[Fernando Tella]

The light looks very nice. With the texture it will look great.

By the way, I've been trying to render some leaves with diffuse transmittance too on single sheet surfaces and the results were quite weird too.

[tokiop]

the backlighted plants look very nice !

[Bubbaloo]

I let this one cook overnight to s.l. 25:

[Fernando Tella]

Still a bit too transparent?


Edit: duplicated post? I clicked "edit". Hmm.

[Fernando Tella]

Still a bit too transparent?

They look fresh though.

[Bubbaloo]

Yes, I think you are right.
Yes, I think you are right.

[Richard]

Thanks Rusteburg and numerobis!!

Well... I don't want to take this thread in the wrong direction, but...

What is single sheet subsurface scattering? It's an interpolation of a physical property. It's interpolating what "should" be happening beneath the surface of a solid object. It doesn't exist in the physical world. In our real 3d world, even the thinnest objects have thickness. Wouldn't using S5 make Maxwell Render non-physically correct? That's what I like the most about Maxwell Render. The more accurate your model, the more accurately the materials can react to the lighting in your scene. Plus, with better hardware and low memory prices, keeping a low poly count is becoming less important. Just my thoughts on it...

Mate some great tests there! Those with SSS are very convicing!!

I must say I sort of disagree with the suggestion that all has to be physically correct - we are doing digital models so to start with they are NOT physically correct!!!

I don't think that the wanting for faster, lighter and more simple options is at all a bad thing. So way of achieving (faking) SSS is just another way to speed and simplify processes. If it means closer to realistic results at little cost - to me it's not a bad option!

I applaud your tests and the results but at the same can see benefit of other solutions!

Look forward to seeing much more mate!

[Richard]

Actually mate I thought to back up this suggestion I'd run a quick test on the possibility of faking SSS.

I'll post the results when complete! What I've actually done is to produce a plastic type material (semi glossy 2 BSDF layers) with two additional low weighted BSDF layers, to each of these I have used a fine B & W noise map in the transmission channel - one normal and one inverted so that from both layers light transmits directly through > one being inverted means this transmission is uniform over the material.

I can see from the test running now the material has these two additional layers weighted to high though the results are showing nothing of the problems of just applying transmittance to a single plane material. I'd suggest however that once made effective the structure of the leaf could also be mapped to the noise map to give greater realism there. If you look at your last (very nice image and test BTW) the leaf structure is lost using just SSS.

I will admit that i don't think the results initially will produce the same quality as your examples though I feel unless the material is being viewed very close up it may produce rather acceptable results without the need for additional geometry or slower times created by the use of SSS.

[Bubbaloo]

I'd like to see your tests. Post them here if you want. I have been swamped here at work so I haven't been able to continue my tests. One thing I've noticed about using SSS is that a small amount of it in a material (such as a thin leaf) doesn't affect the render time hardly at all. But there is still the matter of more polygons required that is a drawback.

[Richard]

Mate here is my test. Sorry I didn't really set up a good one nor did I chose a great plant or for that matter do a great job on the material but it does give some idea I guess.

I didn't place the map to the leaves well but did put some stripes on the map so it shows how the viens or skeleton of the leaves could be achieved. It isn't so evident though you can pick them up a bit (the diagonal lines running accross the leaves).

BTW the left plant (no mapping) the right one (mapped as tested).

I'll try to set up a scene and play with better materials and leaves to see what I can come up with!

[Richard]

Mate here is another quick test with a better material. It is easier in this one to note the skeleton of the leaf has been preserved by combining the bump map with the noise map to create the transparent areas. No noise (white) areas have been placed where I want the leaf to be non transparent.

Again I've possibly given too much weight to the transparency layers.

[Bubbaloo]

Here's a close-up of the leaves with sss:

[Richard]

Like all you've posted mate they look stunning!!!!

Do all the leaves have added geometry for thickness - it looks like the leaf is clipped?

What I was refering to in loss of leaf structure was in that of your last posted image where the leaves with the sun behind become homogenous wi the SSS. Maybe a transmittance map for that may work or even differing levels of SSS weighted.

Great work mate!

[Bubbaloo]

Thanks, man. All the leaves have 1/128" thickness and you're right, I did use a clip map on these. Also, the sss layer has a weight map so that the effect isn't uniform.

[Richard]

WOW!!!

I didn't think the result would be right if the leaves where clipped to a volume. I would have thought that using a clip on two planes would have just produced two opposing planes with an open edge?

But maybe not so!!!!

As you can see from my image all the leaves become transparent (though I over did it) but still the result isn't right, though doesn't have the added geometry issue!

Would be great if you could apply a thinkness to a mapped surface similar to how displacement adds height!