Hello!

Is there a fix for this issue? We are currently experiencing this and I would love to know a more solid solution, at the moment we are just editing the .usda in a text editor

Brilliant, Thanks mate!

Hi noticed this in the release

"Lama nodes exist in this version of MaterialX, but they are not fully functional. Until Lama is completed, the equivalent Houdini nodes are hidden, to avoid confusion."

I there ant was to access the nodes despite them not being fully functional.

Best,
Rich

Hey,

I was wondering how I could render a Sheen AOV in Karma using Mtlx.

It would be great if we would be able to label certain lobes in MaterialX too. If this is possible, please could someone let me know?

Best,
Rich

unfortunately this is not possible with this node. The mtlx definition is also not great as it seems to have an additive overlap, causing the corners to inflate. If you want to get the desired effect with an adjustable falloff, you will have to build/write the node yourself. this is doable with the materialX nodes so it shouldn't be too big of an issue!

Can you send me over the release note describing the fix please?

Doesn’t seem to be affected by dicing at all, seems to occcur from rendercam and at any distance with any dicing setting, restarting the render too also doesn’t cause it to update in viewport which usually fixes dicing artifices so I am clueless to what is causing it

Just wanted to see if nyone else has noticed a bug when mixing BSDFs with mtlx mix

when set to 0 it shows input1



and when set to 1 it also shows input 1



if you use a value of 0.001 it does show input0



this seems to occur specifically with the BSDF mode.

this makes shader building difficult. It seems when a constant is plugged in with the value of 0 the issue does not occur



would be great to see if anyone else has had this issue?

Best,
Rich

I have noticed when using the mtlxheightntomal node there is some significant artifacts.

I am able to remove them by subtracting the artifacts but they shouldn't be there in the first place.

I was wondering if anyone had any ideas what could be causing this issue.



In the image above you can see the artifacts clearly, bare in mind there is no texture, just a value of 0 as an input



here is the world normals, the result we would expect from a value of 0





and here is the height normals subtracted by the world normals, isolating the artifacts.

I've noticed this issue on multiple meshes.

Best,
Rich

Is there any way we can access derivatives functions in karma for MaterialX at the moment or is this something that has to be supported on the MaterialX side of things? It would be incredibly useful to make screen space curvature and other useful nodes for shading.

It would be incredibly useful to have these functions if they are not yet implemented so it would be great to hear if this is something sideFX can/has implemented or if this needs to be suggested to the MaterialX team.

Best,
Rich

Hey,

Since there is no ambient occlusion definition that ships with MaterialX other than an incomplete osl definition I was wondering how this node works?

I would like to write my own definition of Ambient Occlusion as well as other useful shaders such as curvature so I would love some insights into how this node works.

Also I was wondering how exactly karma interprets MaterialX shaders since it doesn’t use HdMtlxCreateMtlxDocumentFromHdNetwork So I’m assuming it doesn’t convert the shaders to GLSL. I’m also assuming it doesn’t convert it to Vex due to the lack of compatibility with vex shaders so I’m wondering what the materials are being converted to become usable with Karma?

Best,
Rich

Hey,

Just did a quick test and it seems the mtlx standard surface run great with prman but not the bsdfs, mainly the surface node cannot be translated to osl with the current version of MaterialX shipping with houdini

I've also tried lama defintions generated from mtlx2hda in renderman with no luck throwing a similar error

I'm assuming this is due to the osl generations not being fully supported at the moment in 1.38.1. in this case waiting for sideFX to ship with a later version may cause it to work but due to a lack of closure support I thought renderman couldn't use osl for
bsdf shaders. I haven't looked into getting MaterialX to work with renderman too heavily so im not too sure what exactly is going on under the hood to allow for MaterialX support for standard surface and not the bsdfs/lama nodes as at the end of the day, the standard surface is defined from the bsdfs. I dont know if theres some sort of translation under the hood mapping the standard surface to pxrsurface or something like that instead. probably a good idea to ask someone on the prman team.

I know there are some pretty significant changes happening to OSL at the moment to allow greater support for MaterialX so that may also be something that will make this easier to deal with. but I'm probably not the best person to ask as I haven't done a huge amount of research into MaterialX support for prman.

I hope you find a solution! and if so id love to hear how you went about it!

Best,
Rich

Hey Oliver,

Due to the lack of support for Absorption in karma it is not possible for the time being to have the exact definitions working as expected for dielectrics.

I used Mtlx2HDA and managed to successfully create all the nodes needed to recreate the definitions present in 1.38.2 and onwards but due to karmas lack of functionality for absorption, it’s currently unsupported.

I also ran the lama definitions through Mtlx2HDA with no luck for all nodes apart from layer, mix etc.

For the most part, all other Lama Definitions work fine! When reconstructing with the nodes in the nodegraph Eg. Lama Diffuse, lama sss, lama layer, lama mix, and so on.

I have had some minor issues with the Conductor Definition as for some reason the Thinfilm BSDF cannot take a node input despite having input ports on the node. I’m assuming this is a bug on sideFX side. This means you can’t have the same ior/thickness calculations described in the lama definitions. But if you wanted to exclude Thinfilm and have a limited lama definition than it’s possible.

You also cannot layer a Thinfilm BSDF over a conductor in karma as described in the lama Defintions so you have to layer the Thinfilm on a dielectric and then layer that on top on the Conductor to get it to work so this will need to be supported too.

That being said, these seem to all be very minor bugs which will be addressed in upcoming releases.

So all in all, yes, you can have the lama nodes operate “fine” with limited functionality for both dielectric and conductors. And full support for all other nodes according to the current Mtlx Definitions. There are some limitations like karmas lack of custom lobe naming for use in AOVs. Rendermans implementation has some additional features not defined in the Mtlx Lama Defintions and according to the MaterialX Slack there may be some updated defintions to come to match the Mtlx Defintions closer to the Renderman Defintions.

I hope the conductor issues due to the Thinfilm BSDF will be fixed when updating to 1.38.2 or higher definition as it seems like a small bug but for the dielectrics we will have to wait until absorption is supported in karma. even if the hdas are updated to 1.38.2 definitions this won’t work until absorption is supported.

Hope this clears some things up!

Best,
Rich

Hey Rafal,

appreciate you doing a deep dive into this!



As you expected, the 2nd option (using the subnets for <nodegraph> and <output>) works perfectly in MaterialXView although as you said there is issues reimporting through Usd Mtlx file format plugin back into Houdini.



In the third example (which allows for BSDFs to be re imported) works great for reimporting into houdini but unfortunaly is not able to be imported into MaterialXView due to syntax errors.

I thought including this inside a subnet may fix the issues but I get a message saying:

"nodegraph_subnet contains a subnet subnet1, this is not supported by MaterialX currently"

I'm assuming this is due to MaterialX not supporting nested node graphs?


Thanks again for all your help as at least now we know a method of saving out BSDFs and as a .mtlx between Houdini files which is a great thing to know!

Best,
Rich

Hey,

So I used mtlx2hda to update to the latest pbrlib_defs.mtlx and all the nodes seem to work as expected apart from the layer.

When updating the HDA we do get the VDF option but for some reason we cannot layer a dielectric on a VDF and get a result.

I'm assuming this is due to the limitation that Karma does not currently support Surface and Volume materials on the same prim.

Best,
Rich

Thanks! this was really useful! appreciate it!

And do I run this in the Python shell?

Unrelated to the general use of the BSDF but where my main issue was stemming from was that it seems although the thinfilm BSDF can take floats. when plugging a constant into the thickness and IOR it seems to break so I am going to submit a ticket in regards to this.

Hi,

you can use the configure primitive node in lops to set the kind of a primitive.

Great!

I can seem to get it working now with the Dielectric, I just had the scatter mode on transmission not transmission and reflection or just reflection. When the reflection is enabled it works as expected so that's a silly mistake on my end as the thinfilm bsdf just modifies the specular reflection lobe as mentioned in the spec. This is why it shouldn't work with diffuse but should work with dielectrics and conductors.

Still not too sure why it doesn't work with a conductor though, this will need to be changed since the lama node defs have thinfilm layered on top of the conductor so for any kind of lama support this would need to work. I'm hoping this is just to do with the nodedefs that created the VOP nodes and when updating to 1.38.2 and above then regenerating the VOP nodes this will work.

Best,
Rich