Questions about V-Ray

[stevejjd]

Hi guys.
I just wanted to say that I love V-Ray so much. I had a great experience doing a freelance job with a client who was screen sharing. I was able to make so many quick changes and the look dev was a really easy an quick process. Im very happy with V-Ray and also have some questions too.

as a quick side note, I would recommend you try learning something like RenderMan, as it actually amplifies your knowledge of each renderer. My question come from having learnd a little in Renderman.

1. Can V-Ray make their version of the Disney BRDF? https://disney-animation.s3.amazonaw...f_notes_v2.pdf
- All Renderman shaders have an automatic GGX specular inclusing the skin shader. For some reason the quality of the secular GGX in the Pixar shaders look better and are way more easy to control. Can we mimic this in V-Ray? I am trying to reseach how Renderman treats the GGX specular in relation to the IOR. I really dont like the VRay GGX in comparison as it has a differnt quality and controlling it so hard. I also dont like blending two phongs to fake GGX. Can the default material have an easier to use SSS control like the disneyBRDF ? - Just basically mimic the Pixar shaders please haha .

2. Can V-Ray make the Marschner hair model? http://graphics.pixar.com/library/Da...ing/index.html
- Can we drive hair color with a melamine percentage and dye color like this?;

w = max( pow(floatInput1, 2.0) * 33.0, 0.02);
exp(w * -[0.187, 0.4, 1.05])*colorInput1

# compute a perceptualy linear absorption weight
# from melaninDensity
#
$weight = max( pow($melaninDensity, 2.0) * 33.0, 0.02);

# The fiber color is actually an absorption term. As light
# travels through fibers, melanin particles filter out some
# wavelength. We compute it using Beer's Law.
#
$fiberColor = exp($weight * -[0.187, 0.4, 1.05]);

# Now we model the dying process as a thin transparent
# film over the fiber that will filter transmitted light.
# This is simply a multiplication.
$dyedFiberColor = $fiberColor * $dyeColor;

# Output our result
#
$dyedFiberColor

3. Can Metals please work without the need for custom fresnel curves? A person told me "renderman does this, Nd is your usual IOR in LM materials, k = extinction coefficient in LMMetal. To use measured data from http://refractiveindex.info put n value into refractive index of LMMetal and k value into extinction coefficient; for dielectrics even simpler, just put n into refractive index of LMPlastic if it is solid or LMGlass if it is transparent. K is responsible for most of this "kink" in metals, so if you have both n and k, a complex IOR is kinda sorted, there is no need to adjust something by hand IMO, you will not notice difference in the best case and most likely make it LESS realistic if anything." Another person said to me "I use the custom curve inside Vray all the time because by default it has no way to have accurate metal fresnel. Vray is behind the times and hasn't caught up to the current shader models like Arnold and Renderman and Disney's principled shader. Heck even Mantra has a full PBR shader. Vray needs to get with the times"

4. Denoise images, you can render images with noise and denoise them.

5. Does Vray use multiple impotence sampling?

6. is it a good idea to have progressive rendering in the bucket rendering mode?


Please dont get me wrong, I love VRay, its just after starting to learn Renderman I feel like Vray needs some improvements. One thing I did notice is that I would never use Renderman yet to do professional projects, its just not as fast as VRay. Vray is like beast! it punches so far above its weight. I just hope for some improvements inspired by the amazing RenderMan.

What do you guys think?

[Svetlozar Draganov]

Regarding questions 4,5 and 6:

4.Denoising images is already supported into the latest nightly builds.
5.Multiple Importance Sampling is used in many V-Ray features.
6.Would you please give us more information about that approach?

Regarding the first three questions I don't think we have a support about them but I'll invite someone from our team with more knowledge on the subject to confirm that.

[^Lele^]

as a quick side note, I would recommend you try learning something like RenderMan, as it actually amplifies your knowledge of each renderer. My question come from having learnd a little in Renderman.

Eheh, that's debatable. (i have worked with renderman, on purpose, as well. And came out with a few good ideas, and a lot of underwhelming experiences...)

1. Can V-Ray make their version of the Disney BRDF? https://disney-animation.s3.amazonaw...f_notes_v2.pdf
- All Renderman shaders have an automatic GGX specular inclusing the skin shader. For some reason the quality of the secular GGX in the Pixar shaders look better and are way more easy to control. Can we mimic this in V-Ray? I am trying to reseach how Renderman treats the GGX specular in relation to the IOR. I really dont like the VRay GGX in comparison as it has a differnt quality and controlling it so hard. I also dont like blending two phongs to fake GGX. Can the default material have an easier to use SSS control like the disneyBRDF ? - Just basically mimic the Pixar shaders please haha .

Well, the Disney shader is cute, but it doesn't, for instance, deal with any refraction, nor, incredibly, is Energy Preserving (it breaks it without second thought in no time.).
But back to your question: I scripted one, for the fun of it, a year and a half ago.
Impressively, it still works, and i still use it personally, but i stopped short of adding the texturing support to the various layers and channels, purely because of the wall of text that'd have resulted in, and the fact the scripted shader can make a copy of itself in non-encapsulated form, so that it can be then edited like a normal vrayBlendMtl.
Back then, it was "acquired" by ChaosGroup, but a little later *I* was acquired by ChaosGroup, and ended up caring about very different stuff altogether.
In any case, a good starting point is here:
http://forums.chaosgroup.com/showthr...153#post657153
So we'll have to see what to do with the shader (you'll notice the edit in the post saying "THEY" will know what to do. Never i thought i'd be part of them...), but the point is that i'd personally MUCH prefer a radically different approach, rather than pre-packaged ubershaders, so i wouldn't vote for one such approach... :P
If you're a Max user, however, feel free to PM me, and i'll send you the script, provided you will NOT ask for any support on it, for none will be given (what doctor would want to know about the decaying health of a zombie, after all?).

2. Can V-Ray make the Marschner hair model? http://graphics.pixar.com/library/Da...ing/index.html
- Can we drive hair color with a melamine percentage and dye color like this?;

w = max( pow(floatInput1, 2.0) * 33.0, 0.02);
exp(w * -[0.187, 0.4, 1.05])*colorInput1

# compute a perceptualy linear absorption weight
# from melaninDensity
#
$weight = max( pow($melaninDensity, 2.0) * 33.0, 0.02);

# The fiber color is actually an absorption term. As light
# travels through fibers, melanin particles filter out some
# wavelength. We compute it using Beer's Law.
#
$fiberColor = exp($weight * -[0.187, 0.4, 1.05]);

# Now we model the dying process as a thin transparent
# film over the fiber that will filter transmitted light.
# This is simply a multiplication.
$dyedFiberColor = $fiberColor * $dyeColor;

# Output our result
#
$dyedFiberColor

The hair shader in v-ray IS based on marschner's paper, among others.
http://docs.chaosgroup.com/display/V...rencesandLinks
You can assume the melanine amount is taken into account internally.
Once rMan 20 (19 had hair completely broken) got its act together, it rendered exactly like us. Just, two years and a bit later.
I have visual proof of it, if you really want it, but would have to ask if it's ok to show it publicly, or just under NDA.

3. Can Metals please work without the need for custom fresnel curves? A person told me "renderman does this, Nd is your usual IOR in LM materials, k = extinction coefficient in LMMetal. To use measured data from http://refractiveindex.info put n value into refractive index of LMMetal and k value into extinction coefficient; for dielectrics even simpler, just put n into refractive index of LMPlastic if it is solid or LMGlass if it is transparent. K is responsible for most of this "kink" in metals, so if you have both n and k, a complex IOR is kinda sorted, there is no need to adjust something by hand IMO, you will not notice difference in the best case and most likely make it LESS realistic if anything." Another person said to me "I use the custom curve inside Vray all the time because by default it has no way to have accurate metal fresnel. Vray is behind the times and hasn't caught up to the current shader models like Arnold and Renderman and Disney's principled shader. Heck even Mantra has a full PBR shader. Vray needs to get with the times"

http://docs.chaosgroup.com/display/O...Fresnel+shader

4. Denoise images, you can render images with noise and denoise them.

It's being refined as we speak, but it denoises faster and WAY better than rMan (again, data in hand.), as it is right now in the nigthlies.
3.40 will have it included, with a supremely usable UI (especially when compared to the rMan approach), and a LOT less fiddl;ing required by the user.
Also, it works each time you want it to, not when it feels like it, and it works on individual AoVs as well, if that's your choice, so you can move the denoised AoVs to post.

5. Does Vray use multiple impotence sampling?

Are you referring to the fact V-Ray has no kids, there? If so, not to worry, it's Potent enough, just hasn't found the right partner to commit to, yet.

Jokes apart, you can assume that any modern raytracer (read: after Pov-Ray, or thereabout) uses MIS.

6. is it a good idea to have progressive rendering in the bucket rendering mode?

It has pros, and cons.
Those which benchmarked both approaches preferred to do progressive sampling on the whole frame at once.
It also works in DR mode, although the performance isn't likely as good as bucketed rendering, but then again, it wouldn't get all that better sampling progressively by the bucket (it's good to give remote machines, and cpu cores, a wee bit of work to do at once.)

Please dont get me wrong, I love VRay, its just after starting to learn Renderman I feel like Vray needs some improvements. One thing I did notice is that I would never use Renderman yet to do professional projects, its just not as fast as VRay. Vray is like beast! it punches so far above its weight. I just hope for some improvements inspired by the amazing RenderMan.

That was the amazing SPIDERman.
Don't get confused, now.

Please keep the questions coming, it's healthy and good to have a debate.

[vlado]

1. Can V-Ray make their version of the Disney BRDF? https://disney-animation.s3.amazonaw...f_notes_v2.pdf

I have thought about it; I am not sure yet we want to do that. There are many choices in that BRDF that are arbitrary and not physically correct.

2. Can V-Ray make the Marschner hair model? http://graphics.pixar.com/library/Da...ing/index.html

The V-Ray hair shader is already based in part on that; you can put that code into an OSL texture if want to use it to drive the hair color. Let me know if it's worth it

3. Can Metals please work without the need for custom fresnel curves?

This is somewhere on the "to do" list, but in the meantime you have the OSL texture here:
http://docs.chaosgroup.com/display/O...Fresnel+shader

4. Denoise images, you can render images with noise and denoise them.

See here:
http://forums.chaosgroup.com/showthr...noiser+testers

5. Does Vray use multiple impotence sampling?

Yes, since about 10 years now.

6. is it a good idea to have progressive rendering in the bucket rendering mode?

It depends on how your renderer is implemented and is mostly irrelevant to the final result.

Best regards,
Vlado

[stevejjd]

Thank you Svetlozar, Thank you Lele and Thank you Vlado ~

Okay I'll keep some questions coming in...Ill try and hit you a few more times

Lele: That shader you wrote looks really tantalizing but if your saying its a decaying zombie then I'll have to leave it. I was hoping for a really simple shader I can use in production that has all the good things of a RenderMan and V-Ray shader but none of the bad things. In terms of seeing hair comparisons, I would love to see any research from Chaos group that may help me, I don't mind signing any non disclosure papers.

Vlado I like the idea of non physical shaders as long as they look good and are easy to use. haha I can tell your eyes are rolling. I love the idea behind the physically plausible BRDF...I love the principals that made that Disney shader about being artist driven and plausible not necessarily physical.

Request: I know this request may be outrageous but, would you be able to provide me this "OSL texture" with that hair melanin code in it? That code I got from a Pixar guy who got it from a Weta guy. Im not smart enough to do that. Here is the weta paper and the code:

http://www.eugenedeon.com/wp-content...4/egsrhair.pdf
w = max( pow(floatInput1, 2.0) * 33.0, 0.02);
exp(w * -[0.187, 0.4, 1.05])*colorInput1

Q: If the V-Ray hair is based on the Marschner technique then can the user access various controls like eccentricity and cone angle (and more controls as seen via the link)? https://renderman.pixar.com/resource...chnerHair.html

Q: Do you know if there is a difference between how V-Ray calculates specular as compared to Disney or Pixar shaders? The Pixar ones have an automatic GGX model on all shaders but I think there is something else at play which gives easy to control beautiful specular reflections. I would like V-Ray to try mimic whatever that is. Many times I get a glossy sheen coat that appears on grazing angles of models when using V-Ray on GGX and sss shaders.

Q: Will Chaos group implement "Multiple-Scattering Microfacet BSDFs with the Smith Model"?
https://eheitzresearch.wordpress.com/240-2/

Q: I have a V-Ray licence so does that mean I can access the nightly releases and test denoising?

Q: Can there be a check-box to scale the lights independent of intensity. So you can use scale for shadow sharpness without having to change the intensity?

Suggestion: I have another very, very important suggestion that I hope you take notice of. I see very amazing example scenes shown by Chaosgroup showcasing various V-Ray demonstrations which are of a very high standard. I wish that the V-Ray help files also had these professional quality scenes with brief tutorials on the how and why they are put together. This is probably the most important thing. It would be completely unique among render engine companies; to provide very high quality scenes and breakdowns.

Suggestion: Would it be too insane to consider partnering with Hair-Farm, and helping them to develop a new viewport for 3ds max which would help visualize hair grooming in the viewport without the need for rendering? I have already posted this on Hair-Farm forums but I thought, 'what if there is a chance Chaosgroup could partner up or collaborate to make something incredible?' (XGen and mayas Viewport 2.0 is an awesome combo - 3ds max viewports suck so bad. I wish chaos group could make an awesome viewport for max (not RT - a viewport) in such a way to allow for better hair grooming and view-port feedback. - I know this suggesting is very far fetched.

Thank you...Ill add more questions and suggestions as I think of them.

All the best.
Steve.

[stevejjd]

"impotence" LOL sorry, you are a funny bastard Lele hahaha sorry I often make spelling mistakes. - These forums are a real suppository of wisdom. thanks for those links.
actually I was very impressed with your story, about how you ended up working for chaos group. Congratulations.

Will there be any announcements for VRay 4 ? Will I be able to upgrade since I own a licence?

[^Lele^]

"impotence" LOL sorry, you are a funny bastard Lele hahaha sorry I often make spelling mistakes. - These forums are a real suppository of wisdom. thanks for those links.

Lol, glad you took it in the right way!

actually I was very impressed with your story, about how you ended up working for chaos group. Congratulations.

I'm VERY impressed too, by how LUCKY i am.
As to finding merit, or deserving congratulations, lol, that is the difficult bit to grasp, at least for me.
So i keep working, and pretend it's just another job... XD

[vlado]

Request: I know this request may be outrageous but, would you be able to provide me this "OSL texture" with that hair melanin code in it? That code I got from a Pixar guy who got it from a Weta guy. Im not smart enough to do that. Here is the weta paper and the code:

Yes, when I get a bit of free time.

Q: If the V-Ray hair is based on the Marschner technique then can the user access various controls like eccentricity and cone angle (and more controls as seen via the link)? https://renderman.pixar.com/resource...chnerHair.html

We could, but I don't know how to explain those parameters in plain words to artists and I don't think it is really needed. We will be adding a highlight shift parameter at one point though.

Q: Do you know if there is a difference between how V-Ray calculates specular as compared to Disney or Pixar shaders?

From what Lele and I could test, there is no difference in the GGX hilights/reflections specifically.

Q: Will Chaos group implement "Multiple-Scattering Microfacet BSDFs with the Smith Model"?

Not in the immediate future; as that paper points out, the implementation has severe performance impact and in many cases it doesn't seem justified. If they can improve the performance, then we can certainly think about it.

Q: I have a V-Ray licence so does that mean I can access the nightly releases and test denoising?

Yes, you can email me to vlado@chaosgroup.com for more information.

Q: Can there be a check-box to scale the lights independent of intensity. So you can use scale for shadow sharpness without having to change the intensity?

This is already there - just choose Lumens for the light intensity. Then you specify the total emission of the light regardless of its size.

I wish that the V-Ray help files also had these professional quality scenes with brief tutorials on the how and why they are put together.

Yes, as we are updating our help files right now, we also try to provide the scenes where possible.

Suggestion: Would it be too insane to consider partnering with Hair-Farm, and helping them to develop a new viewport for 3ds max which would help visualize hair grooming in the viewport without the need for rendering? I have already posted this on Hair-Farm forums but I thought, 'what if there is a chance Chaosgroup could partner up or collaborate to make something incredible?' (XGen and mayas Viewport 2.0 is an awesome combo - 3ds max viewports suck so bad. I wish chaos group could make an awesome viewport for max (not RT - a viewport) in such a way to allow for better hair grooming and view-port feedback. - I know this suggesting is very far fetched.

I think it would be best to direct this request to Autodesk; the viewport code is a closed book for us.

Best regards,
Vlado

[stevejjd]

Thanks Vlado, I really appreciate it.

[sharktacos]

Request: I know this request may be outrageous but, would you be able to provide me this "OSL texture" with that hair melanin code in it? That code I got from a Pixar guy who got it from a Weta guy. Im not smart enough to do that. Here is the weta paper and the code:

http://www.eugenedeon.com/wp-content...4/egsrhair.pdf
w = max( pow(floatInput1, 2.0) * 33.0, 0.02);
exp(w * -[0.187, 0.4, 1.05])*colorInput1

I have the code for OSL here: http://docs.sharktacos.com/vray/osl.html#melanin

[stevejjd]

I have the code for OSL here: http://docs.sharktacos.com/vray/osl.html#melanin

Thank you!