Hey all,

I keep running into the fact that if I put down a reference LOP in 19.5 it errors with a "layer not found"
19.5 seems to be using the 2.0 version of the asset where 19.0 was on the 1.0 and works fine.
If I revert back to the 1.0 definition in 19.5 it works fine as well.
I thought it was probably me doing something wrong because other LOPs that use it under the hood work fine (like assetreference lop) but those seem to be using the 1.0 version inside of it...

Is the 2.0 version just broken? I cannot get it to work. I cannot finds any post about this, but I cannot be the only one with this problem. Looks to me like this is also why quixel bridge for 19.5 still isn't working (as it brings in the 2.0 of the asset) because if you set it to use the 1.0 version it works fine.

So is this a bug or am i missing something?

This time I did finish in time, and i’ve recorded my entire process in this 6 hour video.
Rendered in Karma (CPU) and composited in COPs

Hope you like it.




Get the project files here:
https://timvanhelsdingen.gumroad.com/l/mardini_bulb2023 [timvanhelsdingen.gumroad.com]

https://www.patreon.com/posts/80290347 [www.patreon.com]

Had to leave last night before I could get a final render out, so a bit late but here is my render for CRASH and a recording of the process:

Rendered in Karma (XPU)

Download project file:
Gumroad [timvanhelsdingen.gumroad.com]
Patreon [www.patreon.com]

jsmack
Has anyone tried disabling native command queueing to see if it fixes the crashes with OpenCL?

HOUDINI_OCL_FEATURE_DISABLE=CL_DEVICE_DEVICE_ENQUEUE_SUPPORT

Just tried this on my 4090 and this seems to solve it! What exactly does this do?

Guess i'm not the only one with this issue. I had instabilities with a certain Vellum setup, but noticed that when switching to openCL CPU it was all fine. But in GPU it crashes. 4090 drivers have been a mess... There's also a memory leak issue for redshift where it can only use 16gb of vram, wonder if it has anything to do with eachother.

STONE.
Redshift, and some minor compositing in COPS (houdini)



also did a recording of the entire process. Here's the sped up version:

GROW

Pretty happy how this one turned out. but i'm not 100% sure if I qualify because I only just now read that I'm not allowed to use any megascans assets, and I used some for the environment. The growing plants themselves are made by me though. So hopefully this still counts. Rendered in Redshift.

edit:
Ok aparantly you're also not allowed to composit in anything else than cops? Well ok I guess that disqualifies this as I comped it in Fusion. So I'll withdraw my entry for contest then, but I'll leave it here for funsies. But might I suggest removing the comp rule, because it doesn't make a lot of sense to not alow any other compositing package.

Does anybody know how to get motion blur to work on something with changing topology (so from the velocity attribute) I cannot get it to work with any renderengine (tried karma and redshift)
I have it turned on on the rendergeometrysettings but I cannot get it to work even on a basic particle sim. Tried with cache sop, tried saving to .USD, nothing seems to work.

I'm probably just overlooking something but i've been trying since yesterday and I can't figure out why it's not working. Does anybody know what i'm doing wrong?

BrandonA
I noticed a couple of things in your .hip file:

Because it's a simulation, you will want to turn on “All Frames in One Batch” under the ROP Fetch tab of the ROP Geometry node. Without this enabled, it is rerunning the simulation each time up to the frame that is being cooked. So the cook time for each frame will get progressively longer because it will have to resimulate the previous frames. “All Frames in One Batch” ensures that all frames are done in a single batch, and there is no additional overhead.

I would also disable “Cache Simulation” on the DOP network since it's not necessary because the sim is being run out of process and the files are being written to disk. For this sim, it's less important because it's very lightweight but if you're running multiple concurrent heavy duty sims, you might eat up all your RAM quickly.

One other thing – although you should definitely enable “All Frames in One Batch” for a simulation, it was no where near 2 minutes per frame for me when running the ROP geometry with the original settings. What kind of machine specs are you running with?

Hey brandon,
Thanks for your reply. I think i've tracked down the issue. So the “All Frames in One batch” indeed stops it from re-simming the entire thing, but the other problem was that apparently loading in network paths inside houdini.env severely slows down the startup time of TOPs (I had all my presets and OTL's on a network drive), even more then normal houdini startup time. Moving all my otl's and presets away from a network location and to my local drive has reduced startup time for TOPs from 2 minutes to about 35s. Still more then a clean install but at least this is workable.
Thanks!

Okay, i've been trying to get something going for the past 2 hours but it's just not working. I've setup some wedging, which works correctly. But whenever I run the rop output driver it just takes absolutely FOREVER to cook. I've attached a very simple example scene with 5 wedges of a shattered sphere falling. Runs in realtime if I run this directly, but takes 2 full minutes PER FRAME when running through tops.
Can anybody point me out what the hell is going on or what i'm doing wrong?…

NimaNolan
Hello my friend.
When did it happen?
Maybe isn't bad to use System Restore.

problem is fixed now with the help of support. Had to delete pythonpath inside of windows and now it works again

Hey guys,

So i'm running into a strange issue here.
My shelf tools are giving me this error:


It's doing it in all my desktops.
I already tried deleted my houdini folder from my documents, and reinstalling houdini alltogether, but it doesn't seem to do anything.

Any ideas?