[zdeno]

Hi all maxwellians !

I am still strugling with mxm system

I found something like this on web :

Reflective Values

pure aluminum /polished/ 80 - 87 %
polished aluminum 65 - 75 %
matte aluminum 55 - 75 %
aluminum painting 55 - 65 %
steel 25 - 30 %
INOX 80-90%
polished chrome 60 - 70 %
high polished copper 60 - 70 %
high polished brass 70 - 75 %

light oak /polished/ 25 - 35 %
dark oak /polished/ 10 - 15 %
wood chipboard 25 - 40%
white paper 70 - 80 %

granite 20 - 25 %
lime stone 35 - 55 %
polished marble 30 - 70 % (depending on color)
light stucco 40 - 45 %
dark stucco (rough) 15 - 25 %
concrete /rough/ 20 - 30 %
bricks red/new 10 - 15 %

glass 5 - 10 %
silver mirror 80 - 88 %
high polished mirror 92 - 95 %
white tiles 75 - 80 %
white enamel 65 - 75 %
white lacquer 80 - 85 %

so ... If I understood this correct.
When I want to make concrete material , my texture should be "clamped" "cliped" to RGB 77,77,77 (about 30%) ?

or should I use all RGB spectrum up to 242,242,242 and do some trick with Nd ? (but still it has a small factor when roughness is high

or should I use all RGB spectrum up to 242,242,242 and use another layer to cut reflected energy in some way ?

[zdeno]

now everything is clear !
http://citeseerx.ist.psu.edu/viewdoc/do ... 1&type=pdf

[zdeno]

Tue Nov 27, 2007 9:23 am

Maxwell applies 2.2 inverse gamma on input and corrects it back with 2.2, so the image you view in Maxwell viewer is correct. It looks like your image viewer on the right is applying a preview gamma to the images without gamma information.

so ... If I want to get 18% "kodak" reflectance ... should I use 118,118,118 RGB ? or bump it high to "compensate" gamma conversion?

are You there My MXM Master ?

RBG 0,0,0 means no reflected light RGB 255,255,255 means 100% reflected light. so is it linear ? do RGB 64,64,64 mean 25% reflection? and 192,192,192 - 75% ?

[tom]

Equivalent RGB = ((reflectance/255)^(1/gamma))*255
Therefore, you should set RGB 186 for 50% reflectance.

[zdeno]

thx Tom ! I just ran few test and made a little pratice with curves in photoshop to avoid this scary mathematical expresions.


Could anyone share a tutorial to squize HDR in photoshop to get sky back in 8bit format? I ran few approach but only funny colors and megadark corrners shows;)

and Bigger one

[zdeno]

and another quest

I fight with rooftiles last half year. as in test simball enviroment it looks delicious in hard light it looks like c..p.
yellow tint from sun and bluisch from sky, turn this in green/barbie pink monster.

I tried almost everything . 1BSDF 2BSDF 3 BSDF . BSDF + coatting, additive mode , normal mode. roughness from 10 to 90 , , changed sky and sun conditions to soften shadows.
nothing ... just have no patience.

still i get flat , contrastless pixels.

after limitless test i back to roots , and what looks , almost what I want.

normal mode
1 BSDF -(100%)- REFL 0 and 90 - HSV - 24,215,190 roguhness- about 85-90
2 BSDF -(30%) - REFL 0 about (24-190) RGB REFL90 - (128-255) RGB tested - roughness about 35% will add glossy not polished effect

I add some picturess





I tested a few materials from mxmsite, to sneak masters, but as good in simball as bad in sun .

So . ? anyone have solution?

[Bubbaloo]

I think if you want this roofing to look great, you'll need to put a lot of work into variations in the tiles. Also, flat colors never look as good as real photo referenced textures with specular and bump maps.

[zdeno]

You have right Bubbaloo but I work with flat colors to finally squezz out how this material system works .
If I can do convincing materials with flat colors, with textures I would rule the world

I found what is the hardest thing to do GOOD materials , not this cr..py scanline look from wizard.
It is hard to find balance beetwen roughness, reflect colors, weight layers, and Nd.
Mayby it is only me who have issue with judging what this all thing are when I see reference picture or have a plastic ball im my hand.

I analyzed thousand times any objects , but when I sit to maxwell I get confused. What is Nd of this brick? it is quite reflective at creasing angle , but what reflectance color? what is roughness of "diffuse" layer ? is this color more saturated? and after all this test renders I get with nothing in my material library.
oh ! mayby roughness have to be low but small bump is what I need ?
zonk! another failure.
so more saturated colors in roughness 80 and more light from second "specular" layer?
baaaad!

In other engines it is WYSIWYG but in maxwell nooooo it is always guess and predict.

Anyone have a simple workflow , how to decode "deasemble" materials ?
some tricks to KNOW is this light is from High Nd factor ? or low Nd and higher refl. colors?

and my another test with rooftiles little pimped in phshop

[zdeno]

I will try again to post .... a few day ago I tried to post this and lost connection with maxwell site and forum for almost week. So I will be carefull now

*************

Hello again!

Can anyone help me with understanding roughness?
Is there any thread on this forum that describes roughness value vs reflectance? I have searched but failed.

How is it possible ?

If I put perfect sphere inside skydom. Light is identical from all directions.
and all reflected light from this sphere is different with different roughness. Measured in photoshop. reflectance color is always the same. so why there are differences?
When I set reflectance to about 235,235,235 I assume It will reflect THIS SUM of energy and roughness will change only directions of rays. But in overall it should all add to 235,235,235 (in perfect calibrated scene) because from every possible angle there comes identical sum of light to bounce back.

it is even worse. If in refl 0 and relf 90 is the same color so why there is "shading" even on perfectly clear 0 roughness , and 30 and 50 and lambert. almost "flat ramp" is with about 70 roughness value. I tried with many Nd values from 1 to 1000 and with. r2 file 90-0 to get rid of fresnell effect.

how in the name of Lord can I predict what will show in framebuffer if Nd has it ramp. roughness has it ramp , there are 2 colors depending on angle of view.

there is a great research about Nd vs reflectance from Thomas Anderson. Mayby there is something about roughness ?

and again 255,255,255 sphere lit by 255,255,255 skydom have "shading" so why there are info in NL materials ? "if you set 255 value color ALL light will be reflected back"

so ? anyone can help?

[gotoxy av-media]

hi zdeno,

i tried to reproduce your setup:

perfect sphere inside skydome with same color in refl0 and refl90 slot...is this correct?

then i rendered this scene with different roughness values applied to the sphere...
and i saw what i expected a solid color sphere!
so i cannot confirm what you have posted.

munch

[gotoxy av-media]

it is me again,
i have to correct myself!

after doing some more tests i have to confirm that behavior...

my sphere inside the skydome has refl0 and refl90 with RGB128,128,128 and roughness 0.
and when i measure the center of the rendered sphere it shows RGB 38,38,38 and at the edges it goes up to 45,45,45 ?!

first i thought it caused by scattering or vignetting, but even setting these values manually to 0 and hit the refresh button there is no change...

so no idea how to fix this?!

i can live with this small gradient...

munch

[zdeno]

thx for confirmation but roughness 0 is not the biggest problem. This gradients comes from Nd value (but doesn,t disappear even with 90,0 .r2 file). but shinning grazing angles is normall.

Most annoyin is +70 roughness behavior ... where Nd shouldn,t be noticable.
with ther same refl .0 and 90 . there is some gradients too (even with .r2 file - Nd should be turned off).

With -70 roughness grazing angles are going to black and +70 are going to be lighter. WHY ? I don,t know.

and lambert material reflects about 20% more light than refl 99 material with THE SAME color.
Measured as mediana on all sphere. This is strange, and makes work with maxwell more complicated. You have to do hundreds of test renders to achieve real like looking materials.

I think this is why There were big cry when mxmsite crashed. Peole still don,t know how to make good materials . It causes all this scanline looking renders in WIP.

[Bubbaloo]

Version 1.7 is inferior to Version 2's roughness model. Big improvements were made.

[zdeno]

I don't claim 1.7.1 version roughness model is bad . I said only I can't understand it. So mayby there is any explanation on the web.

[Mihai]

The roughness model has been greatly improved in V2, it's one of the things the developers spent a lot of time on to get right, because it will affect the shading in all your scenes. You can check this thread for a good example (exactly what you did, with spheres under skydome): http://www.maxwellrender.com/forum/view ... 97&t=32630

Scroll down a bit to see the new roughness examples. It wasn't easy also to get a seamless change between roughness 100 (lambert) and 99 roughness, but now it works perfectly. A good example of why this is crucial is an interior where you don't want to use lambert for the walls because you lose shading definition, so you set them to 97-99 roughness. If the change isn't seamless, the shading on the walls will look completely different. They will darken too much depending on viewing angle.