userIbl shader from mr 3.10

[Kzin]

keep in mind that the geo shader and the way it sends its data to mr is xsi related and not a mr problem. i wanted to make this clear one more time that ad is responsible for things like this, passing data to mr.

ok, tested it more deeply, builtin_ibl is working now properly, BUT only for render region. means in other words, useless for final rendering because the scene is not correctly updated and passed to mr.
i think its the new render region feature that makes this happen. now you can only update the region when you move the cam for example, so ad had to change the way the region sends data to mr. but they forgot to implement this new way also for passes rendering (i am pretty sure they dont even know what they do, looks alot like this).

set fb managment to none makes the whole thing working, it looks like this. but will also have to make more tests.

EDIT: it works now, also on renderpasses with animated env intensity. of yourse final rendering in a farm needs to be tested seperately.

EDIT2: no, it does not work, dont know whats going on, it works in one second and failed in another one.

[Kzin]

ok, would be great if some of you could test this because i dont trust my own tests anymore in terms of user_ibl. ;)

in the rendersettings, under shadows, set the type to enabled instead of segment and render. it looks like the user ibl shadow shadow mode 2 is working with this setting, tested it in a simple scene with transparent red glas and all is working like expected.

[Maximus]

Yes it works. At first i didnt say anything because i've modified my mental ray options spdl to start always with Shadows enable, since this is how it should have been by default. So everything was working to me.

Anyway i can confirm that if you switch from enabled to segmented they dont work anymore. So keep it at enabled.

[Kzin]

Yes it works. At first i didnt say anything because i've modified my mental ray options spdl to start always with Shadows enable, since this is how it should have been by default. So everything was working to me.

Anyway i can confirm that if you switch from enabled to segmented they dont work anymore. So keep it at enabled.

argh, you should have stated this some time ago, perhaps we had found the problem earlier. ;)

[Kzin]

mia_material color falloff for reflections dont work in "enabled" shadow mode that must be used for transparent shadows with ibl.
for this, use the "mib_dielectric" shader with its absorbtion color feature. it can be find in rendertree under mental ray-->material. use this node as input for additionel color slot in the mia mat.

[Nox]

Good day,

I'm testing this shader and have a problem with noise. There is no specular effects in scene. I've tried different materials (Arch, Mila, Mia Advanced, legacy) and hdri images, high sample values but ugly noise is still there.

My setup: create area light, turn on visible, select shape as user, connect user_ibl_env with hdri and set samples along with light samples. Am I missing something?

[Kzin]

Good day,

I'm testing this shader and have a problem with noise. There is no specular effects in scene. I've tried different materials (Arch, Mila, Mia Advanced, legacy) and hdri images, high sample values but ugly noise is still there.

My setup: create area light, turn on visible, select shape as user, connect user_ibl_env with hdri and set samples along with light samples. Am I missing something?

no spec, yep, thats the BIG problem. their is actually no consistence workflow with mr's ibl and shaders in general and spec/reflect workflow because nothing is connected because of the outdated mr integration.
you would need a native implementation of booth, mila and ibl (also for mia, but let us stop to talk about mia because its heavily outdated) and of course different visibility options that follows the new lpe paradigm.

[Nox]

Yes, I'm aware that MR is all about this. But I see nice images above and wonder what's wrong? I have that noise even on simple plane + sphere...

[Kzin]

Yes, I'm aware that MR is all about this. But I see nice images above and wonder what's wrong? I have that noise even on simple plane + sphere...

noisefree renderns will take a long time, its true for every renderer which is using brute force method.
the question is how much noise is acceptable for you and your client.

can you post an image and your ibl settings?

[Nox]

Sure, this is really basic stuff. Light comes from one user_ibl, 2 arch mats with 0,318 diffuse. I think settings are overkill for such scene, but as you can see it's extremely noisy anyway.

[Kzin]

try these unifed settings:
min2
max100
quality 3

[Fer]

User shaders are buggy. they don't play nice with unified sampling. specially environment shader, bump up the samples to 32 and try a unified setting of: min of 1, max of 32 and 2 or + for quality.

If it's not enough bump up the max value.

Don't forget light relative scale.

[Nox]

Thanks for tips, 100 samples on unified and light cleared most of the noise. But I don't think it should work like this, probably this shader is bugged.

[Kzin]

Thanks for tips, 100 samples on unified and light cleared most of the noise. But I don't think it should work like this, probably this shader is bugged.

its working correctly. you always need samples (local) to solve noise for user_ibl. or what are you expecting?
the builtin_ibl is working better with unified (and because it prebakes the env compared to user_ibl), but you also need more samples to solve the noise.

your unified samples were really low, to low to solve things that needs more rays to cleanup. and you cant replace these low samplevalues with a high quality value.

[Fer]

Here my test with 32 samples in the ibl env, and 1-50-2-0.02 in the unified sampling ones.

[Nox]

Nice,

so I've made a test with these settings and...



Oh snap