New Mental Ray shaders development

[Kzin]

last idea would be to uninstall SI completely incl all user folders and reinstall it.

[Daniel Brassard]

if i should apply or not the multiply by PI

for mia and legacy shaders: apply the 1/Pi (0.3183)

for the new shaders like mila: don't, its already taken care of internally.

This is all relative to the texture you are using sRGB, Jpg, BMP texture have Gamma, EXR and HDR are linear.

Anybody tried EXP vs sRGB on the new mila?

Dan

[Maximus]

Layering system is very nice.
this is 6 layer including transparency and subsurface, 2 scratches map, bump, 2 glossy layers and 1 coat on top, I couldnt manage to render this with mia_material in a reasonable amount of time.
Pretty impressive

[Tekano]

** edit ** sorry rray, the string option shader is working properly as a global string , I just had to setup the user_ibl_env light properly to see an effect as described. thanks again!


@ Maximus. green scattering only huh? got me for a while as the auto ppg generates labels for front radius and back radius as X Y Z , when really it means separate radius for R G B ! wonder if its easy to fix this.

also, think the default values in scatter resolution should be 0.5 not 1 -100 from reading from this thread but just methodically going through docs now to check up on stuff

[rray]

That looks great.

Tekano - dnt.w.! good to hear it's working!

*edit* moved shader description to the post with the shader

[Tekano]

yeah sorry rray did not have lights setup correctly I believe. was just thinking I also could have turned on verbose and looked out for the 'light relative value' here is the obligitory before after shot of it working. top left is mila shader , bottom right is split render mia shader with the global string as 0.318 on left and 1 on right, can clearly see it is brighter at 1 on mia shader whilst either value had no effect on mila shader as expected. phew!

cheers!

[Maximus]

Don't know if this works but for the light relative setting try this addon, should allow to set arbitrary string options...
__
*edit* looks like it's working

btw the shader ppg is strings, internally it's rewired like this:

numbers only in string (394) -> sets an integer string option
numbers and decimal point in string (232.1) -> set a float string option
comma in the string (20,10.34,105.0) -> sets a vector2, 3 or 4 string option - depending on the # of commas
string = on or true -> sets a boolean string option (true)
string = off or false -> sets a boolean string option (false)
in all other cases, or if the sting is inside double quotes -> set a "string" string option

Sorry i was absorbed into mila world and didnt see this, thanks a lot Reinhard for your effort, this is really nice!

[Mr.Core]

Don't know if this works but for the light relative setting try this addon, should allow to set arbitrary string options...
__
*edit* looks like it's working

btw the shader ppg is strings, internally it's rewired like this:

numbers only in string (394) -> sets an integer string option
numbers and decimal point in string (232.1) -> set a float string option
comma in the string (20,10.34,105.0) -> sets a vector2, 3 or 4 string option - depending on the # of commas
string = on or true -> sets a boolean string option (true)
string = off or false -> sets a boolean string option (false)
in all other cases, or if the sting is inside double quotes -> set a "string" string option

Maybe a stuped question - but can it help to enable iray mode with more controls than in the actual ctrlGhost ? Like with some script for 3dsmax ( i forgot its name =\ ) which had ability to use various combination CPU+GPU

[rray]

No that's a good question but I don't know, this would be interesting. It could be from it's UI-less design, that it can access some extra variables that ctrl.ghost couldn't access.

[AlexanderM]

It works!

[Mr.Core]

But unfortunately seems to be that doesnt understand keplers, only CPU
http://www.dreamwalks.com/MEDIA/product ... ons_24.pdf

[gustavoeb]

thanks rray!

I tried progressive mode here... I wish it would downsample better (when I bring progressive/unified Min under 1 it stops working), so we could see things FASTER, but it still is something that could be implemented, and everybody would be spoiled by it easily. I noticed it glitches from time to time (at least here) when switching between Preview and Render Region, but no biggie.

The layering library looks great. Appart from the fact that XSI just explodes when you connect something in the wrong place and it cannot deal with the exceptions, but that is expected of its beta nature. The scattering shader and its speed is what impressed me the most! I can only wonder what this shaders would do with Multi Importance Sampling.

Lastly, SIs light shaders prove theire age again. Having been built in a age where no one cared about energy conservation or linear workflow, in my tests reflections look either too bright or too dim depending on the area light mode being using in the glossy reflections layer. The portal light proved much more consistent, and, fast (general feeling, not scientific tests, still confused as to why).

But this statement by Bart sounds interesting: "We think we may provide a set of mila lights to guarantee the right light transport. This may be a second step."

[Kzin]

thanks rray!

I tried progressive mode here... I wish it would downsample better (when I bring progressive/unified Min under 1 it stops working), so we could see things FASTER, but it still is something that could be implemented, and everybody would be spoiled by it easily. I noticed it glitches from time to time (at least here) when switching between Preview and Render Region, but no biggie.

The layering library looks great. Appart from the fact that XSI just explodes when you connect something in the wrong place and it cannot deal with the exceptions, but that is expected of its beta nature. The scattering shader and its speed is what impressed me the most! I can only wonder what this shaders would do with Multi Importance Sampling.

Lastly, SIs light shaders prove theire age again. Having been built in a age where no one cared about energy conservation or linear workflow, in my tests reflections look either too bright or too dim depending on the area light mode being using in the glossy reflections layer. The portal light proved much more consistent, and, fast (general feeling, not scientific tests, still confused as to why).

But this statement by Bart sounds interesting: "We think we may provide a set of mila lights to guarantee the right light transport. This may be a second step."

use the user ibl_rect shader with hdr map applied. way better looking reflections and it renders alot faster then standard area or portal or physical light because of its importance sampling.
you do the same in arnold, so does it in mr.

and wrote a mail to ad that they have to overwork the whole old preset stuff, we need more people to write such mails. and you can then add that ad should implement progressive with correct api in xsi too.
and i dont want to hear the "no, i will not write, because ad is not listening" thing. ;)

[Tekano]

replicating what ive posted on the mental ray forum , but worth posting here anyways

heres some more scattering / portal light / user_ibl testing



and a quick sequence test, early days and fumbling through the settings so wouldnt worry too much about the flicker just yet

[Rez007]

and wrote a mail to ad that they have to overwork the whole old preset stuff, we need more people to write such mails. and you can then add that ad should implement progressive with correct api in xsi too.
and i dont want to hear the "no, i will not write, because ad is not listening" thing. ;)

I agree, people definitely need to write to AD more, whether you hear back from them is another story...I wrote a couple of times in the past about lighting, and never heard anything back. Recently I wrote two different things (one was about 4 pages long) of ideas/concerns/suggestions. I sent it to the AD online suggestion area for Softimage (which Softimage was finally added to the list a little while ago...)

http://usa.autodesk.com/adsk/servlet/in ... id=1109794

I heard back from one guy there, and asked if he could pass it around to maybe some of the higher-ups as well - never heard anything back from the higher-ups (which I didn't expect too much anyway), and the other longer one went to Cory Mogk - never heard anything back.

He shares his email address on the Area Blog of his "re-introducing" thread.

"(my name is Cory Mogk and our email structure is: firstname.lastname@autodesk.com)"

Whether you get a response back or not, you never know, but I am confident that they at least read what people write, as long as you keep it professional and not ranting could probably go a long way.

[Rez007]

@Tekano - those clouds look really nice!