Bye Bye diffuse?

3D Studio Max Plugin (Export Script Plugins developed by [gk] and KilaD; Integrated Plugin developed by Karba)
Forum rules
Please post only in English in this subforum. For alternate language discussion please go here http://render.otoy.com/forum/viewforum.php?f=18
Post Reply
User avatar
gabrielefx
Licensed Customer
Posts: 1701
Joined: Wed Sep 28, 2011 2:00 pm

Hi all,

Someone probably already know that with the new glossy node we can create "diffuse" materials setting the specular to 0, roughness 1, index 1.6 (very similar to the Vray default settings)

The curios thing is that the new fake diffuse-glossy mat produce faster renders.

Also I tried to delete all the diffuse materials from my test scene, for light meshes too.

Using a specular mat with a scattering medium we can simulate a diffuse emitter.

Unfortunately the specular emitter produces noisy render.

I suggest to limit the use of diffuse material.

Regards
Attachments
diffuse.jpg
glossy.jpg
diffuse_diffuse.jpg
diffuse_glossy.jpg
specular_glossy.jpg
quad Titan Kepler 6GB + quad Titan X Pascal 12GB + quad GTX1080 8GB + dual GTX1080Ti 11GB
User avatar
t_3
Posts: 2871
Joined: Tue Jul 05, 2011 5:37 pm

as far i see from the images, there is a floattexture node connected to the transmission pin of the diffuse mat. even if set to zero, this already slows down performance, compared to the same diffuse mat with an empty pin there...
The obvious is that which is never seen until someone expresses it simply

1x i7 2600K @5.0 (Asrock Z77), 16GB, 2x Asus GTX Titan 6GB @1200/3100/6200
2x i7 2600K @4.5 (P8Z68 -V P), 12GB, 1x EVGA GTX 580 3GB @0900/2200/4400
User avatar
gabrielefx
Licensed Customer
Posts: 1701
Joined: Wed Sep 28, 2011 2:00 pm

t_3 wrote:as far i see from the images, there is a floattexture node connected to the transmission pin of the diffuse mat. even if set to zero, this already slows down performance, compared to the same diffuse mat with an empty pin there...
the diffuse emitter is for the light only, the floor and the lamp have a diffuse material without any emitter in that example.
quad Titan Kepler 6GB + quad Titan X Pascal 12GB + quad GTX1080 8GB + dual GTX1080Ti 11GB
User avatar
t_3
Posts: 2871
Joined: Tue Jul 05, 2011 5:37 pm

gabrielefx wrote:
t_3 wrote:as far i see from the images, there is a floattexture node connected to the transmission pin of the diffuse mat. even if set to zero, this already slows down performance, compared to the same diffuse mat with an empty pin there...
the diffuse emitter is for the light only, the floor and the lamp have a diffuse material without any emitter in that example.
_transmission_ not emitter ;)
the diffuse node setup says "transmission 0.0" so there must be a floattexture node? this slows down octane for every single diffuse mat in the scene. at least the last time it tested how octane deals with that node, and why it is empty by default...
The obvious is that which is never seen until someone expresses it simply

1x i7 2600K @5.0 (Asrock Z77), 16GB, 2x Asus GTX Titan 6GB @1200/3100/6200
2x i7 2600K @4.5 (P8Z68 -V P), 12GB, 1x EVGA GTX 580 3GB @0900/2200/4400
User avatar
gabrielefx
Licensed Customer
Posts: 1701
Joined: Wed Sep 28, 2011 2:00 pm

t_3 wrote:
gabrielefx wrote:
t_3 wrote:as far i see from the images, there is a floattexture node connected to the transmission pin of the diffuse mat. even if set to zero, this already slows down performance, compared to the same diffuse mat with an empty pin there...
the diffuse emitter is for the light only, the floor and the lamp have a diffuse material without any emitter in that example.
_transmission_ not emitter ;)
the diffuse node setup says "transmission 0.0" so there must be a floattexture node? this slows down octane for every single diffuse mat in the scene. at least the last time it tested how octane deals with that node, and why it is empty by default...
are you sure? In 3ds Max the diffuse material is a single node.

The transmission is 0 but there is no second node attached to it...

I'm missing something?
quad Titan Kepler 6GB + quad Titan X Pascal 12GB + quad GTX1080 8GB + dual GTX1080Ti 11GB
User avatar
t_3
Posts: 2871
Joined: Tue Jul 05, 2011 5:37 pm

hm; i don't know how the max/maya plugs handle this, but i tested it again in the standlone: some object, rendering at ~16.8ms/sec (pmc) with a single diffuse mat and with no node attached to the transmission pin (=default). then created a floattexture set to zero, connected it to the transmission pin and speed went down to ~16ms/s - not that much difference you encountered between "fake" diffuse and diffuse, but still a some ~5% ...
The obvious is that which is never seen until someone expresses it simply

1x i7 2600K @5.0 (Asrock Z77), 16GB, 2x Asus GTX Titan 6GB @1200/3100/6200
2x i7 2600K @4.5 (P8Z68 -V P), 12GB, 1x EVGA GTX 580 3GB @0900/2200/4400
User avatar
roeland
OctaneRender Team
Posts: 1823
Joined: Wed Mar 09, 2011 10:09 pm

We should solve this issue by making a diffuse material with transmission set to 0.0 more efficient. I find for most objects a diffuse material without transmission is faster than a glossy material.

Since the fake diffuse material has a slightly different reflection pattern, the amount of reflected rays hitting the object again may change slightly. More rays hitting the object again will slow down the renderer, so the performance increase or decrease depends on the scene.

The speedup when using the fake emitter occurs because Octane is not tracing any shadow rays anymore. This means less work per sample and thus more megasamples per second, but this will usually lead to more noise after the same time.

--
Roeland
paoloverona
Licensed Customer
Posts: 435
Joined: Thu Mar 29, 2012 8:40 pm

Hi Gabriele,

otoy should pay you for all the useful tricks that you share with us for improve the quality or the speed of our renders, really thanks! :D

did you try some interior shot?, seems that the faked diffuse material gives a darker color in ninteriors.

I tested 3 interiors, the speed encreased from 10% up to 60% :shock:

...more tests need to be done ;)

Cheers, Paolo
intel i7 3820 3.6GHz, 16Gb 1600Hz, windows 7 professional 64bit, gtx 580 3Gb x2, Octane 3dsMax 2.58
User avatar
roeland
OctaneRender Team
Posts: 1823
Joined: Wed Mar 09, 2011 10:09 pm

You can't compare speed if the brightness doesn't stay the same. It is normal for the rendering to speed up if the materials get darker, you can reproduce this by just making a diffuse material less bright.

--
Roeland
User avatar
gabrielefx
Licensed Customer
Posts: 1701
Joined: Wed Sep 28, 2011 2:00 pm

Thank you, Otoy already pay me with this awesome software.

Regarding the diffuse vs. glossy the winner (for me) is the fake glossy diffuse. Obviously in interior white spaces the light is clamped using many glossy mats, then we can balance it with exposure.
With glossy white mats we get less noise. This is not valid if we try to simulate metals.
I think that Octane should use different methods of sampling.
I remember that Roeland was working on this feature.

Regards
quad Titan Kepler 6GB + quad Titan X Pascal 12GB + quad GTX1080 8GB + dual GTX1080Ti 11GB
Post Reply

Return to “Autodesk 3Ds Max”