I solved it with a sopsolver that stores the activation frame on the geometry, so the points know when to start, and the gasopencl was simple, no need for extra geometry
but still would be interested if this is possible at all. I hope at some point the gasopencl dop is like the sop version
Hello I need to control a mpm with some additional geometry to control the effect. it works fine with a wrangle but it is slow. I tried to port it to a gas opencl, but I can not see a way to import a second geometry, like in the new opencl sop, that has 4 inputs ports or a wrangle DOP that also has additional inputs.
is there a way to do this with the gasopencl dop node?
Hello, I want to fix points at the beginning of the sim and I use the "pintoanimation" attribute for this and animate it so some points are not not pinned anymore over time. Not sure if this is the right way to do this. It works, at least it gives me no errors, but the adjacent simulated points are moved into the pinned ones as the simulation evolves and by the time the pinned ones join the sim, there is already to much going on inside, so it kind of explodes. Is there a way to prevent the simulated points to move inside the pinned ones or is releasing pinned points to mpm a bad idea anyway? thank You Martin
Hello, great Masterclass helps a lot. One question I still have is there any reasonable way to combine the results of the mpm with other solvers, like vellum, bulltet or flip the onlu way to I can imagine right now to get interolated colliders woult be the points, but the mere amount seems a no go
Hello I would like to hide parts of a simulation dynamicly and control it by the sim points. In sop import there is a option to import a attribute as a usd custom attribute, so I can drive visibility by this. I wonder how to us a attribute imported by the TransformBySopPoints in the same way.
I often need to hide geometries in a sim at a certain point and it would help a lot if I could use a point attribute for this.
is the growing done with a consistent geometry, I mean, do all branches exist at the beginning, keep their vertex order and topology the same and growing just by transforming points?
i want to create a fx layer for rendering that adds point attributes to a existing geometry. for example I get some fire wood geometry and I want to add a animated point attribute, that has some animated values to blend in a ember shader for lookdev. Is it posssible to just write out this added attribute with it's primopath, so it can be layerd in later?
right now I get allways the complete data set for the geometry
If things work fine and then suddenly start to cause problems or stop working, most of the time some environmental issues had changed. Did some things change like hardware, additional software, updates. Keep in mind the automatic OS updates. A workstation is often a changing environment due to updates alone. As of this complicated nature I guess You are lucky to find the problem, but holding someone responsible for this is another thing. This might be the case for systems that advertise this kind of feature, but this often works only in a very closed system with minimal system changes and very slow development.
If that kind of stability is important to You, I recommend to work on a very limited system with close to zero changes, beside of security critical updates, not automated. Redundancy would also help.
Hello, sorry for the delayed response. We use houdini 19.5.640 and Renderman 25.2.
Snapshots work until You view the Renderman Lop, as it sets the Color AOV to ci, if You view the node before the snap shots work again, and it is not really the snap shot, it is the thumbnail. The actual Snapshot that opens is correct and You can see that the Aov is called ci instead of color
Hello, when working with Renderman snapshots create black thumbnails, it seems that the snapshot tool uses the color channel to create the thumbnail. Is there a way to use Ci instead of color? thanks Martin
Hello, I try to gather all files needed for a usd layer in tops with usd_import_files. It worked so far for most cases, but I had to switch to stitched clips and now it does not get all files needed. it stops at clip, manifest and topology. The actual usdc files that are in the clip do not show up as items in the graph. Unless I display the clip file in a Solaris view, then pdg is able the get all the files down to the lowest level. When I scrub the time line even the complete range is collected
I prepared a little demo setup to show the behavior
It would be best if the usd_import_files now would be able to collect all files that are needed for this layer
if anybody has a idea how to solve his without displaying the clip usda in a viewer, it would be super helpful