VRay leaves...

[tolgahan]

this effect is still trying to catch

[tolgahan]

I'm writing a short tutorial on that, I'll post it a bit later.

Best regards,
Vlado

Wow great..

[windowlicker]

2 sided doesnt simple mix front and back. you can set easy different/seperate speculars and reflections for both sides. (make sure to have "back side" disabled in reflection)
to my knowledge and testing it does take the material aspects into acount too (surface specularity/reflectivity) - if you make the opposite site strong reflective, only few light travels through the 2 sided, as this is in real. it is easy to test yourself (make it very strong reflecting and almost no light will travel, if only little reflection (best with ior) light will travel a lot more, as expected in reality.

so i think it is very realistic, also how it deals with shadow and transporting light from one side to the other, more or less blurred depending on distance of the object is quite nice.

stefan

Well, I beg to differ. I set up a simple scene to demonstrate the problem. The scene can be downloaded here: http://www.mediafire.com/?fq9cqlq3idwhvzx Let me know if you think there's something wrong with my setup.
The scene consists of a light, a completely black atmosphere so no ambient GI will affect the objects. A checker texture is used as a reflection override to see where the reflections are present and not. The object is a plane with polygons facing towards the camera. 2 bend modifyers are used to give the plane a slight curve.

The front material is as follows: diffuse-220RGB, reflect-255RGB, refl.gloss-1, fresnel-1.6, reflect on backside-disabled.
The back material is as follows: diffuse-128R 8G 8B, reflect-64RGB, refl.gloss-0.25, fresnel-1.6, reflect on backside-disabled.
They are loaded into three different 2sidedmaterials - one with 0%translucency, one with 50% and one with 100%. "Force single-sided sub-materials" is used in all of them.

This is the render using 2sided material:


In the front lit bottom row, the front material disappears and the reflection diminishes as the translucency increases. In the backlit top row, the reflections of the front materials also disappear when the translucency increases.

I included the reflection pass from this render where it is even more visible.



This is the result I would expect from the 2sided material, if it was to be used as paper or leaf material.

The reflective side would keep its reflection no matter how strong the backlight or how high the "translucency" (a value that in vray is a rough estimate of the materials thickness and translucent properties, really). You would also expect color to transmit through the material. In this case I have one white side and one brightly red. In a real case, a backlit "paper" would be red, no matter from which side you wiew it, as long as light travels from one side to the other. You can easily recreate this in real life by getting a glossy magazine, and viewing it in different lighting and reflective environments (for example with a window behind you, and the light in front of you, shining through the magazine cover).

I include one last render where I switched positions of the red and white materials where you can observer the lack of color bleed.



All of this is to show the basic "flaws" of the 2sided material. It is in its current state however the best option for rendering leaves, but you have to keep the translucency at a very low value to get rid of the dull look which comes when you crank it too high. This will of course make the leaf transmit less light trough, and the foliage will be darker than it should be. This can be compensated for in post, of course.

One other interesting thing with leaves is that they are full of chlorophyll, which is intenselt green and thus tints the transmitted light. I usually set the translucent color to a mild green to get that effect, especially if the backside leaf material is grayish.

The 2sided material might be working as intended, but it is not working as a piece of paper or a leaf. Atleast if you ask me. :P

[Macker]

I'm willing to bet that "conservation of energy" is the problem here.

[vlado]

Vlado, any tips on achieving hich specularity and high translucency at the same time in a 2sided material?

But leaves are not highly specular, certainly not compared to mirrors and metals.

Best regards,
Vlado

[vlado]

Here is what I got so far:

http://www.spot3d.com/vray/images/stuff/leaf_tutorial/

I forgot to take a photo of the actual plant at home for comparison, will do that tomorrow.

Best regards,
Vlado

[Macker]

Also; does the 2sided mat actually let light through onto other leaves beneath, or does it just show the lighting on the back face through the mesh?

[vlado]

This is the result I would expect from the 2sided material, if it was to be used as paper or leaf material. The reflective side would keep its reflection no matter how strong the backlight or how high the "translucency".

Eh, no, not really. This would break the energy conservation law, if done in exactly this way. However, nothing stops you from putting the 2sided material as a base material inside a VRayBlendMtl material, where the reflections are in their own coat layer. In that way, the translucency value of the base 2sided material will not affect the reflections in any way. You'll have to uncheck the double-sided option in the coat material to avoid reflections on the back side.

There is indeed a slight flaw in the 2sided material (its BSDF is not symmetric in math terms), but it's completely different thing and it doesn't really affect the results visually so much. The reflections would still behave in the same way anyways.

Best regards,
Vlado

[vlado]

Also; does the 2sided mat actually let light through onto other leaves beneath, or does it just show the lighting on the back face through the mesh?

If you use GI, then it will transmit light, yes.

Best regards,
Vlado

[BBB3]

nothing stops you from putting the 2sided material as a base material inside a VRayBlendMtl material, where the reflections are in their own coat layer

Good point.

[windowlicker]

Eh, no, not really. This would break the energy conservation law, if done in exactly this way. However, nothing stops you from putting the 2sided material as a base material inside a VRayBlendMtl material, where the reflections are in their own coat layer. In that way, the translucency value of the base 2sided material will not affect the reflections in any way. You'll have to uncheck the double-sided option in the coat material to avoid reflections on the back side.

There is indeed a slight flaw in the 2sided material (its BSDF is not symmetric in math terms), but it's completely different thing and it doesn't really affect the results visually so much. The reflections would still behave in the same way anyways.

Best regards,
Vlado

Well, I'm no expert on the physical calculations and the programming of vray - I'm just trying to recreate materials and objects I know in reality. So far I have always had trouble with recreating paper and leaves with vray2sided, since I lose reflection and color when working with it.
I'll see if I can get around to doing a photographic study sometime. In the meantime I can just make the suggestion to pick up a piece of paper - glossy or not - but preferably with different colors on different sides and then observing the results. I have never seen IRL what I see when using the 2sidedmtl.

But if we for a second leave the reflective part aside, and look at the transmission of light, which really is the main issue, imo - what can we expect from the 2sided material?
I made a second version of the scene, this time with the same white material back and front. I added a thin object to cast a shadow on the back of the backlit objects.



Holding a paper up to the light you would expect to see something like the top left object - 100% translucency works fine here. But when that material is lit from the front, it's completely unrealistic. You basicly only see the back of that paper.
The 50% material is decent, but you still get a much darker front lit paper than expected. Also the transmission of light through the paper could be better.
The version with 0% translucency looks good from the front, but lets no light through at all.

Without being able to express this in correct mathematical terms, I'd say the light transmission through thin objects works like this: Front light component + (back light component - whatever light is being absorbed by the transmission). So if the front light component is 1 and the back light component is 0, the end result will be 1. If the front light component is 0 and the back light component is 1, the result will be nearing 1 but never actually 1, since some light is always being absorbed in the transmission. This being based on real life observation... with a paper and a desk lamp.

[tolgahan]

I am trying.Great results.Comptex getting better results by using the translucency is increased.

[lllab]

windowlicker, i do not think a material woudl behave like you suggest in reality.

maybe an example from real world: if there is light energy incoming on the surface(front), lets call that 100% energy, it can get absorbed , but can't increase, it has to be 100% or in reality much less, so if the reflection would stay icentical, the translucency cannot get more, otherwise light (energy) would get more by itself out of nothing, which obviously isnt possible in reality (beside self illuminated things/lights). so its always a balance what happens withthe 100% light enegry, some gets reflected, some absorbed, and some is part of the translucency. if you use IORs in specular this shoudl be even behave correctly to the angle of the lights and camera.

[vlado]

Holding a paper up to the light you would expect to see something like the top left object.

No, I don't really expect that Like I said, there is a slight flaw, but you haven't nailed it so far (whereas our internal QA did it from the first try )

Without being able to express this in correct mathematical terms, I'd say the light transmission through thin objects works like this: Front light component + (back light component - whatever light is being absorbed by the transmission).

No, unfortunately it doesn't work like that. If it did, it would break the energy conservation law and would produce artifacts all over the place.

Best regards,
Vlado

[windowlicker]

Haha you guys just won't break! I like that you're taking the time to discuss this. I'm gonna take some photos of what I mean, then we can take it form there.

But you see nothing wrong with the front lit render of the 100% translucent material? That it picks up almost nothing of the light coming from the front, and instead showing the unlit backside? And yet it is clearly visible when backlit? And that there is no red front color showing at 100% translucency and a white back material? I'd like to see those situations in real life.
Is there a limit to how much translucency you can use before the renders start becoming unrealistic?

For me the problem has always been that I'm not getting enough light through the leaves I render. The thing I do then is I simply increase the translucency. What happens then is that the leaves get dull and lose reflection. The ideal situation would be that you could increase translucency without losing color and reflection. I'll try your new tutorial though. It looks rather interesting.