[ARCHIVE] Pre-1.0 Daily builds

[Bigguns]

Me again haha. If you think this post is not in the right lpace you can move it .

I think Keymaster you should really integrate the papers of this guy, there are the best I have seen  http://www.eugenedeon.com/

hair and skin shader are really good.. Weta used his work in their own renderer... I can see a skin shader like that but upgraded, with the first layer wich is refractive so we can see the fur connecter into the skin going  through (like in reality) and it with ABC reflection model ... that would be the killer skin shader of any apps!

some paper links:
http://cgg.mff.cuni.cz/~jaroslav/papers/2014-zerovar/
http://s2014.siggraph.org/attendees/talks/sessions/sampling

Apparently what renderman has made on the skin shader model is more accurate then the one of  eugene :  http://renderman.pixar.com/resources/current/rps/improvedSubsurface.html


Also his work on : Importance Sampling Microfacet-Based BSDFs using the Distribution of Visible Normals   look's very interesting... I let you check it Keymaster I'm sure it worth  integrate these.

anyway lot's of cool stuff to nitegrate :)

[Ondra]

hmm and more recently, there is a new (august 2014..) shader model that emerged ... is it even  better then ABC shader? http://www.cs.virginia.edu/~gfx/pmwiki/index.php?n=Pubs.Brady2014DNA

I saw the presentation of that paper at siggraph. I actually had the same idea 3 years ago, but never got around to implement it. Too bad :/.

The basic idea behind these papers is trying to fix measured data as closely as possible. Which is cool, but there are still some harsh facts:
1) using better fit for some measured data will not make you a better artist. What is currently in Corona is probably "good enough". I doubt BRDF alone would ever prevent you from achieving what you want in Corona, once GGX is in. Most of the times, the difference is not even visible.
2) BRDFs that are good fit for some measured data might not be good fit for other data, and may not be customizable or simple to use
3) the more advanced BRDF, the worse it is for importance sampling. Good importance sampling is essential for fast rendering. There  are some very advanced models that capture reality well, but since we cannot importance sample them, they are unusable in reality (unless you want to get 10 times more noise than before). Corona uses Ashikmin-Shirley and GGX, two BRDFs for which good importance sampling strategy is known, and that have intuitive parameters.

tl;dr: BRDFs are now probably good enough and it is time to deal with more pressing matters.

[Ondra]

Me again haha. If you think this post is not in the right lpace you can move it .

I think Keymaster you should really integrate the papers of this guy, there are the best I have seen  http://www.eugenedeon.com/

hair and skin shader are really good.. Weta used his work in their own renderer... I can see a skin shader like that but upgraded, with the first layer wich is refractive so we can see the fur connecter into the skin going  through (like in reality) and it with ABC reflection model ... that would be the killer skin shader of any apps!

some paper links:
http://cgg.mff.cuni.cz/~jaroslav/papers/2014-zerovar/
http://s2014.siggraph.org/attendees/talks/sessions/sampling

Apparently what renderman has made on the skin shader model is more accurate then the one of  eugene :  http://renderman.pixar.com/resources/current/rps/improvedSubsurface.html


Also his work on : Importance Sampling Microfacet-Based BSDFs using the Distribution of Visible Normals   look's very interesting... I let you check it Keymaster I'm sure it worth  integrate these.

anyway lot's of cool stuff to nitegrate :)

Dude, slow down ;)


anyways... you are linking paper by Jaroslav, who is now also developing Corona. In fact he did the SSS and GGX implementation in Corona. The last paper you mentioned is already implemented in Corona's GGX. It is not new BRDF, but optimization of GGX, and the reason why GGX is usable in production rendering.

[Bigguns]

Thank's a lot for the reply Keymaster, it clarify's some things, I'm not a programmor so I'm sending you things that I found look good and will be good in Corona but it's you that know if it worth it and all those details :)

Do you have any idea when the skin shader will be ready to test in the dailybuilds? :) I can't wait sorry, I have characters begging for it haha



Edit : I'm very please to know it's  Jaroslav Křivánek doing the SSS integration :) I'm confident it's going to be quickass skin shader as the rest of Corona is :)

[Ondra]

sorry, no ETA on the skin shader. It will be sometimes next year. For now, the basic SSS can be used for skin too.

[Bigguns]

Ho sad news :(

No the basic sss right now just look like plastic.. nothing to compare with a true skin shader. In volumetric mode it get closer to a skin shader.. but we can't use textures.. so..

I will have to wait a true skin shader, I think you underestimate the importance of that ( or maybe it's just impossible for now to do because of some constraint I don't know ^^ ). Vray took years to get one.. and it was only used for archiviz and those kinds of things.. because no studio in the VFX could use it without a skin shader (and hairs is really important too) except if they could write their own.. anyway , but now that they have one we have seen Vray growing more and more in VFX production Studios..(and artist like me that do characters, before most peoples were using Mr ONLY for that reason.. now that vray has a good skin shader.. what happened to MR... quite dead..)  To me skin shader and good sss is as important as GGX or any other mtl. I think you should get it the sooner so peoples can do super realistic render in Corona while in Alpha.. it would make big studio and peoples want to try Corona and you'll sell a lot of it if you have a good skin shader that has made his proof (by having some nice character render in your gallery) in the first release.

It's just based on my observations and friends in the fields but I think it represent a lot of peoples. I know archviz guy don't care that much about it but big studios and C.A. do.

[lacilaci]

Ho sad news :(

No the basic sss right now just look like plastic.. nothing to compare with a true skin shader. In volumetric mode it get closer to a skin shader.. but we can't use textures.. so..

I will have to wait a true skin shader, I think you underestimate the importance of that ( or maybe it's just impossible for now to do because of some constraint I don't know ^^ ). Vray took years to get one.. and it was only used for archiviz and those kinds of things.. because no studio in the VFX could use it without a skin shader (and hairs is really important too) except if they could write their own.. anyway , but now that they have one we have seen Vray growing more and more in VFX production Studios..(and artist like me that do characters, before most peoples were using Mr ONLY for that reason.. now that vray has a good skin shader.. what happened to MR... quite dead..)  To me skin shader and good sss is as important as GGX or any other mtl. I think you should get it the sooner so peoples can do super realistic render in Corona while in Alpha.. it would make big studio and peoples want to try Corona and you'll sell a lot of it if you have a good skin shader that has made his proof (by having some nice character render in your gallery) in the first release.

It's just based on my observations and friends in the fields but I think it represent a lot of peoples. I know archviz guy don't care that much about it but big studios and C.A. do.

Relax. I think that it is better for everyone to wait until Ondra and Jaroslav have as much time as possible to dedicate it to building some super nice skin shader which could be actually used in big production, instead of doing it quickly and then quickly fixing it every year and still nobody would want to use it cause someone would have better, faster and easier to use alternative...

And for the characters, there is still no hair shading (which i guess will also take a while to do, if it's meant to render quick and be usable) so skin shader wouldn't really find that much use right now.

I'm not saying by this that it is not important, but because it is important, it will be better if development takes enough time to do it right.

[tomislavn]

Skin shader is not quite that important for a renderer that is marketing itself as an arch-viz/product viz solution - For those branches a simple sss shader is enough, for now :) Then again, when Corona grows up completely and goes into commercial flow, I am sure that a lot of those "additional" features will follow. Same happened to V-Ray. You cannot have it all over night.

I would rather want to see some real dispersion added so I don't have to fake it or use v-ray for it :)

[Ludvik Koutny]

Ho sad news :(

No the basic sss right now just look like plastic.. nothing to compare with a true skin shader. In volumetric mode it get closer to a skin shader.. but we can't use textures.. so..

I will have to wait a true skin shader, I think you underestimate the importance of that ( or maybe it's just impossible for now to do because of some constraint I don't know ^^ ). Vray took years to get one.. and it was only used for archiviz and those kinds of things.. because no studio in the VFX could use it without a skin shader (and hairs is really important too) except if they could write their own.. anyway , but now that they have one we have seen Vray growing more and more in VFX production Studios..(and artist like me that do characters, before most peoples were using Mr ONLY for that reason.. now that vray has a good skin shader.. what happened to MR... quite dead..)  To me skin shader and good sss is as important as GGX or any other mtl. I think you should get it the sooner so peoples can do super realistic render in Corona while in Alpha.. it would make big studio and peoples want to try Corona and you'll sell a lot of it if you have a good skin shader that has made his proof (by having some nice character render in your gallery) in the first release.

It's just based on my observations and friends in the fields but I think it represent a lot of peoples. I know archviz guy don't care that much about it but big studios and C.A. do.

1, there is absolutely no difference between VolumeMTL and CoronaMTL when it comes to  SSS. Only difference is naming, But if you either set CoronaMTL opacity to 0, OR set refraction level to 1 and IOR to 1.0, CoronaMTL volume will behave exactly same as VolumeMTL.

2, First Corona release will not be aimed at VFX, so that's why Skin shader is not a priority. It's way down on the priority list. Sure we can have skin shader, but we would have to give up one of lot more important things, such as flicker-free secondary GI, multilayered material or interactive rendering.

[Ondra]

yep, volume mtl will be probably removed, it is now causing too much confusion.

[juang3d]

Hey the skin shader can be used in character animation too... And that is not VFX.

IMHO Corona should not be targeted towards arch in its first release, everybody does the same thing, I think Corona is able to deliver a better product that can demonstrate it's enabled for other tasks, even without skin shader, which I don't think is so so important while you have SSS.

But hey, it's just my opinion, Keymaster and team know their product better than anyone :)

Cheers!

[tomislavn]

Hey the skin shader can be used in character animation too... And that is not VFX.

IMHO Corona should not be targeted towards arch in its first release, everybody does the same thing, I think Corona is able to deliver a better product that can demonstrate it's enabled for other tasks, even without skin shader, which I don't think is so so important while you have SSS.

But hey, it's just my opinion, Keymaster and team know their product better than anyone :)

Cheers!

I wouldn't worry too much about it - Pretty sure its gonna have it all, sooner or later :)

[Ondra]

I would like a ton of things. But we are working on a limited budget: monetary, time, and energy. So we have to make compromises, like the first release being archviz-oriented.

To be honest, lately I am getting frustrated from getting feature requests or suggestions that are outside any realistic time plan, always supported by that one sentence/paragraph "and this is the most important thing because...". There are like 40 of those floating around. Please realize that we are also only people and while there are many important things, we cannot handle them all at once.

[zzubnik]

I would like a ton of things. But we are working on a limited budget: monetary, time, and energy. So we have to make compromises, like the first release being archviz-oriented.

To be honest, lately I am getting frustrated from getting feature requests or suggestions that are outside any realistic time plan, always supported by that one sentence/paragraph "and this is the most important thing because...". There are like 40 of those floating around. Please realize that we are also only people and while there are many important things, we cannot handle them all at once.

There will always be too many suggestions from users, you guys seem to know what you are doing, so keep on as you are. There's always time later to expand on what you have built. Everyone I've showed Corona too has been stunned; you are doing something right!

[juang3d]

Hey Keymaster don't understand me wrong, I don't mean you have to add any feature more to have a character animation based mnarketing for Corona, Corona has the most importnat things for that, Render Elements, shadow catcher, motion blur in deformable objects and camera space and SSS... what else do you need to conform a character animation oriented render engine? It's not a VFX render engine because it lacks some basic features for it like a proper volume material with density inputs or the ability to render billions of particles points in a superb speed, but that is VFX, character animation is a completely different thing, and any small/medium studio can deal with it using corona in it's current state (using HDCache with the corrent setting syou won't get any flicker)

Yes a skin shader could be good, and as you say tons of things could be good, but what I mean is that there is no need to orient Corona towards arch&viz exclusively, Corona is capable of doing much more in it's current state.

Cheers.