For as far as I know this is a consistent bug in 5.5.4, and is resolved in the later versions. That being said, according to UE and USD importers, it might resolve switching off the Never Needs Cooked Collision Data toggle under the Collision>Primitives tab.

I can't figure out how to achieve holdout with transmission, even after trying to use LPEs. Is there any way to accomplish this, or is it a limitation of Karma?

Alright, I just realized that I should first assign LPE tag to the holdout geometry to achieve this.

I imported images(EXR) rendered in Karma into Nuke, and they also used the same ACES color space, but the colors were completely different,Why?

Houdini21.0.440

I attach one frame

To me it looks like you're using an older Nuke color config. Check the one that says ..aces-v1.3_ocio-v2.1 in the dropdown of your project color settings.. then the Nuke viewer config should also say the same as mplay: ..SDR video (sRG - Display) instead of sRGB(ACES).

See if that improves it.

Hello,

According to the H21 Update Karma XPU supports "shadow linking". I have a scene with a background asset, a hero asset and a big diffuser asset. I don't want the background asset to recieve shadows from the diffuser asset. Is this possible to achieve with shadow linking, and if so, where can I set it? I could not find it in the "render geometry settings" node.

Karma XPU now supports shadow linking for up to four geometry subsets per scene. Shadow linking means that only a subset of a
geometry will contribute to the object's shadow.

Thanks in advance

Did you find something, I am also looking for this???

tamte
you can make a function that combines masks and use it in for loop for each mask or whatever you are doing

If the number of loop iterations is high, you may get better performance by putting the conditional outside of multiple loops.

e.g., instead of:

It may be faster to do:

This reduces the number of conditions that need to be checked per iteration and can help the compiler optimize.

Note that I also replaced the if chain with an else-if chain. This allows the code to short-circuit, meaning that if one of the upper conditions is true, it doesn't need to check any of the lower conditions. e.g., if op == 0 is true, it doesn't need to check op == 1, op == 2 or op == 3, as op can only have a single value.

Thanks for both replies, I apreciate!

@tamte : What is that line about? I've done vex for many years and never came across this before:

Bonus question: if I want a toggle for +-*/ so any click on one toggle will de-click the other options?

Hello, I am trying to make a wool tearing simulation and I couldn't figure out how to freeze the deformations after vellum cloth breaking. I've added a video to show an example. The cloth falls down, I want it to freeze slowly after breaking so it looks more like a wool tear and not a chain.

Hi everyone ! I'm pretty new to Houdini and getting my a** handed to me by Solaris. I've been trying to render a VDB with Redshift. Adaptive Sampling and Denoising are ON. I'm using Houdini 21 and Redshift 2025.1.1.

- Half of the time, I get the same render in viewport than in the final render (Image 1, left). It's way too bright and very blocky, which I can't manage to get rid of. The renders use Adaptive Sampling, as expected, and take around 10 min. For info the density and Cd values of my VDB are all between 0 and 1, I corrected that already after seeing this fix under another post; and changing the VDB resolution doesn't change anything.

- The other half of the time, my viewport looks like image 2 (right). The render uses bucket rendering with a bucket size of 256 (it's set to 128 in the Redshift node). It also goes waaay faster (finished in seconds), but with a longer initial startup time. The look is closer to what I'm going for but doesn't seem to use denoising.

In the viewport, I can switch from case 1 to case 2 by toggling "Set Missing Widths = 0.0001" ON or OFF in my SOP Import node. Sometimes it works when I turn it on, sometimes when I turn it off. At some point, minutes or hours later, it "randomly" switches back to case 1.

The final render (Render to Disk) systematically goes to case 1.

Can someone more experienced tell me what is going on? I feel like a caveman witnessing fire for the first time.

use Vellum Hair Constraint
strong bend stiffness, (if necessary also increase substeps and/or constraint iterations)
play with plastic deformation to have it remember distorted shape to a degree

olivierth
@tamte : What is that line about? I've done vex for many years and never came across this before:
float combine_masks( float mask0, mask1; int op){

it defines a function [www.sidefx.com] so that you can use it later instead of typing the whole thing again, so later you can just do

Is there no way to freeze the points as soon as they break? Like inside of the vellum solver?

there is, you can set i@stopped to 1 if that's what you want, that's essentially what sleeping is using, not sure that's applicable to your effect thogh
you can also increase drag or do a lot of other things to force it to slow or stop

but from the effect you described I'd go with suggested approach, maybe with a bit of drag on top

AAAaaahhhhhh! Thanks!

Thank you for your help, I think I am just misunderstanding something, because I am completely comfortable working with dop networks but when it comes to vellum, I just can't figure it out I feel like it's still static even though I am inside the vellum solver node and also it doesn't even show any data in the geometry spreadsheet. I guess I shouldn't use the houdini vellum solver tool and just create it from scratch in dops

but the drag modifier really helps, so I guess ill stay with that for now, thanks again.

you should see stuff update when within Vellum Solver SOP, and since the dive target is DOP context you can do a lot there

however there were builds where diving inside any solver just stops updating viewport, so its likely you are on one of those builds

in such cases I usually lock the viewport when in SOPs looking on the output of the solver and then dive inside and do some work, while you can't see the guides, it's still better than no update at all