lucavfx
I followed your instructions to the letter, the nparticles cache is created on disk,
however upon attaching the nparticles cache to the duplicated nparticles transform node nothing appears when scrubbing in the viewport. I get the following warning I suspect might be related:

// Warning: file: CProgram Files/Autodesk/Maya2018/scripts/others/doImportCacheFile.mel line 99: Channel names did not match selected objects. Using channel houdini_engine_particles_v0011_0Shape_id on object houdini_engine_particles_v0011_1Shape. //

Hm, here are a few things I can think of that you can check/try:

1. Take a look at the cached files, and make sure the file sizes do look like sensible. For example, the file size shouldn't be too small (less than a few kilobytes would be too small).

2. Before exporting the cache, do a sync to make sure the nParticle node has the exact same set of per-particle attributes that your asset is outputting. There might be issue if your asset is actually outputting less attributes than your nParticle is expecting. Otherwise, the nParticle node would have attributes with no data, and I'm not sure how nCache would react to that.

3. I wonder if some per-particle attributes are causing issues during nCache export. The simple way to test this is to try removing all the point attributes in the asset itself. This means the particle should have no other attributes except P. Then, in Maya, redo the sync, export cache, and attach cache steps. If this works, then there's likely some attributes on the particle that's causing issues with nCache.

4. If there's no particles in the beginning of the simulation, try exporting the cache from a frame that has particles. (I don't expect this to be an issue, but worth trying just to be sure.)

“Use Asset Object Transform” is probably only useful when the asset is an Object asset. An Object asset has its own world transform, which isn't exported to Maya by default. Enabling that option will export the Object transform into Maya.

In the case of a SOP asset, Houdini Engine creates a dummy Object node, and puts the SOP asset inside. However, this Object node is always at identity. So when this transform is exported to Maya, the asset looks like it's always at zero.

I see your geometry has the “shop_materialpath”. There was a performance bug when outputting string attributes. This was fixed in 16.0.785 or later. You could test if this is the issue by deleting the “shop_materialpath” attribute in the asset, and see if there's any speed difference.

Another potential slowdown is when the output geometry from the asset tries to preserve the input Maya geometry's locked/unlocked normals and soft/hard edges. I don't know if this is happening in your case, because this depends on the input geometry from Maya. You could test this by deleting the “maya_hard_edge” and “maya_locked_normal” vertex attribute in the asset.

Using a switch node would be the way to do it. Many people use the switch node to turn off and on parts of the network.

Maybe another approach is to use multiple asset It's also possible to connect multiple assets together. This means you could have a separate asset for each stage of your network. Then, you'd connect them on the Maya side.

Does the Houdini asset take input geometry from Maya?
How many polygons is the output geometry? What attributes are on the geometry? There's some overhead when converting geometry from Houdini asset into Maya, so it won't be as fast as inside Houdini.
If you could post a .bgeo of the output geometry, I can test if it's slow for some reasons.

Which Houdini and Maya version are you using? There was a few performance fixes recently, so you could try out the most recent daily build.

It looks like something is trying to modify the connection at render time. The Maya plugin shouldn't be making any connection changes unless it's doing a sync. Do you have auto sync on? But even when doing a sync, I wouldn't expect this error to happen.

Does this error happen if you render locally?

It'd be really helpful if you can post an asset and scene file that can reproduce the problem.

What primitives is the asset outputting? And are you using “Split Geos By Group”?

When “Split Geos By Group” is off, and the asset only outputs polygons, all the polygons should be outputted to the same “_0” Maya node, and shouldn't be outputting “_1” and “_2” mesh nodes.

CMake seems to have trouble finding the MSVC compiler.

depperm
To use the NMake generator with Visual C++, cmake must be run from a shell
that can use the compiler cl from the command line. This environment is
unable to invoke the cl compiler. To fix this problem, run cmake from the
Visual Studio Command Prompt (vcvarsall.bat).

As the error message suggests, I believe you need to run cmakefrom inside the Visual Studio Command Prompt (vcvarsall.bat). This should allows cmaketo find the compiler that you want to use.

How is the Maya plugin registered with Maya? Did you set the MAYA_MODULE_PATHenvironment variable? If you've only set MAYA_PLUG_IN_PATH and MAYA_SCRIPT_PATH, this is not sufficient. In your case, make sure MAYA_MODULE_PATHincludes hfs16.0.491/engine/maya.

I made an asset that attempts to identify/clean these unsupported polygons. Maybe give it a try and see if it helps.

tricecold
I tought some how for each in import fields was breaking the vels, so i fetched each field seperately with dopimport1, display flag was set properly on where the data was.

On Maya's fluid shape node, if you enable “Display -> Velocity Draw”, are you able to see the velocities drawn in the viewport?

tricecold
I would be very happy to get back BGEOs also, but a filecache node doesnt work, because it cannot tell maya to change frame, unless there is a way I dont know, I would be happy to hear that.

String parameters are evaluated as Houdini expressions. So you can use variables like $F and $F4 in file paths to allow changing frames.

tricecold
I checked the mcx cache from Maya and it outputs color channel, which I already tried renaming rest from Houdini, but it overrites the values. I wonder if the connection editor would allow me to reqire these manually.

Does the color appear in the fluid shape that the asset created in Maya?
After caching, when you import the cache, you also need to make sure “Color Method” on the fluid shape node is set to “Dynamic Grid”.

tricecold
The Problem is when I cache out maya fluids cache and load back in the cache, i cannot see the velocity field.

Is the velocity field being outputted in the display SOP? The Pyro shelf normally creates a “pyro_import” object that contains two SOPs. One has the display flag set, and the other one has the render flag set. The displayed SOP is the one that'll be outputted to Maya. But the displayed SOP is used for visualization in
Houdini, and doesn't contain the velocity field. In that case, you probably want to set the display flag on the rendered SOP.

tricecold
My other problem is the scale, by default fluid container size is set to 1 and its transform is scaled instead it also get transforms, this becomes a problem for some of our tools, voxel position is wrong in objectspace. Was this an intended design ?

This was the intended design at the time, but I would also prefer it to have the proper object size. I tried to change this a long while back, but I ran into some issues. Let me take a look at it again, and see if this can be changed now.

tricecold
Before my output node,
Do I have to rename vel to velocity, or engine fluid will do the conversion?

The Maya plugin should handle the conversion. Maybe the displayed SOP isn't outputting the velocity field?

tricecold
What workflow would you suggest to pass through the rest field, i currently rename it to color as maya fluids have color caching, so i was thinking to replace that since thats a vector too.

Unfortunately, your options are very limited, since Maya's fluid shape node only has a fixed set of fields that can exists. Like you said, the best bet is probably to store the Houdini field in one of the available fluid shape field. You can do this in your asset by renaming your volume to Cd..

By default, Maya loads the Houdini Engine backend into the Maya process, which is probably causing conflicts in Maya 2018. It's also possible to load the Houdini Engine as a separate process, which would avoid library conflicts all together. Normally, you'd select it from the Houdini Engine preferences. But since you aren't able to load the Maya plugin to begin with, you could run the following MEL to set the same preferences:


Then, load the plugin.

The benefit of supporting parallel evaluation is to allow multiple Maya nodes to evaluate simultaneously. So this doesn't stop Houdini Engine's internal cooking to be multithreaded. And like Houdini's cooking, Houdini Engine's internal cooking is already multithreaded. So there's likely not much benefit from the Houdini Engine side.

But since the global evaluation mode needs to be set to “DG” in order for Houdini Engine evaluate correctly, this means no Maya nodes can make use of parallel evaluation, which is a performance improvement in recent Maya versions. So the main benefit for Houdini Engine to support parallel evaluation is to allow other Maya nodes to evaluate in parallel if those nodes support it.

So the performance benefit would depend entirely on what nodes are in a Maya scene. I suspect there's not much benefit if the Maya scene is mostly doing dynamics. Maya scenes that have heavy rigging and animation might be the ones that see performance improvement. The best way to see whether there would be improvement is probably to take a Maya scene that doesn't use Houdini Engine, and toggle between DG and Parallel evaluation mode.

As you saw, the parameter interface is only updated during a sync. Unfortunately, there's no way to trigger a sync from parameter changes right now. The Maya plugin should probably try to detect these hidden/disabled parameters. I've submitted a RFE (84792) for this.

I tried creating an asset like you described, but I can't seem to get the same behavior that you're seeing. I've attached the asset and scene file that I created. With those files, are you able to reproduce the same issue? The files are created with Houdini 16.0.557 and Maya 2016.

I'm not sure what that strange cooking is. When would this strange cooking start? Does it happen after you change a parameter? Clicked a button? Changing the current frame?

It'd be helpful if you can post a simple asset that can reproduce this issue.

The licensing mechanism relies on a few pieces of network information, like the hostname, IP address, and MAC address. So it's possible that a license that you get while on cellular data doesn't work when you're not connected to cellular data. I'm curious to why the licensing suddenly stopped working though. Have you changed your hostname recently?

A few other things to check:
You should be using local licensing mode. If not, there's information on how to switch here:
https://www.sidefx.com/faq/apprentice-local-licensing/#faq74 [www.sidefx.com]

Also, check that the license file is writable by your user account.
https://www.sidefx.com/faq/apprentice-local-licensing/#faq78 [www.sidefx.com]

While connected to WiFi, regenerate the Apprentice licenses. You could do this by clearing out the content of the license file (licenses or licenses.local). Then run Houdini again to get the prompt for installing a free license.

If none of those worked, try contacting support. There are instructions here:
https://www.sidefx.com/faq/apprentice-local-licensing/#faq79 [www.sidefx.com]

Check out the attached asset. It uses the following:
0.8+0*$F+0*chs("../../parm")It has to be on a numeric parameter on the material node.

Note that the parameter for the input objects is an array. Similar to the “Set to Selection” button, it's possible to pass in multiple objects of different types, which is a new feature in 16.0. The objects would be merged together on the Houdini side.