PDA

View Full Version : Questions for Octane users, about surfacing and lighting



S0nny
10-20-2016, 09:55 AM
Hi guys, I'm taking some hours to test octane, I have the demo here since quite some time, but never tested in depth.
I have some questions, not in a particular order:

Lighting

- is it possible to match Octane light properties with custom shaped poly-lights using the texture emitter? They both have power, efficiency and color but the Octane light has also the range/nominal distance from Lw lights panel.
- any way to tell octane to not take into account the range/nominal distance of the light? It's a feature that doesn't make sense.
- also, does the emitter take into account the Inv dist sqr falloff?
- the emitter is faster or slower than Octane lights? There's no spherical light in the built in shapes, strange.

Can't find anything in the manual for this.

About the shaders

It seems to me that the Glossy shader behaves really wrong. It's probably me but:

- if I try to build a simple colored metal using the built-in IOR I get something super fake because the shaders takes into account the diffuse values even when the specular is at 100%.

Aren't this shaders supposed to be energy conserving or what are they? Because it also looks totally wrong even if I put 0 as diffuse, it's like the specular is multiplied over the diffuse or something.

- also, the roughness power is waaaay too much aggressive, like in Lw. Is there a way to use something like the POW node to remodulate the effect of the roughness? It behaves very badly with textures.

- one more on roughness: the metal material produced with the glossy shader at 100% roughness is still something very metallic, which is obviously wrong.
If octane have the right nodes I can manage that automatically (didn't check everything yet). I also noticed that if I put a IOR of 1.0, which in my experience means no fresnel, which means just the diffuse component, the shader produce the fresnel effects based on the specular color.

I think it's all for now, need further testing for the rest.

Sorry to post much questions here, but can't post on octane forum if you are not a licensed user... I love otoy already.

10-20-2016, 11:03 AM
Can't answer them all but I'll stab at a few.
No, I don't think emitters have control as you suggest. You should ask for a freq from juanjon on that. Or perhaps he might explain why it isn't in there already.

Make a ball. Make it an emitter. There goes a real spherical light. The built-in disk of a light comes close, no? What more are you needing?

My experiments suggest to start at the low end of the numbers in order to get something usable: Octane's defaults are quite high for many an effect.
I would strongly suggest piping scalars into anything you want to control as it is more precise and clearer in use. That is to say, Yes to the POW node remodulating effects.
The specular node is ... complicated. Ha, not really but it has one or two catches you have to keep up with. This note will demonstrate that I am not up to speed at the moment and Octane is at home such that I can't check.
Specular or diffuse works in a complementary way. I think it is specular: the color yellow will get you blue reflections and highlights.

Are you reading the OTOY forum? There is some valuable info there. As well, look at the manual; a lot of your questions are answered there.
Robert

S0nny
10-20-2016, 11:27 AM
Thanks Robert, I just saw that there's an open board for the demo users on otoy forum, I'll ask there.

The glossy node doesn't look so complicated actually, it just doesn't make any sense. To get an acceptable looking metal I have to use 1.0 IOR and drive the reflection with the color, but it's totally wrong using any IOR values for fresnel effect.

The problem with scalar nodes is that there you can't use the Lw ones for the most part, octane requires the float node as scalar. For example, can't use the POW node for remodulate the roughness.

I don't know, it doesn't seems to start very well, for now, I'll see what happens with more testing.

S0nny
10-20-2016, 03:16 PM
Well, the post is in moderation. As for now I solved the doubt with the light, I didn't noticed that nominal distance hasn't effect in the octane lights, and is the Lw intensity that just act as multiplier.

m.d.
10-21-2016, 10:44 PM
Hi guys, I'm taking some hours to test octane, I have the demo here since quite some time, but never tested in depth.
I have some questions, not in a particular order:

Lighting

- is it possible to match Octane light properties with custom shaped poly-lights using the texture emitter? They both have power, efficiency and color but the Octane light has also the range/nominal distance from Lw lights panel.
- any way to tell octane to not take into account the range/nominal distance of the light? It's a feature that doesn't make sense.
- also, does the emitter take into account the Inv dist sqr falloff?
- the emitter is faster or slower than Octane lights? There's no spherical light in the built in shapes, strange.


Octane lights will never have range/nominal distance, as they are physically based. Light in the real world is inverse squared. So all lights and emitters follow that rule. The range/distance/falloff have no effect on octane lights, they are just leftover in the shared light GUI.

They should be the same speed....as emitters governed by the sampling rate of the light rate, but a sphere is casting more rays in more directions than a cube or rectangle....so it may slow it down a bit because of the multiple ray normals.
But all things being equal they are sampled the same

3dworks
10-23-2016, 04:28 AM
computing IOR on surfaces which are not dielectric usually needs additional 'N' and 'K' factors to be taken into account, which the octane material system doesn't have, at least currently. changing just the IOR for non dielectric materials will most likely produce wrong results in octane. usually metal looks good with a factor of 1, but you are free to experiment, of course. maxwell for example always has had a much more realistic material system where you can set these factors to get a physically correct result. in octane the process involves visually tuning using material mixing, falloff maps, etc. i also wish one day we get a more physically based material system in octane maybe in next big version. from a practical point of view, it's however very easy to get the desired look very fast while twaeking - and results then look very constant under different lighting conditions. you also can tweak good looking materials from the material DB...

EDIT: as i found out now, there is a way to script shaders in the standalone, see this thread:

https://render.otoy.com/forum/viewtopic.php?f=73&t=43973&p=292582

i wonder if it is possible to use this also within the LW plugin? i will ask juanjo and keep you updated...

cheers

markus

S0nny
10-24-2016, 03:09 AM
Octane lights will never have range/nominal distance, as they are physically based. Light in the real world is inverse squared. So all lights and emitters follow that rule. The range/distance/falloff have no effect on octane lights, they are just leftover in the shared light GUI.

They should be the same speed....as emitters governed by the sampling rate of the light rate, but a sphere is casting more rays in more directions than a cube or rectangle....so it may slow it down a bit because of the multiple ray normals.
But all things being equal they are sampled the same

With the gpu I have here I can't do more complex tests, but it seems like the way it works.

S0nny
10-24-2016, 05:15 AM
computing IOR on surfaces which are not dielectric usually needs additional 'N' and 'K' factors to be taken into account, which the octane material system doesn't have, at least currently. changing just the IOR for non dielectric materials will most likely produce wrong results in octane. usually metal looks good with a factor of 1, but you are free to experiment, of course. maxwell for example always has had a much more realistic material system where you can set these factors to get a physically correct result. in octane the process involves visually tuning using material mixing, falloff maps, etc. i also wish one day we get a more physically based material system in octane maybe in next big version. from a practical point of view, it's however very easy to get the desired look very fast while twaeking - and results then look very constant under different lighting conditions. you also can tweak good looking materials from the material DB...


This is a very big disappointment, state of the art physically correct render sounds like a pretty strong statement without physically correct shaders honestly, expecially at its 3rd major release.

Probably it can be tweaked, but i'm still not sure looking at the nodes available for octane if it'll be so easy. I'll try for sure. Still I don't get why the developers think that it's a good idea to build a shader like this.
The glossy shader drives the fresnel using IOR 1.0 + spec values in a different way that using a high IOR value. It takes into account the diffuse, even at zero values: it doesn't matter much if there are not extinction coefficients taken into account, it's just wrong.
Also, why bother to add a coating layer if it's not a real coating: I mean, the thin film has its own fresnel, but not its own specular or roughness? It doesn't make sense.



EDIT: as i found out now, there is a way to script shaders in the standalone, see this thread:

https://render.otoy.com/forum/viewtopic.php?f=73&t=43973&p=292582

i wonder if it is possible to use this also within the LW plugin? i will ask juanjo and keep you updated...


Nice, maybe he knows something about the octane developement to fix this problems also. I still can't see the post I did three days ago in otoy forum, I guess I'll never see it.

3dworks
10-24-2016, 06:35 AM
if you need to simulate materials in a very accurate way - for example for jewellery, optical experiments etc. the only solution on the market is maxwell, imo. for general archviz and probably 99% of industrial design viz the current standard octane materials render realistically enough, you won't notice those details easily.

also, juanjo told me that the use of scripted material nodes as locally stored LiveDB node should work, this permits accurate simulations even in octane, but you have to setup the parameters in the main octane application and then save the node to local LiveDB. i've tried this, but (at least on mac) there seems to be an issue currently, which makes layout crash if handling that node. juanjo is informed and i guess he will get this resolved soon. also, it is likely that future versions of the plugin will incorporate this directly as a node, if there are enough requests for this!

cheers

markus

S0nny
10-24-2016, 10:03 AM
Well, I think it's not much about the accuracy here, it's just that is a step backwards in the way I'm used to setup materials. I'm looking for consistency and I need a shader that is predictable in how a fresnel works, or a micro-faced roughness works, to say just two very simple rules actually. With this Glossy shader the roughness will not look right as others with micro-faced models, for example. There's no GGX brdf model also, which is also very difficult to replicate via nodes. The coating also is not lacking of accuracy, it's just lacks of its basic functionality. Even the carpaint node in Lw is more advanced.
Don't know, maybe I have too much expectations, but this is stuff that other renders can do pretty much in one click.

For the script, I can't test because the LiveDB is disabled for the demo.

tcoursey
10-25-2016, 03:02 PM
I can't speak to the scientific or real world properties of light and such but I can say that after getting familiar with Octane you'll find it easy to get some great results.

Keep in mind most everything in Octane is based on real world, so as mentioned inv. sq is not an option, it's the rule. So there is no turning it on or off. This goes for many things throughout the system.

You will find the falloff node something to look at in regards to the glossy node. This will help make metal like materials have the frensel affect you may want.

I too find it hard to pick a color from the picker for a glossy material and get it to look right when lit, but I assume that what is going on in the render process is real world. So having IPR up when fiddling around with colors helps. Typically you will have the Saturation of a color most of the way up to get some nice rich colors.

I find myself using actual lights rather than emmitters as it helps me OPTIMIZE render times with SAMPLES on different lights in different places, rather than all CAN LIGHTS or whatever having the same SAMPLEING. What I mean is when you have a large or multi room scene the lights that are in the other room (surface emmitters or actual lights) will be sampled and cost you render time. So when I save out multiple scenes for a space, if the light that isn't contributing to the scene at that moment isn't visible turn it's sampleing rate to 0. You can also fiddle with having different lights have different sample rates.

Using IES on actual lights can help the look quite a bit too. Again, these octane lights will have the IES properties, so you can get a SPEHERICAL like IES and put it on a light.

Hope that rambling helps a bit. It took me a bit to really get going with Octane, making some presets etc...but I really, really enjoy having the two Titan's pulled up in an IPR and getting artistic with my scenes now!

Enjoy!