3Delight Laboratory Thread: tips, questions, experiments

Mustakettu85

dracorn said:

Mustakettu85 said:

dracorn said:

I was playing around with Scripted 3Delight and found some interesting results. 

The first is standard G8 3DL materials with AoA Distant and Ambient light. 

The second is Scripted 3DL with Point Based Occlusion, default settings and identical lighting.  Does anybody know anything about Scripted 3Delight?

 

*raises hand*

Anything you could possibly want to know. The question is: what exactly do you want to find out? How to use this particular script to the best extent? How to write your own? Ask away.

How to use it.  I've never played with it before and am curious.

Thanks!

Alright here we go! (c)

There are a number of example scripts in there, but the most interesting is the "Point-based Occlusion" (PBO for short) one.

Why?

The standard example literally does nothing out of the ordinary because it's just an example; the outline example is sorta broken (get LineRender9K in the store to have a fixed and majorly updated version), while the "PBO" one is fairly well thought-out and integrated.

So the first step in my world is to always check if "gamma correction" is "on" in the "vanilla" render settings. This control carries over to "scripted rendering" thankfully, same as setting up motion blur. But only the "on/off" section carries over; the exact "gamma" value will need to be matched.

Then we select "Scripted rendering" and the "PBO" script. Let's take a look.

The first "subfolder" is "Renderer options". This is fairly standard fare, so set it up the way you prefer. Let me remind you to check the gamma slider here and match it to your "vanilla" setting.

Here's what my standard setup looks like regardless of specific mode:

If you want, I can explain each setting.

The next "subfolder" is "Simple occlusion light".

Now, see that "on" button there? Disable it.

This "simple occlusion light" is like a glorified headlamp. It will only mess with your scene lighting.

What you need is UberEnvironment - it is able to use the point clouds that this script generates.

But before you add it, it's best to save out a "render settings preset" for the next time you want to experiment with the script. Just do it as usual.

Before we proceed, I'll try to explain why this script exists at all. See, there are multiple ways of rendering things, especially more "complex" effects like AO and other forms of GI. A neat "hack" is to represent the scene as a "cloud" of points storing specific data that could later be interpolated to get certain effects calculated quicker than if we were to raytrace them. 3Delight can do it, but it needs to be explicitly told to do it. This is what the script achieves: it first tells the renderer to create our "cloud" and then it tells the renderer where to find it.

UberEnvironment is a shader that understands this "where to find" the cloud.

So, when you add an UberEnvironment, you have the options of just letting it do "point-based" AO instead of "raytraced" AO, which is not exactly interesting. You can try; the results will look a little different, and depending on what and how you render, point-based AO may or may not be faster.

But what _will_ be faster is point-based indirect light as compared to raytraced indirect light (when we're talking about UberEnvironment in particular, since it's so old its advanced raytraced effects are very slow today).

You can set up your UE like this:

The intensities on top that didn't fit in the screenshot are the default. The important bits are:

• The mode is IDL with soft shadows. I'm not using an HDR map to light the scene, so using "IDL with directional shadows" makes no sense.
• IDL colour = black because I am not using a map here to light the scene, it's enclosed in a sphere that has ambient active (with GI, ambient-based materials will emit light).
• Occlusion samples: it's a quality control. Technically we shouldn't be raytracing anything, but this value may be used for interpolation. Hard to say without source code. Either way, this is what to increase if you get blotches.
• Maximum trace distance: set to the distance you want one object to bounce light on other objects. My scene is inside a 10 m diameter sphere, so I used 700 cm = 7 meters to make sure the walls bounce onto the center and a bit beyond.

Here is what you get, left to right: - no UE, vanilla render; - no UE, the PBO script (these should be identical with the same settings); - UE set up for IDL, the PBO script. Render time for this last one is about 7.5 minutes on my office PC (Intel Core2 quad Q9300). As you can see, you don't have to fiddle with actual point cloud settings at all.

The figure is using mild SSS for the skin.

wowie

Mustakettu85 said:

Another way would be to head over to Wowie's store here at DAZ3D and buy some of his recent work which is more advanced in terms of light presets and has a somewhat different approach to setting up skin materials.

I recommend NOT buying it. Reason is simple - the new shader is leaps and bounds easier to use, renders faster and much closely follow physically based rendering/physically plausible shading. Packed with the usual material presets (not just skin) and a light set. I just need to finish the thing. Right now it is still just 373 lines of code. I want to add switchable BRDFs, just in case Ashikmin-Shirley works with anisotropy in DS 4.7.

Mustakettu85 said:

It is a DS bug.

It's literally a single extraneous AttributeEnd. One wrong API call. I even sent example RIBs in.

So, that's at least two unsolved tickets on the same subject. As with the 'other' bug reports we've filed. Strangely enough, when I tested 4.9.10, 3delight final rendering fails everytime (but not IPR).

Reworked the transmission component - it is much simpler now and also easier to use. Finally added proper specular color handling depending on material type (dieletric or metal). The 2nd spec is now separately controlled, though it is not linked to raytraced reflection. I also did some profiling on various character texture sets to see how the shader works with various DAZ assets. Basically, trying to find values that works and compare them to references. Much, much easier with a shader that has proper energy conservation.

These values work and falls inline with the shader parameters. Metal is different since I'm using reflectance/edge tint workflow.

Found on the XSI section of 3delight forums. Basically you want to set IOR between 1.4 to 1.6. 1.6 for 'oily' or 'glossy' surfaces and 1.4 for rougher, almost matte surfaces. If you find yourself using 2, the specular map is too dim.Disney used specular/reflection strength of 62.5% (without maps) to get an IOR of 1.5.

Some of the old Elite textures actually have very good luminance for the specular maps. I used 15% for Amy and 30% for Lana on both spec. Total spec is 30% for Amy and 60% for Lana, coat isn't used.

I set the diffuse roughness to around 25%, while spec 1 is 30% and spec 2 is 50%. Everything except the hair, iris, sclera is with the new shader. Test render one distant light and UE, bounce GI at 12x12 pixel samples - 8 minutes 23.68 seconds. Specular ray depth at 12 and diffuse ray depth at 3, courtesy of Mustakettu's script. For IPR tests, I usually just use 4, which is just high enough for thin glass objects and get most of the diffuse bounces and still renders quick enough for testing purposes.

Mustakettu85

wowie said:

Mustakettu85 said:

Another way would be to head over to Wowie's store here at DAZ3D and buy some of his recent work which is more advanced in terms of light presets and has a somewhat different approach to setting up skin materials.

I recommend NOT buying it. Reason is simple - the new shader is leaps and bounds easier to use, renders faster and much closely follow physically based rendering/physically plausible shading.

Makes sense, but I'm not sure the poster wants physically based and photoreal. If they're in a hurry and don't like using freebies, why not.

And that reflectance in the chart, it's the heads-on reflectance, right?

wowie

Mustakettu85 said:

And that reflectance in the chart, it's the heads-on reflectance, right?

Yes. Reflectance at 0 angle - head on. For the technically adept - F0. Values for grazing angles (F90) usually are 1 or close to it.

Oh yeah, I found this bit from Anders Langlands.

https://groups.google.com/forum/#!topic/alshaders/IZTbaqJMQBo

Enter 0.65 as the wavelength, then scroll down to the Reflection calculator (in Refractiveindex.info). The value 0.64915 is the red channel of the Reflectivity parameter. Then do the same thing at 0.55 and 0.44 to get the green and blue values. This gives me Reflectivity = (0.64915 , 0.6103, 0.53092). To get edge tint, change the angle of incidence in the reflection calculator to 80 from 0, and do the same thing again (check that the 80 angle doesn't get reset to 0 when you change the wavelength), using the non-polarized value, R. This gives me Edge tint = (0.64184, 0.64359, 0.6353).

 

The values are different now, but the workflow is good. My metal presets used the same first values, but slighlty different second values.

Mustakettu85

wowie said:

Mustakettu85 said:

And that reflectance in the chart, it's the heads-on reflectance, right?

Yes. Reflectance at 0 angle - head on. For the technically adept - F0. Values for grazing angles (F90) usually are 1 or close to it.

Oh yeah, I found this bit from Anders Langlands.

https://groups.google.com/forum/#!topic/alshaders/IZTbaqJMQBo

Enter 0.65 as the wavelength, then scroll down to the Reflection calculator (in Refractiveindex.info). The value 0.64915 is the red channel of the Reflectivity parameter. Then do the same thing at 0.55 and 0.44 to get the green and blue values. This gives me Reflectivity = (0.64915 , 0.6103, 0.53092). To get edge tint, change the angle of incidence in the reflection calculator to 80 from 0, and do the same thing again (check that the 80 angle doesn't get reset to 0 when you change the wavelength), using the non-polarized value, R. This gives me Edge tint = (0.64184, 0.64359, 0.6353).

 

The values are different now, but the workflow is good. My metal presets used the same first values, but slighlty different second values.

Thanks!
Oh yeah, that's a cool bit. Probably gives the closest match we can get with just RGB.

dracorn

Linwelly said:

has anybody tried the AOA lights in the latest DS version ( official, not the public built)? I believe to remember they had some more dial options in the light settings than light colour and light intensity, shadow is not even active... I tried the advanced distant light, the advanced spot light and the Ambient light.

For the record, I am in 3delight for this and I selected the lights from the light presets in the light tab and went to the editor to adjust them.

Could be I am just sitting on my brain right now but can someone check on that, please?

I just did this test render with V8 and AoA's Distant, Ambient and 2 Spot Lights.  I don't see any missing dials.  Shadow is not active in Ambient, but is active in Distant and Spotlight.  Lighting Control is still there. 

The red marble vase is lit by Ambient light and the second spotlight only.  It's Diffuse is set to 99%.  Ambient lighting control = none.  Second spotlight is bright green and is set to illuminate only the 99% diffuse.  The Distant and other Spotlight are set not to illuminate the 99% diffuse.  V8 is lit by all lights except the second bright green spotlight.

Looks to me that AoA's lights still work.

dracorn

Mustakettu85 said:

dracorn said:

Mustakettu85 said:

dracorn said:

I was playing around with Scripted 3Delight and found some interesting results. 

The first is standard G8 3DL materials with AoA Distant and Ambient light. 

The second is Scripted 3DL with Point Based Occlusion, default settings and identical lighting.  Does anybody know anything about Scripted 3Delight?

 

*raises hand*

Anything you could possibly want to know. The question is: what exactly do you want to find out? How to use this particular script to the best extent? How to write your own? Ask away.

How to use it.  I've never played with it before and am curious.

Thanks!

Alright here we go! (c)

There are a number of example scripts in there, but the most interesting is the "Point-based Occlusion" (PBO for short) one.

Why?

The standard example literally does nothing out of the ordinary because it's just an example; the outline example is sorta broken (get LineRender9K in the store to have a fixed and majorly updated version), while the "PBO" one is fairly well thought-out and integrated.

So the first step in my world is to always check if "gamma correction" is "on" in the "vanilla" render settings. This control carries over to "scripted rendering" thankfully, same as setting up motion blur. But only the "on/off" section carries over; the exact "gamma" value will need to be matched.

Then we select "Scripted rendering" and the "PBO" script. Let's take a look.

The first "subfolder" is "Renderer options". This is fairly standard fare, so set it up the way you prefer. Let me remind you to check the gamma slider here and match it to your "vanilla" setting.

Here's what my standard setup looks like regardless of specific mode:

If you want, I can explain each setting.

The next "subfolder" is "Simple occlusion light".

Now, see that "on" button there? Disable it.

This "simple occlusion light" is like a glorified headlamp. It will only mess with your scene lighting.

What you need is UberEnvironment - it is able to use the point clouds that this script generates.

But before you add it, it's best to save out a "render settings preset" for the next time you want to experiment with the script. Just do it as usual.

Before we proceed, I'll try to explain why this script exists at all. See, there are multiple ways of rendering things, especially more "complex" effects like AO and other forms of GI. A neat "hack" is to represent the scene as a "cloud" of points storing specific data that could later be interpolated to get certain effects calculated quicker than if we were to raytrace them. 3Delight can do it, but it needs to be explicitly told to do it. This is what the script achieves: it first tells the renderer to create our "cloud" and then it tells the renderer where to find it.

UberEnvironment is a shader that understands this "where to find" the cloud.

So, when you add an UberEnvironment, you have the options of just letting it do "point-based" AO instead of "raytraced" AO, which is not exactly interesting. You can try; the results will look a little different, and depending on what and how you render, point-based AO may or may not be faster.

But what _will_ be faster is point-based indirect light as compared to raytraced indirect light (when we're talking about UberEnvironment in particular, since it's so old its advanced raytraced effects are very slow today).

You can set up your UE like this:

The intensities on top that didn't fit in the screenshot are the default. The important bits are:

• The mode is IDL with soft shadows. I'm not using an HDR map to light the scene, so using "IDL with directional shadows" makes no sense.
• IDL colour = black because I am not using a map here to light the scene, it's enclosed in a sphere that has ambient active (with GI, ambient-based materials will emit light).
• Occlusion samples: it's a quality control. Technically we shouldn't be raytracing anything, but this value may be used for interpolation. Hard to say without source code. Either way, this is what to increase if you get blotches.
• Maximum trace distance: set to the distance you want one object to bounce light on other objects. My scene is inside a 10 m diameter sphere, so I used 700 cm = 7 meters to make sure the walls bounce onto the center and a bit beyond.

Here is what you get, left to right: - no UE, vanilla render; - no UE, the PBO script (these should be identical with the same settings); - UE set up for IDL, the PBO script. Render time for this last one is about 7.5 minutes on my office PC (Intel Core2 quad Q9300). As you can see, you don't have to fiddle with actual point cloud settings at all.

The figure is using mild SSS for the skin.

Wow, thanks Mustakettu85 ​for the detailed explanation!  I'll have to play around with this now.

Mustakettu85

dracorn said:

Wow, thanks Mustakettu85 ​for the detailed explanation!  I'll have to play around with this now.

You're welcome! Would be cool to see what you come up with.

Mustakettu85

K folks, updated the trick list some more.

https://www.daz3d.com/forums/discussion/comment/1827726/#Comment_1827726

It's also linked to from the first page of this thread, in huuge letters =)

Stopped on page 47 as of yet.

RAMWolff

Thanks so much.  Bookmarked

wowie

OK. Here's another one for those with US2.

This is how (I think) US2's SSS colors (scatter and absorption) are calculated:

Scatter Strength * (255 - Scatter Color)/255

Absorption Strength * (255 - Absorption Color)/255

Example:

The actual value you want to pass is for apple. That would be:

SSS_scatter = color(2.29, 2.39, 1.97)
SSS_absorption = color(0.003, 0.0034, 0.046)

From 'A Practical Model for Subsurface Light Transport', the original Jensen, Marschner, Levoy and Hanrahan paper.

Omnifreaker's US2 apple preset uses these values

Subsurface Scatter Color (11, 0 , 45) with Scatter strength 2.39

Let's tackle the green first

2.39 * (255 - 0)/255 = 2.39 * 1 = 2.39

So that works. How about the red and blue?

Red: 2.39 * (255-11)/255 = 2.39 * 244/255 = 2.39 * 0.9568 = 2.286 which is pretty close to 2.9. Might well be the same if i used the longer result

Blue: 2.39 * (255-45)/255 = 2.39 * 211/255 = 2.39 * 0.82745 = 1.97. Pretty much the same.

The same formula is also used for the absorption color. The difference is he used 0.050 for the absorption strength rather than 0.046.

Mustakettu85

Yes that matches the formula from the US2 rendertime script (thankfully not encoded).

wowie

And, some new tricks built into my shader.

Oren Nayar diffuse translucency with proper bleedthrough shadows. No user intervention required.

 

Mustakettu85

wowie said:

And, some new tricks built into my shader.

Oren Nayar diffuse translucency with proper bleedthrough shadows. No user intervention required.

Looks cool! Is there a way to turn it off for "heavier paper"?

wowie

'Heavier' paper means thicker material, less translucency, right? So just scale back the translucency. The shadows will honor any changes to translucency (strength, color). Since translucency itself only shows up with thin objects (infinitely thin polys), they don't show up on models with volume.

algovincian

wowie said:

And, some new tricks built into my shader.

Oren Nayar diffuse translucency with proper bleedthrough shadows. No user intervention required.

 

Great stuff, wowie! Simple renders, but there's a lot going on there, and it looks fantasic.

- Greg

AndyS

Hi wowie,

please, what is US2? Sounds like a new Uber Shader. I only know UE2, the environment light sphere of onmifreaker.
And I think it is time to share your fantastic shader. OK - you're not to do. But with the existing, in DAZ included surface shaders I'm not able to reproduce what you show; not able to follow what you do.

Thank you.

wowie

algovincian said:

Great stuff, wowie! Simple renders, but there's a lot going on there, and it looks fantasic.

- Greg

Thanks.

AndyS said:

Hi wowie,

please, what is US2? Sounds like a new Uber Shader. I only know UE2, the environment light sphere of onmifreaker.
And I think it is time to share your fantastic shader. OK - you're not to do. But with the existing, in DAZ included surface shaders I'm not able to reproduce what you show; not able to follow what you do.

Thank you.

US2 is UberSurface 2, a paid, more advanced version of UberSurface, also made by omnifreaker.

DAZ standard shaders is woefully out of date so no wonder people's renders have been stuck in the 90's. You can do something similar with light linking and geo shells but yeah, it's not fun. Which is why I want to put that feature in.

As for eta on the shader being publicly available, if all is well, probably in about a month.

Since kettu's shaders are still not out yet, I'll probably going to distribute the base shader via DAZ, for free. However, that's only going to be the shader and a short, barebone manual. Hopefully, this way a lot of people will pick it up and content makers (commercial or freebies) will use the shader as default materials. I only ask those who get them to file a feature request to DAZ on enabling, or at least, exposing raycache options into the 3delight renderer settings and update 3delight to a newer build. As i pointed out some pages back, DAZ is two years behind the latest release.

For those who want to support further development on this (and other stuff I'm not ready to say yet), there will be a commercial bundle of the shader, material presets and lightset using UE2 and shader light based dzSpotlights with a more thorough, detailed manual and scene examples.

There are some extra stuff I haven't been able to include on this version. Namely - thin film interference (iridescence), GTR brdf support with tail control and separate single subsurface scattering/multi scattering. Those will likely come at a later date, since they need a newer version of 3delight than the one in DS 4.7.

I'd say progress wise, it's about 70% done. Anisotropy for Ashikhmin Shirley BRDF is in, and you can control anisotropy direction with a map. Practically the same way as you would do it with 3delight material for Maya/Max. Those with DAZ Studio 4.8 will likely be able to use anisotropy with GGX (no support for anisotropy with Cook Torrance). SSS is already working though I still need to find a good way to control backscatter. After that, it's just bump/normal/displacement support and housekeeping stuff.

The optimizations are, well, extensive. One example : you can now use fully raytraced reflection on materials with opacity maps, even with with a max trace depth value of 12. Yes - raytraced hair is not only possible, but renders in about the same time as using US and US2 with occlusion override (with no raytracing).

Second example, with translucence enabled, you can now have thin fabric/material with some light passing through. The insides of clothing won't be completely dark, unless you want them to be.

One final note on the shader - it's made with physically plausible settings and global illumination in mind. So, when you apply the shader to a material with dsDefault material or other shaders, the values transferred over will likely be off. More so if you're just using ambient occlusion.

Here's a shot of the user interface (minus SSS since this is an older build).

There's a 'Transmission Roughness' to control refraction blurriness on the bottom, but isn't shown on the shot.

AndyS

Thank you so far.

One remark to your latest example:
I tried to reproduce this setup with several available materials in real world physics in my household.  --> This is unreal! You never get the clear shadow transition. Only a strong scattered light / shadow distribution.
OK - Perhaps if your material has got an opacity significantly below 1 (Opacity mixed with translucency). So a texture with a dense array of many mini-holes in the surface. But even the fabric of an umbrella doesn't show this outcome.

For fantasy pupose this looks nice - however.

omnifreaker

AndyS said:

please, what is US2? Sounds like a new Uber Shader. 

Not so new, but a layered Uber surface shader: https://www.daz3d.com/ubersurface2-layered-shader-for-daz-studio

wowie--looking forward to seeing your new shader!

RAMWolff

Will the commercial bundle be a merchant resourse at all?  I am looking forward to this release too.  I still see so much potential for 3DL but with iRAY here it's really tempting to move away from 3DL but upon making characters for HW3D and wanting to include 3DL settings that I like I have to say I like some of my 3DL pre promo renders better or as much as my iRAY pre promo renders.  My only niggle with 3DL is the really slow rendering time for anything with transparency.  I'm not a patient person when it comes to renders so having tricks to speed things up is good but if this shader has those tricks built in I'll be over the moon! 

AndyS

OK,

omnifreaker said:

AndyS said:

please, what is US2? Sounds like a new Uber Shader. 

Not so new, but a layered Uber surface shader: https://www.daz3d.com/ubersurface2-layered-shader-for-daz-studio

looking at the SKU number it's pretty old. So somehow I missed it.
Does it also dramatically slow down the render as your Über Volume shader does?

Mustakettu85

wowie said:

'Heavier' paper means thicker material, less translucency, right? So just scale back the translucency. The shadows will honor any changes to translucency (strength, color). Since translucency itself only shows up with thin objects (infinitely thin polys), they don't show up on models with volume.

Okay, I've run into a vocabulary deficiency here =D I'll try to explain...

Your render is great for - I think it's called "tracing paper" in English? The one that is almost "transparent", like cellophane/thin plastic. Those materials will actually transmit a significant amount of light and colour it. They're "internally smooth".

And if you take even a newspaper - supposing newspapers elsewhere are printed on the same type of paper as in Russia, of course - it's paper that is thin almost like "kleenex" tissue. It will be translucent even when there is a few layers, and you will see the shadow of the object on the other side of the paper, but the cast shadow will still be solid. So it's more like "internally fuzzy paper", "scatter-y paper" that I meant.

So there are times when you want the translucency on the material itself but you don't want (extra computation in the) more complex shadows. I'd throw in a switch =)

Mustakettu85

wowie said:

Since kettu's shaders are still not out yet...

And they won't be. Not the traditional way people understand "out" in this community, at least. I just don't have the time to deal with extensive technical support, and I sure as hell don't want my name smeared in mud behind my back by those who were unable to figure something out, as I've seen happen to other people, freebie providers and commercial vendors alike.

I see that out of my alpha team, when we only consider those "pure artists" who do not write their own shaders, there is only one person as of now who is actually investing their time into working with my kit. Others figured that it's not worth their time, given how much stuff out there is for Iray and/or for "oldschool" 3Delight (yes even when those presets do not make sense - at least they use familiar parameters etc). There's a lot of "inertia of thought" going around.

It's not surprising, actually, given that my team are hobbyists for whom 3D is not a "scientific exploration" but "a vehicle for their imagination" for which they want shortcuts. So I have no grudge against them.

Szark

that is sad that others have made you feel that way...(alternative to swearing) I am rather angry about this type of thing of disrespectful behaviour to someone that just wants to help and share information.  

RAMWolff

Agreeing with Szark!

omnifreaker

AndyS said:

Does it also dramatically slow down the render as your Über Volume shader does?

No--render times more in line with UberSurface that comes with DS.

Mustakettu85

Szark said:

that is sad that others have made you feel that way...(alternative to swearing) I am rather angry about this type of thing of disrespectful behaviour to someone that just wants to help and share information.  

 

RAMWolff said:

Agreeing with Szark!

Thank you guys. It hasn't yet happened to me personally in the 3D world, but I've had my share of it IRL.

Have you seen my friend's new release here BTW?

https://www.daz3d.com/futuristic-glass-elevator-and-building-kit

I helped her with the 3Delight stuff - UberSurface-based materials, a light preset, a render preset to get the best speed out of vanilla DS. So if there is any constructive criticism regarding that part of the product, it should go straight to me! =)

wowie

omnifreaker said:

wowie--looking forward to seeing your new shader!

Ah, thanks omni. Nice to see someone else than the usual people here in the thread.

RAMWolff said:

Will the commercial bundle be a merchant resourse at all? 

Hmm, I haven't really thought about that. What I have in mind are roughness presets, metal presets (basically, reflection and color edge for various types of metal) and index of refraction preset for dieletrics (water, glass, plastic). I'd say if you want to make your own material preset, like maybe rough plastic, you can basically apply my preset, then save the IOR and roughness values (for base roughness or diffuse/specular/coat) into a shader preset for your users/customers.  I'm OK with that.

RAMWolff said:

My only niggle with 3DL is the really slow rendering time for anything with transparency.  I'm not a patient person when it comes to renders so having tricks to speed things up is good but if this shader has those tricks built in I'll be over the moon! 

Same here.

My shader has explicit handling of opacity maps. You still need to use kettu's script (which will be bundled with the base shader) and use progressive rendering. The shader optimizations are geared for use with UE2's indirect light or bounceGI mode. It doesn't do that much in ambient occlusion mode. Honestly though, with kettu's script indirect light and bounce GI is as fast as ambient occlusion. And if DAZ actually allow raycaching to be enabled in the renderer's settings, it will work without kettu's script (allowing really, really fast IPR renders). So put in the feature requests.

I wanted to push 3delight as fast as possible in these areas - raytracing, opacity handling and physically plausible materials. Once this shader is done, I'm going to work on some new lights (not lights setup).  As I noted to kettu, hard coding Russian roulette into the shader wasn't an easy task, but that extra effort is responsible for really accelerating raytracing. And it's worth every compile error along the way. But really, the 3delight devs should be commended too for just how fast, powerful 3delight really is.

Mustakettu85 said:

And if you take even a newspaper - supposing newspapers elsewhere are printed on the same type of paper as in Russia, of course - it's paper that is thin almost like "kleenex" tissue. It will be translucent even when there is a few layers, and you will see the shadow of the object on the other side of the paper, but the cast shadow will still be solid. So it's more like "internally fuzzy paper", "scatter-y paper" that I meant.

With layered models, the resulting shadows gets quite dark (as you expect). I don't have a proper layered paper model in my runtime, but I saw this behavior on hair props.

Mustakettu85 said:

So there are times when you want the translucency on the material itself but you don't want (extra computation in the) more complex shadows. I'd throw in a switch =)

Something like this?

 

https://www.nda.ac.uk/blog/translucent-concrete/

A feature request? Sure, I can add a switch to control the resulting translucency shadows. I'm guessing you want a slider to control the intensity as well?

AndyS said:

One remark to your latest example:
I tried to reproduce this setup with several available materials in real world physics in my household.  --> This is unreal! You never get the clear shadow transition. Only a strong scattered light / shadow distribution.
OK - Perhaps if your material has got an opacity significantly below 1 (Opacity mixed with translucency). So a texture with a dense array of many mini-holes in the surface. But even the fabric of an umbrella doesn't show this outcome.

For fantasy pupose this looks nice - however.

Yep. That test scene is set with the plane set to be 100% translucent with a very strong distant light. Something like this?

Since your input is similar to kettu's, I can integrate a switch to allow users to turn on/off the effect.

RAMWolff

WOW WOW WOW.. no wonder they call you WOWIE!