All posts relating to Maxwell Render 1.x
User avatar
By Ernesto
#362452
In order to solve the shadows issue, I tried lots of things.

Image

1
Instead of using a color map, with white background in the transmitance box, (in the 3rd figure) I used a black and white map.
I was surprised that the sparks rendered blue anyway, despite that map was black and white...
This means that the lights colours that we see in the render, are from the mxi file, and the colours in the colour map in the transmitance box are the colour of the shadows.

2
So the next step, I thought that if the colour map in the transmitance box, would be white, the shadows would dissapear, and I was right! but the lighting sparks dissapeared too...

3
Then I thought that the shadows could be due to a diference between all the masks, so I used the same file for all the mask, inverting it when necesary, but the shadows were still there. The masks are in the 2nd and 4th figure.

4
The next time I looked carefully at the shadows and they were in the parts of the masks that were pure white or pure black, so I edited all the masks so that they would be grey, without black nor white, but the shadows were still there.

5
The next reasoning was that perhaps the WHITE parts in the MXI file were causing it, so I edited the MXI file so that there would be NO WHITE in it. But again I failed, the shadows were still there.

I am sure that someone here knows the LOGIC of these materials.... At least the guy who created this system should know!
All I need is an explanation on which is the logic, in order to be free to create the desired effect, as in this case.
I must say I have been asking this for years... without success up to now
Perhaps the guys who are programing Maxwell are too bussy to answer one question from a single person, so I will make it more profitable for the Maxwell team. If the programmer guy who must know this could explain it to me, I can write a tutorial or even a manual for everybody so that Maxwell Materials could end to be a mystery, and became a tool, instead of being an obstacle in certain cases as the present. Remember that I have been strugling with this problem for more than two month now. If this would be a commercial project it would be not possible since nobody would accept an undefined deadline. On the other hand if I work on it forever, I would lose money...

I think it is a good offer!
I hope you (the programmer guy) could think the same, and we both (and everybody else) could benefit from this "effort" I am asking you!

Sincerely

Ernesto
User avatar
By Ernesto
#362455
Although I still wonder about the logic of Maxwell materials, I am still trying any possibility hoping that by chance I might find it.
I am very fortunate, because I did found a temporary solution although I stay totally ignorant about the reasons.
Here is what I did:

First I realized that the colour map, (in the 3rd window) defined the colour of the shadows I want to remove.
The first Idea was to use a white map, I thought that a white map would show no shadow and i was right but the consequence was that there was no emiter, when using a white map.
To make things simplier I wondered why a map was necesary, and found that it was NOT necesary.
I removed the colour map, and assigned a transmitance colour, and verified that the shadows rendered the same colour as the transmitance colour.
Then I tried a white colour, and again the emiter dissapearaed in the render.
I do not know why I tried a grey colour, and then a lighter gray colour and I saw that the shadows got more and more transparent.
Then I decided to use a colour that would be CLOSE to white but NOT WHITE. So I used a H=0 S=0 V=254 where 255 is white and 254 is ALMOST WHITE. BiNGO!!! it seems to work. The emiter is still there and the shadow seems unnoticeable up to now.

The Million Dollars question is still the same:
Which is the explanation?
Why?

So I still wait for the specialist to enlight my knowledge about Maxwell Materials.

Ernesto
By zdeno
#362457
it is some cruel funny. Just look at You strugling ;) I see myself about 2-3 years ago ;)

first of all verision 1.7 is not supported anymore and studying 1.7 material system is useless. In 2.x versions clipmasking is much more easier.

so You have to tell me if You are going to update to 2.x ? or stay in 1.7 world?
User avatar
By Ernesto
#362470
I see your point Zdeno.
Let me "translate" what you said:

1) You seem to say that It is not me that cannot understand the logic of 1.7 Maxwell materials, but the problem could be the lack of logic of 1.7 Maxwell Materials...
2) In such a case you seem to say that upgrading would solve the problem.

To tell you the true when I started this post, I had the hope, it was me that was not able to understand a perhaps tricky logic, so my way was to keep working hard to overcome the obstacle.

But now I wonder if 2.x Maxwell materials has any logic either, before upgrading. Is there any documentation or manual or tutorial that could explain 2.x Maxwell materials predictable behaviour? I really wouldn´t want to go through a process of upgradding to change an old group of problems for a new group of problems. Do you understand me? In other words I was disappointed by the final product Maxwell Render 1.7, and Maxwell do not seem to ensure that the problem was solved and that I wouldn´t be dissapointed again. To tell you more I haven´t found any official indication accepting that problem, nor any notification telling that it was solved, so that there is no hint that I could be sure I would not be disappointed by the finnal product Maxwell Render 2.x again.
The distrust is increasing if I assume the problem is not mine, because then I cannot understand why there is nobody else accepting the responsibility?

Anyway, there is a Spanish saying:
"Si uno se quema con leche, cuando ve una vaca llora!"
that would translate more or less like this:
"If you were burn with milk, when you see a cow, you cries!"

So I would like to be sure the milk is not burning this time
Yours

Ernesto
Last edited by Ernesto on Wed Nov 14, 2012 7:14 pm, edited 1 time in total.
User avatar
By tom
#362471
Ernesto, I've offered the solution in Aug 2006.
Please follow the link: http://resources.maxwellrender.com/sear ... de=1&id=50

The idea is very simple:
As a rule, every material has a base BSDF and this means emitters also do have it. For this reason, your emitter casts a shadow. So, you only need to modify emitter's BSDF to make sure it lets all the incoming rays pass through. It means setting Trans=(255, 255, 255) and Nd=1 which is called Ghost BSDF. An emitter having such a base BSDF will not cast a shadow in the final render and it will get completely invisible when faded out using Multilight. The example mxm is a complicated version of the same idea which is working with color transmittance and color emission but it's not necessary.
User avatar
By Ernesto
#362473
Thanks Tom,

I followed your link, and found a fire material in the database.
I hope i have understood your explanation ok. (I am doubting about myself at this stage :-))
I tried to see what you said, but as far as I can see the fire material in the link,

Image

is prety similar to the others I tried before:
In the first BSDF the colour is set as you said Trans=(255, 255, 255), but it is not relevant because there is a colour map checked.
If I remove the colour map, and use a Trans=(255, 255, 255), the emiter would not work anymore, as I explained in previous posts.
The partial solution I found is to use a Trans=(254, 254, 254) so that the shadows would reduce to a minimun, and the emiter would still work, but I know this is not the right solution, because the shadows are still there!

Yours

Ernesto
Last edited by Ernesto on Wed Nov 14, 2012 7:42 pm, edited 1 time in total.
User avatar
By tom
#362474
Ernesto wrote:If I remove the colour map, and use a Trans=(255, 255, 255), the emiter would not work anymore, as I explained in previous posts.
It really works here. So, send me your mxm and let me fix it for you.
User avatar
By Ernesto
#362475
tom wrote:
Ernesto wrote:If I remove the colour map, and use a Trans=(255, 255, 255), the emiter would not work anymore, as I explained in previous posts.
It really works here. So, send me your mxm and let me fix it for you.
Thank you so much Tom!

I will send it tonight, because I have no access to my hardrive from my work.
Thanks again!

Ernesto
User avatar
By tom
#362477
Here's an example:
Image

Maybe your problem is not the material and it's the fact "Emitters only emit in normal direction". Because, my above example has duplicate (normals inverted) emitter geometry for illuminating also in the opposite direction. :)
User avatar
By tom
#362484
It's better you send me a simple scene and I fix it in your own example. So, you can understand how it works.
By zdeno
#362489
tom wrote: Maybe your problem is not the material and it's the fact "Emitters only emit in normal direction". Because, my above example has duplicate (normals inverted) emitter geometry for illuminating also in the opposite direction. :)
all true ...
clone emiter objet and use Flip command to change normals direction. in 1.7 it helps that with multilight all meshes with the same emiter are treated as one.

sometimes with more fancy bsdf it is good idea to move new clone a bit (1mm) to avoid overlaping artefacts

and if You want to work with maxwell in future try to upgrade ... it looks like there is a proper promotion for upgrades 50% cheaper and 5 rendernodes free. If I good recall last email.
User avatar
By Ernesto
#362492
Dear Tom,

I have sent the files already in mb and mxs formats and all maps.
I am at home now, and tried again to correct the material assigning a Trans= 255,255,255, and it dissapears the emiter even in the material preview. That is the reason for a Trans= 254,254,254
Another thing you should know is that the poligonal disc, where the mentioned material was set as follows:

Maxwell Settings:
Hidden From Camera= Unchecked
Hidden to secondary rays= Checked
Hidden to Global Ilumination = Unchecked
Unafected by Clipping plane= Checked

At the begining I experienced very bad results, so Unchecking "Hidden to Global Illumination" the material seemed to improve. This was a sugestion by Zdeno.
The poligonal disc, is a thin object with only one face (normals down)
Before sending this message, I tried to replace the poligonal disc, with a very flat cylinder, so that the object would have double face, and normals would be up and down.
The result of this experiment was not Ok. I saw that if the sky would be seen through the cylinder, it would look like an emiter even if dimmered off in multilights.
It is evident I am doing something wrong although cannot see what is it!
I can see it by watching your animated gif sample.
Thanks again for your help!
Yours

Ernesto
User avatar
By Ernesto
#362493
zdeno wrote:
tom wrote: Maybe your problem is not the material and it's the fact "Emitters only emit in normal direction". Because, my above example has duplicate (normals inverted) emitter geometry for illuminating also in the opposite direction. :)
all true ...
clone emiter objet and use Flip command to change normals direction. in 1.7 it helps that with multilight all meshes with the same emiter are treated as one.

sometimes with more fancy bsdf it is good idea to move new clone a bit (1mm) to avoid overlaping artefacts

and if You want to work with maxwell in future try to upgrade ... it looks like there is a proper promotion for upgrades 50% cheaper and 5 rendernodes free. If I good recall last email.

Humm ok! I will try this now, and will post the results as soon as possible.

Half an hour later: I did it, and rendered, but I couldn´t see any diference.
Perhaps I am too tired now...
Or perhaps I am having more than one error, that are combining.

Anyway Thanks Zdeno, I think the best I can do now, is waiting Tom´s diagnostic.
Image
His sample was perfect, we will see which is the way to achieve it.

Yours

Ernesto
User avatar
By tom
#362500
Ernesto, I edited your scene and sent it back to you. It was quite nostalgic I must confess. :D As you've said, the ghost trick didn't work because, I've really forgotten the high-end tricks on this old system so long ago. But, I must make it clear the ghost recipe is working nicely with 2.x series. Anyway... So, the solution was in fact using a color transmittance map @ Nd=1 and Roughness=0. No reflectance maps, weightmaps etc... Also, you should uncheck all the visibility flags. You really don't need them for the purpose. And finally duplicated the geometry and flipped the normals to illuminate the other side. That's all... :)

However, you may fail if you attempt to try creating it from scratch because, I didn't tell something very important, yet. The trasnmittance texture has pure black pixels which is not friendly with this trick because, those pixels will block the light on some parts of the texture. In order to avoid it, use RGB Clamp under texture picker. See, it's now 1 to 255 instead of 0 to 255. ;)

And finally do something good for yourself, really... Upgrade to the final version (currently 2.7.20) and cut this pain. We've gone so far since 1.7.1 now I realized once again.
Will there be a Maxwell Render 6 ?

Let's be realistic. What's left of NL is only milk[…]