[Half Life]

Hi Ernesto, I took some time to try to make a new material that fully meets your needs and I found I was having some weird issues like you are having -- after working the material over changing just about every configuration trying to find the problem I got the result expected (no refraction/reflection).

I'm not exactly sure what happened but the breakthrough came after I started working with the texture properties -- most notably I switched out the layer "Mask" image for the HDR (formerly used by the emitter) which immediately gave black bar corruption (unless I enabled "Interpolation") -- even then the HDR did not provide a very good mask... I think there is definitely a bug in there somewhere.

Anyway, after I brought back in the regular layer "mask" image everything worked fine and I was no longer able to break the material (make reflections/refraction appear again). Here is the final material which should meet all of your requirements -- just change out your layer mask image for mine, and change the color of the emitter to suit your needs.

https://dl.dropbox.com/u/41250852/new_emitter.zip

Best,
Jason.

[Ernesto]

Oh! Thanks for your help, Jason!
I will try it tonight, and will post the feedback!
Thanks again
E

[tom]

OK, received the scene and I must say I still don't understand why are you choosing the worst option.
Here's a comparison of dielectric emitter vs ghost emitter.

Your version SL8


My version SL8


I don't see a reason not rendering this scene as in my fix without pain and noise.
And if the only reason is having a masked object id pass of those streaks, why not running a quick second pass for it.
Eventhough, you should also tell me what would you really do with such an output.

[Ernesto]

Tom,

That is a great diference! I knew you were going to find something!
You are right, I do not know either?!?
I am sorry I have no reason for having choosen the worst option. It is probably that I do not know how to do it yet...
I really do not want pain nor noise at all! I promise!!
Where could I find the ghost option?

Tom wrote: The difference comes from making it out of ghost or dielectric. If T<>255 and Nd<>1 it will not be invisible in Material and Object Id channels. So at worst case, you need to hide them and make a separate Id pass, easily.

Tom Wrote: Because, what you call "Hidden from secondary rays" is actually "Hidden from Reflections/Refractions" as clearly denoted in Studio UI. In the failing material the base BSDF is ghost (T=255 and Nd=1) but, in the suggested material I intentionally use a high attenuation dielectric (T=254, Atten=high and Nd=1). Doing so makes hiding something from reflections/refractions available. In short, seeing something through ghost is not a refraction because ghost is a special condition for the engine. Ghost is hidden from "hidden from"

Could you explain me this in more detail?

As far as I know the sparks material was created as follows (T=254, Atten=high and Nd=1) unless I did something wrong...

To answer your output question, I am planning an animation at the end of the project.
It will not be that big, so I guess I could do diferent composites if necesary.
The ship will "suffer" several deformations, (sculpt and twist deformers) to simulate a "spatial distortion", and I am planning to use certain motion blur. The lighting of the ship will interact with the enviroment which will be prety dark.

E

[Ernesto]

Jason, Your material was great! Thanks!
It also could work in Multilight set as Intensity and Colour!

That configuration made me think that perhaps we could do a plain emiter material that could be full colour and transparent not using HDRImages nor MXI which are causing the interference in the multilight.
Here you have my first attempt:



I separated the colorfull map (upper right corner) into 3 RGB channels.
Each of them was used as the opacity map for a layer where there is a plain emiter of that colour (RG and B)
All the layers were set to aditive blend, and it worked!

At first I set each of the emiters the same intensity but there was something wrong.
Finnally decided to increase the GREEN channel 4 times with respect to the other two channels, and it seemed to be like the original image.
I still do not understand why this special diference with the green channel... (still thinking)

Just for curiosity I tried the Multilight Colour settings and although this material do not offer the option (it was expected since there are 3 emiters with 3 colours) it do not interfere with the multilight at all! It is a rather complex construction but we still have a fully transparent not reflective material, so it should work perfectly.
I will see in the further tests.

Ernesto

[Ernesto]

I am writing the concepts in these Notes on Ghost and Dielectric Materials:
(please correct me in case I am wrong)

Dielectric would be: T<>255 and Nd<>1 (in the BSDF component that is in the same layer as the emiter)
If dielectric we will have all kind of refractions reflections etc

Ghost would be: T=254, Atten=high and Nd=1 (in the BSDF component that is in the same layer as the emiter)
If ghost we will have a really transparent material.

I have rendered it again up to SL 8 as you (Tom) did, and in fact the result is as you showed me as in your upper sample.
Evidently I am having something wrong although I thought that I was following your directions.
So there is a missunderstanding on my part. (I wonder what is it)

E

[tom]

I have rendered it again up to SL 8 as you (Tom) did, and in fact the result is as you showed me as in your upper sample.
Evidently I am having something wrong although I thought that I was following your directions.
So there is a missunderstanding on my part. (I wonder what is it)

Sending the MXS to your email now.

Once again;

Ghost is T=255 and Nd=1 *** No reflection/refraction
Dielectric is 0<T<255 and (any Nd) *** In this case, Maxwell computes refraction even if Nd=1. This is why it converges noisier.

The fastest option is making the emitter's base BSDF out of ghost. So why did I suggest using a dielectric base in the previous posts. Because, you wanted to hide the object from reflections/refractions and it's not possible doing this with ghost. But your current setup doesn't require hiding it from reflections/refractions. Additionally, now I see you even hide these sparks from Global Illumination so you don't even need their emission. They are just there like overlaid in the post. Any other problems left?

[Half Life]

I have a question Tom, how is refraction even possible at Nd of 1 -- isn't that exactly the same as vacuum which is = to Maxwell "air"?

If so, there should be nothing to refract since the light rays would not bend at all... it would be as if the object was not there (ghost as you call it) -- or is there some additional subtlety of physics at work here? The only thing I could think is that is somehow related to the negative Nd (metamaterial) possibilities of V2.x...

I could see the lower transmittance color having a logical effect on attenuation, but not refraction/reflection.

Best,
Jason.

[tom]

is there some additional subtlety of physics at work here?

It's the additional subtlety of coding. As a very simple example, the results of expressions A = B and A = B * 1.00 is the same, while one of them is slower than the other.

[Half Life]

So if I understand correctly, the end result should be identical, but at <255 Maxwell actually tries to be physically correct and dutifully does the full calculations (even though the answer is still zero refraction) -- whereas at 255 we've essentially told Maxwell we already know the answer and don't bother checking for us.

If that's true, can I ask what advantage is there for enabling those calculations at any transmittance value (once the Nd is set to 1) -- other than keeping the Unbiased purity?

The reason I ask is this could potentially be made to be useful at lower transmittance levels with SSS -- after all in a SSS scenario we don't always need refraction calculations (although I suppose that is treading into Biased territory). As it stands now SSS does not work at Nd of 1 and value of 255 -- but if it could work this might make a very fast solution for volumetric atmosphere effects.

Best,
Jason.

[tom]

Bingo Jason



..

[Ernesto]

Sorry for the delay, but this is so tricky that it took me some time to find the answer to the previous question:

"OK, received the scene and I must say I still don't understand why are you choosing the worst option.
Here's a comparison of dielectric emitter vs ghost emitter." http://www.maxwellrender.com/forum/view ... 26#p363549

What happened is that the material was set that way because we had to get rid of other plain transparent emiters getting visible when seen through the volume with the mentioned spark material.

This is the complex scene:


The following is the simplified scene that was first published at the begining of this topic, http://www.maxwellrender.com/forum/view ... 97&t=39792 together with the detailed explanation of the issue:

This last frame is representing the same situation as shown in the complex scene.

Ernesto

[tom]

Ernesto, you didn't send me those ones in the MXS so I did not see your need in doing this. It's OK you can prefer the dielectric base for the upper part but, why do you do it for the bottom part, then?

[Ernesto]

Yes, Tom, it was my fault that in an attempt to simplify the scene, I omited unvoluntarily the other emiters that were causing the problem.

Even in the lower part, when rendered from certain viewpoints, some issues can be seen.
The good news is that now I undertand that I have to choose between having certain noise, or having the transparent emiters issue.
I personally prefer the first option, in case the material should be created based in a HDRImage or MXImage.

The other option is to use plain emiters, with opacity masks, as Jason suggested here: http://www.maxwellrender.com/forum/view ... 55#p363541 (you can download a sample of his material)
The bad thing about these method is that all the emiter masked surface shows the same light intensity, giving a plain look, when compared to the high richness of am HDRImage.
But perhaps it is possible to reproduce the same effect using the same material construction, as this:
http://www.maxwellrender.com/forum/view ... 55#p363558
Only that instead of using RGB colour channels, we could use diferent emiter intensities for each channel, reproducing in theory the same behaviour as an HDRImage, that could be sensitive to adjustments in the Multilight feature.

Ernesto

[Ernesto]

The test of the new material were succesfull.
It was possible to create a materiual which behaviour would be similar to an HDRI based material, without using HDRImage map.
The advantage of such materials are:
In Maxwell V2.7 they do NOT cause the Multilight Colour Malfunction, and they are a Zero noise materials.
The disadvantage is:
That they produce the same problems with the suposedly hidden lights as described at the begining of this post.

Now I have no other choise than working with a heavy noise material.
I really wish you could fix this in a 2.x version.

Ernesto