The 2-sided material - real life study vs the rendered version

[JohnVK]

Hmm.. so because the front material isn't receiving any lighting information directly it is calculated as "nothing", where we would expect the 2sided material to actually let light information of the spot bleed through the object (based on the translucency settings) adding it to the front material so it would show up.

I think I'm beginning to understand why I simply couldn't get some of my 2sided materials to look right in the past!

Looks like a problem in the 2sided material, no?

[svetlozar.draganov]

Hello there,

It is not a problem - it is just work this way. In order to get predictable result with it you have to have a least a bit of light in both sides of your objects.
In the example I post there is a sky override - if you disable it fully 2SidedMtl won't give you the 2Sided effect at all - but if you turn it on even with very low values - 2SidedMtl will start to display both sides of the object.

[JohnVK]

... In order to get predictable result ...

I think that's a matter of perception

In my case it was not what I expected to be honest. It's kind of counter intuitive. If I add a light on one side of a translucent material I expect light to pass through the object, not that I need to add another light in the scene to make the other side catch light as well to be able to see it.

I think it actually makes it more difficult to get a predictable result, don't you think? Since I now need to tweak the sky light / environment as well as the light behind the object as well as the translucency setting in the material? Or is the intensity setting of the sky light irrelevant for the 2sided material, e.g., intensity value of 1 gives the same end result as intensity value of 30?

On the other hand, now that I understand what I need to do and how the material actually works I can work around this "problem" It doesn't happen all that often that I have the material lit from one side only by one light anyway. I pretty much always have some GI going on.

[vlado]

Yep, there is something odd going on there, but I still can't put my finger on the code where it happens.

Best regards,
Vlado

[windowlicker]

Yeah, exactly. I can't put my finger on it - atleast not in any technical terms. I sometimes wish I had studied more maths at the university...

I wrote in the other thread (the one about leafs) that the light from the backlit side should be added to the front, but only if the backlight is stronger than the frontlight. I have no idea on how to describe that in mathematical terms, and you claimed Vlado, that it would break the energy preservation rule. I think the problem comes from the fact that there in reality is an SSS effect in the paper that lights up the actual paper, and thus making the frontside appear lit up. I think this is what has to be emulated in the renderer. Somehow. Being able to "add" the backside image to the front without breaking energy conservation is probably the key here.

I have some more photos that show what happens when you have a backlit paper and shine another, stronger light from the front. The interesting thing is that the front light "takes over" and again showing nothing of the backside.
Left is the pure backlit paper with the green front towards the camera. In the second photo I use a flashlight to shine a light on the front facing part of the paper. Now, you the green ink appears stronger (it's somewhat burned out though, unfortunately). In the last photo i flip the paper around, and here you can see none of the green where the flaslight is shining. Now this side takes over, completely excluding the backlight.

[Macker]

Any progress, anyone?

[vlado]

I'm still pondering this at the back of my head, but I've had a pile of other fires to put down in the meantime. I hope to be able to get back to this soon.

Best regards,
Vlado

[darloman]

Hi Vlado,

We are doing three fully-CG commercials, all featuring CG paper from beginning to end. We have been really stressed out trying to get the double sided material to behave properly for weeks now. Today I found this thread where the original poster clearly identifies the problem we are having and you seem to agree that it is a problem.

Has there been any progress in making the double sided shader behave more like real world materials? We still have two months to go on this production so there is still time to adjust our shaders to a new solution. It would be great to be able to light our shots in a simple way (without the skylight hack mentioned here) and have things just work. There is a LOT of backlit paper in these 3 commercials.

We are also wishing that glossy refractions would work with the double sided material and still respect the opacity map. We are using the hack around where you forget opacity maps and fake them by mapping refractive index and refractive colour both to 1 where you want your leaf (or whatever) to go transparent. Our networks would be a lot simpler if you could just plug in an opacity map and it trumped everything else. That is the beauty of them after all.

We are using Maya v2013.5 on Windows with Vray 2.25.01

Cheers!

Stuart

[Sakdon]

Hi there.

I'm trying other way.

make thin shell ,use 2side mtl, use <UDIM> texture for fornt&back, turn on GI , use one key rect light.

[vlado]

I don't have anything usable yet; Sakdon's approach certainly works although it might be a bit on the slow side.

Best regards,
Vlado

[Pixelcon]

Any progress on this, Vlado?

[vlado]

Yes, for the next service pack of 3.0 we are planning some more shaders and shader improvements so hopefully this will make it in.

Best regards,
Vlado

[Pixelcon]

Oh thank god! (vlado)

[windowlicker]

Awesome! Looking forward to it.

[vlado]

So this is now implemented for the next service pack - the two sided material has an option to filter the translucency effect through the front-facing side.

In the images below, there is just one light source, no GI.

Also, the two-sided material has been optimized to be much faster for GI, even when using color translucency.

Best regards,
Vlado