cast shadows or occlusion or something on the ceiling, which is weird since they are not visible to the camera and shadows are not enabled. Thinking of making the emitters doublesided and mask out the shadows that way, unless you have another solution...and of course, the reflections of them in the windows are now huge compared to the bulbs
which looks very off. Trace groups, anybody? 
*facepalm*
Screenshot of the render in progress:
Edit: I set poly blur to 0 and managed to get rid of the shadows, go figure...
Edit 2: I'm now suspecting I've set up the poly shape all wrong, using a too small radius combined with the poly blur. I've resized the emitters once more to better fit the light bulbs, we'll see how it turns out..
By that, you mean like this?
As I noted, don't use a cube or the environment sphere. You want to isolate your lighting to just the emitters. Do a direct light render by setting both diffuse and reflection ray depth to 0. Adjust your lights till you have enough illumination inside the room.\
The emissive part of the emitter should be inside the light's housing.
What you should get should be similar to this:
Direct light from both the ceiling light and portal/windows. Here's what I have when i disable the portal/window light and just use the ceiling light plus hero light.
Ps. You do know the second emitter have poseable parts/elements? If you scale it up, you can easily make a tent like shape over your scene.