mia_material for realistic leaves

#161

Just found out that I have been missing a very important element for that last years, all my trees/leaves are rendered using the mia_material where the ambient occlusion doesn’t respect the leaves transparency.
I know there are many online tutorials making a maya blinn or lambert shader use of AO that respects transparency, but its almost 2013 and I only use mia_materials, so is there by any chance a way to make the built in AO of the mia_material SEE the material transparency as it should.

#162

Would this be an accurate summary of the reason for the multiply node? I’m trying to summarise this thread as part of a reasoning for my shading network set up.

For the translucency to work accurately there has to be a level of transparency – 0.25 in this case. Setting the translucency to 1 ensures that all light absorbed by the material is used as translucency and not transparency and using an adjusted version of the colour map retains the leaf vein detail as these are darker in the mapped image.

The issue comes with how the transparency setting affects the texture mapped into the ref_trans_col slot – by reducing its intensity and over brightening the light. This can be corrected by connecting the file to an input of a multiplyDivide node and setting its multiply value accordingly to make the final value 1. So with a transparency value of 0.25 the multiply value needs to be set to 4.

#163

There is a very critical issue that I never mentioned and its the AO of the leaves. I think presently we have everything right but we dont have a correct AO on the leaves, since the AO in the mia_material doesnt recognise the leaf cut-out transparency which is wrong because its the whole poly plane of each leaf that is casting the AO.

#164

Roy, could you post a rendering example of this AO issue? I’m imagining it properly (I think) but haven’t run into this issue yet before, myself. I use AO on the branches/twigs/trunk but don’t think I’ve ever tried it on leaves.

#165

AO just doesn`t recognize leaves transparency with the mia_material, it only sees a poly plane.

#166

I thought the issue had been resolved way back, but lo and behold. I’m starting to feel that things like this are becoming the final straw for me with MR. It’s 2014, damnit!..

Here’s a quick test with a checker in opacity. One render where both the floor and the plane has AO (with color bleed), and one with no AO. It’s subtle in this example, but I’ve had cases where it was anything but subtle. It messes up sampling and overall shading.

Granted, if you render with Brute Force FG and Unified Sampling, AO is handled differently, and not needed through the material, but it sure is annoying, and should just work…

#167

what do you mean? I always used Unified Sampling along with the mia AO? what does FG brute force change regarding o AO? what`s the workflow?

#168

Well the idea is that with Brute Force Indirect Lighting AO is redundant, as that happens automatically in a Brute Force rendering setup. Though things like these opacity map issues makes me doubt MR alltogether. So in essence if you render using Final Gather Mode set to "No FG Caching"—this is Brute Force Final Gather. See it as something similar to Arnold, or VRay’s Brute Force mode.

Beware that this is extremely slow in most cases, but also happens immediately with the sampling (so you won’t see FG build before render), but on the other hand usually doesn’t require accuracy settings higher than, say 40, for most well lit scenarios.

Man… I am bummed to see that this Opacity Map/AO bug is still around. The final straws with MR are adding up for me.

#169

“mental” ray is no misnomer, for certain. Ain’t no picnic!

Vray looks more attractive every year. With no new FX chips planned, perhaps I can get the boss to spring for it instead of the yearly hardware upgrade, this year.

#170

So when using Bute force, you simply urn off AO on mateials.
And o do that you simply set FG mode to no cachingwhich doesn`t generate FG flicker in animations, right?

#171

I’m not entirely sure if you need to actually turn it off. It should disregard it altogether during Brute Force, but I’d do a quick test with a checker texture to confirm. You never know with Mental Ray. :wink:
Yes, flicker should be eliminated with brute force. As everything is handled by Unified Sampling. You might need to add some filtering though (start with 1 in filter, and see if that suffices), then slightly up the accuracy as needed. This is to eliminate indirect artifacts which can happen during all things FG, as you probably know. 1 secondary bounce should suffice for 99% of the cases (not only Brute Force, just as a rule of thumb), so set secondary bounces to 0 and work your way up as needed.

Post any findings you make, always good to know.

Ps. Check out these posts on Elemental Ray; https://elementalray.wordpress.com/category/final-gather/

Cheers.

#172

Yes indeed, I’m considering getting 3.0 when it’s released for Maya… Though I tend to use Arnold more and more along side Mental Ray.

#173

Hehe, I recall in a vegetation thread (may have even been this one?) me suggesting tongue in cheek V-ray to you and you adamantly refuting my suggestion :stuck_out_tongue:

Though interestingly, the best vegetation I see is usually done in Mental Ray, I think because by its more difficult nature it requires a better understanding, thus leading to better results IMO.

#174

To be honest, my main issue with landscape scenes isn’t so much mental ray anymore. After awhile you kinda figure out the quirks and things that work and don’t work - hell, I spent two hours this morning messing with the stupid Camera Depth pass to spit out a proper z-depth again! So stuff like that seems to work better in Vray.

But for me, it’s the scattering/placing of instances and such that’s the real time consumer. I’m looking forward to playing with Xgen though, maybe that’ll save the day and let me spend more time creating and less time diagnosing scenes.

Vray would be nice for my arch/viz work, though I’ve used mental so long there really wouldn’t be many advantages at this point.