Hello,
I have a crowd sim that is made of agents starting in T_pose that I want to randomly transition into multiple random clip. Everything works fine but I also want to pick up a random clip time of their new state (so that they don't start walking with the same foot for instance).
In the agent State node in my crowd solver their is a random clip offset slider but it only works if the simulation start on this specific state (which I do not want). I know I have a some options like delay transition and such but because I run a vellum sim on top I would need to keep everything to a minimum warm up time.
Any help appreciated.
Thx,
Edit: with a small scene to illustrate my issue.
Edit: Solution in a video tutorial available here https://youtu.be/H-7E_F0W-O8 [youtu.be]
Found 45 posts.
Search results Show results as topic list.
Technical Discussion » Crowd solver transition not using random offset
- Edmond BG
- 45 posts
- Offline
Technical Discussion » Vellum crowd attach constraint issue
- Edmond BG
- 45 posts
- Offline
For anyone who comes across this issue, turns out this an bug that is now reported.
The full message from the support :
The full message from the support :
Unfortunately the vellum unpack is not properly correcting the target_pt attribute on the constraint primitives to the value in the merged collision geometry. It does correct target_prim, which is why attaching by closest location works.
Attached shows a correction factor for this file. Unfortunately, the precise update depends on what geometry you have merged together, and which agent if you are unpacking multiple agents. Either you need to encode the target point yourself in an invariant manner to unpack (which will not be simple, about the same complexity as fixing vellumunpack directly…), or use the primitive attach until the bug is fixed. In this case there will be a primitive attach value that matches the behaviour of the point attach.
This bug is now logged in our database as (ID# 106732). I will update you when our developers have fixed this.
Technical Discussion » Vellum crowd attach constraint issue
- Edmond BG
- 45 posts
- Offline
Well when you take a look at the second sim, you can see that it is possible to handle objects that do not originate from the agent as collision geo. And you can see that I add the object in the shape library before the crowd sim, so I don't think the issue is coming from here.
It is kind of a bummer to do the 2 sim into different solvers, in my exemple I only have 2 different shapes that have a low chance of colliding but in other setup you could have much more vellum that collide with one another so I would not recommend it workflow wise.
At the end of the way I think that the agent vellum unpack node might just support point attach constraint properly. I will just send a message to support because that is something that should precised in the documentation.
Thx for the help.
It is kind of a bummer to do the 2 sim into different solvers, in my exemple I only have 2 different shapes that have a low chance of colliding but in other setup you could have much more vellum that collide with one another so I would not recommend it workflow wise.
At the end of the way I think that the agent vellum unpack node might just support point attach constraint properly. I will just send a message to support because that is something that should precised in the documentation.
Thx for the help.
Technical Discussion » Vellum crowd attach constraint issue
- Edmond BG
- 45 posts
- Offline
Hey,
Thanks for taking the time too look at my hip file.
I am not sure if I understand you corretly but here are a couple of key elements that you might be missing :
Let me know if there is anything I am not understanding.
Thanks for taking the time too look at my hip file.
I am not sure if I understand you corretly but here are a couple of key elements that you might be missing :
- My main issue is the point of the string of the hat suddenly being attached to the arm in sim1.
- I acutally want to use the grid object as collision geometry to setup the constraint for the sring as I whant the rope to be attached to the hat and not the agent.
- I feel like the vellun agent unpack after ‘create_agent’ already unpack everything that might need to be unpacked.
- The cape indeed works fine even if you uncheck the ‘Use closest location on Primitive’ checkbox which why I believe the unpack for vellum node might be failing the retarget the constraint porperly.
Let me know if there is anything I am not understanding.
Technical Discussion » Vellum crowd attach constraint issue
- Edmond BG
- 45 posts
- Offline
Hello,
I am currently working on puting a vellum sim on top of agents after a vellum sim following this tutorial [www.sidefx.com].
At step 4 they showcase how it is possible to put attach constraint on the agent but puting down an attach to geometry node.
My issue comes when you do that and you don't want to use the ‘Use closest location on Primitive’ checkbox. If I don't use that it seems that that constraints after the post sim unpack are connected to the wront points of the agent.
I usually specify which points are connected to which points and agents usually have 1 or 3 primitives as polygons soups so it is an issue for me to only be qble to connect to primitives and not points.
I join my hip file in houdini 18.0.460 sim1 has the issue and sim2 and the checkbox enabled for reference.
If anyone can confirm that this might be a bug, I will send a message to support.
Any help appreciated.
Thx
I am currently working on puting a vellum sim on top of agents after a vellum sim following this tutorial [www.sidefx.com].
At step 4 they showcase how it is possible to put attach constraint on the agent but puting down an attach to geometry node.
My issue comes when you do that and you don't want to use the ‘Use closest location on Primitive’ checkbox. If I don't use that it seems that that constraints after the post sim unpack are connected to the wront points of the agent.
I usually specify which points are connected to which points and agents usually have 1 or 3 primitives as polygons soups so it is an issue for me to only be qble to connect to primitives and not points.
I join my hip file in houdini 18.0.460 sim1 has the issue and sim2 and the checkbox enabled for reference.
If anyone can confirm that this might be a bug, I will send a message to support.
Any help appreciated.
Thx
Technical Discussion » Vellum doesn't stabilize even after a lot of substep
- Edmond BG
- 45 posts
- Offline
Thanks for taking the time to look at my issue.
I have attached a new hip file that is set to Manual at frame 1 frame before the sim starts.
Thanks again.
I have attached a new hip file that is set to Manual at frame 1 frame before the sim starts.
Thanks again.
Technical Discussion » Vellum doesn't stabilize even after a lot of substep
- Edmond BG
- 45 posts
- Offline
Hello,
I have a setup of a rope hanging the rubber toy and for some reason the simulation does not come to a rest.
It seems that the back of head of the rubber toy keep pushing the string and it pushes the whole rubber toy backward.
I tried to play on drag, the energy loss ratio and even with 20 substep, 1500 iterations , 600 collision iteration the motion is the same.
See hipfile.
I know I could just stich the string to the neck but I was hoping to simulate the knot sliding around the rope till it reaches the neck.
Any help appreciated.
I have a setup of a rope hanging the rubber toy and for some reason the simulation does not come to a rest.
It seems that the back of head of the rubber toy keep pushing the string and it pushes the whole rubber toy backward.
I tried to play on drag, the energy loss ratio and even with 20 substep, 1500 iterations , 600 collision iteration the motion is the same.
See hipfile.
I know I could just stich the string to the neck but I was hoping to simulate the knot sliding around the rope till it reaches the neck.
Any help appreciated.
PDG/TOPs » PyTorch + PDG + Houdini
- Edmond BG
- 45 posts
- Offline
PDG/TOPs » PyTorch + PDG + Houdini
- Edmond BG
- 45 posts
- Offline
Yes, CanonBall900 and I have been working on getting ML into houdini.
It is pretty straight forward when you know what to do, I will put an example hip file that requires a lot of external library installation like pytorch for those interested and will send you a message if you want to do a one-on-one.
It is pretty straight forward when you know what to do, I will put an example hip file that requires a lot of external library installation like pytorch for those interested and will send you a message if you want to do a one-on-one.
PDG/TOPs » PyTorch + PDG + Houdini
- Edmond BG
- 45 posts
- Offline
Hey Francisco,
Welcome to the pytorch and Tops club.
I can not say for sure where the error comes from, it might be an houdini version issue. Are you using houdini python 3?
However I can explain the meaning of @pdg_result. The Top pipeline creates and update values like the path to the files that are stored in attributes. @pdg_result is one of theme and it is different for every wedge (green dot in Tops node).
We can see in your screenshot that the error occurs when the rop fetch is called and @pdg_result has the value “C:/Users/Francisco/Desktop/ml_examples_data/ml_examples_data-preparation/terrain_gen/usgs_data_process/ISGS_NED_13_n47ww144_GridFloat.zip”
So maybe houdini has some issue reading the file ? Or maybe the issue is that @pdg_result is pointing to a .zip that is then send to HF_visualize ? Have you installed all he hda that comes with the archive ?
Welcome to the pytorch and Tops club.
I can not say for sure where the error comes from, it might be an houdini version issue. Are you using houdini python 3?
However I can explain the meaning of @pdg_result. The Top pipeline creates and update values like the path to the files that are stored in attributes. @pdg_result is one of theme and it is different for every wedge (green dot in Tops node).
We can see in your screenshot that the error occurs when the rop fetch is called and @pdg_result has the value “C:/Users/Francisco/Desktop/ml_examples_data/ml_examples_data-preparation/terrain_gen/usgs_data_process/ISGS_NED_13_n47ww144_GridFloat.zip”
So maybe houdini has some issue reading the file ? Or maybe the issue is that @pdg_result is pointing to a .zip that is then send to HF_visualize ? Have you installed all he hda that comes with the archive ?
Technical Discussion » houdini, python3 and matplotlib
- Edmond BG
- 45 posts
- Offline
tshead
Try Toyplot, https://toyplot.readthedocs.io [toyplot.readthedocs.io] … it can send your plots to a web browser, and won’t interfere with the Houdini UI.
And yes, I admit it - I’m the original Toyplot developer.
Cheers,
Tim
I used toyplot over week end and it worked great, thanks a lot.
However it seems that calling it from inside a PDG prevent it from creating a window in my browser.
If you have a work around that, that would be great.
I put the hip file in attachment, I moved back to houdini python2 atm and call python3 like canonball900 showed me here [www.sidefx.com].
Thanks again.
PDG/TOPs » PyTorch + PDG + Houdini
- Edmond BG
- 45 posts
- Offline
Hey,
thanks to your help I manage to catch up and also encoutered issues using PDG and the houdini python3 release.
I tested PDG on a super simple setup (see attachment) in houdini python3 and there might be a python error that prevent PDG nodes from working properly even without calling pytorch.
The error is:
I tested the same setup in houdini python 2.7 and it worked fine, so I guess there is a reason why houdini python3 is not the ready for production.
Also I took at the demo you shared and I do not think they used pytorch inside of a PDG. As the title of the post indicate, they only used houdini for data preparation, aka generating the images to train their neural network. All the pytorch implementation was done outside of houdini.
I too wish to to do machine learning in houdini and will be digging the issue. Please keep me posted if you manage to get pytorch and PDG working.
thanks again.
thanks to your help I manage to catch up and also encoutered issues using PDG and the houdini python3 release.
I tested PDG on a super simple setup (see attachment) in houdini python3 and there might be a python error that prevent PDG nodes from working properly even without calling pytorch.
The error is:
can't concat str to bytes
Error running Python callback ‘onSchedule’:
Traceback (most recent call last):
File “C:/PROGRA~1/SIDEEF~1/HOUDIN~1.475/houdini/pdg/types\schedulers\local.py”, line 988, in onSchedule
job_env = self._generateEnvironment(work_item, cpu_slots)
File “C:/PROGRA~1/SIDEEF~1/HOUDIN~1.475/houdini/pdg/types\schedulers\local.py”, line 553, in _generateEnvironment
job_env = temp_dir + os.pathsep + pth
TypeError: can't concat str to bytes
I tested the same setup in houdini python 2.7 and it worked fine, so I guess there is a reason why houdini python3 is not the ready for production.
Also I took at the demo you shared and I do not think they used pytorch inside of a PDG. As the title of the post indicate, they only used houdini for data preparation, aka generating the images to train their neural network. All the pytorch implementation was done outside of houdini.
I too wish to to do machine learning in houdini and will be digging the issue. Please keep me posted if you manage to get pytorch and PDG working.
thanks again.
Edited by Edmond BG - May 24, 2020 23:34:06
Technical Discussion » houdini, python3 and matplotlib
- Edmond BG
- 45 posts
- Offline
Thanks both of you.
toyplot works great, congrats on being the founder of such an amazing tool.
toyplot works great, congrats on being the founder of such an amazing tool.
Technical Discussion » houdini, python3 and matplotlib
- Edmond BG
- 45 posts
- Offline
Hello,
I recently installed the python3 beta of houdini and I wanted to use pytorch and matplotlib.
After placing all the packages into /install/python/lib/python3.7/site-packages I can call both library without triggering any error.
However the tutorial I was following used matplotlib.pylot to show a graph (pictures). So I call pyplot.imshow() and pyplot.show() and the image shows up in a separate window so I can play around with it.
The issue is that houdini freeze as soon as the figure window opens and even after I close it. My only option is use grep and kill to restart houdini.
Is anyone having the same issue ?
Any help appreciated, thanks.
I recently installed the python3 beta of houdini and I wanted to use pytorch and matplotlib.
After placing all the packages into /install/python/lib/python3.7/site-packages I can call both library without triggering any error.
However the tutorial I was following used matplotlib.pylot to show a graph (pictures). So I call pyplot.imshow() and pyplot.show() and the image shows up in a separate window so I can play around with it.
The issue is that houdini freeze as soon as the figure window opens and even after I close it. My only option is use grep and kill to restart houdini.
Is anyone having the same issue ?
Any help appreciated, thanks.
Technical Discussion » Houdini and PyTorch
- Edmond BG
- 45 posts
- Offline
Thanks ! I just got it to work by copying the files to /install/python/lib/python3.7/site-packages.
I will keep you posted if I get more success working with PDG but I might take a bit of time to get here.
Again thanks a lot.
I will keep you posted if I get more success working with PDG but I might take a bit of time to get here.
Again thanks a lot.
Technical Discussion » Houdini and PyTorch
- Edmond BG
- 45 posts
- Offline
Hey do you mind going into details ? I understand that I have to put the pytorch lib inside my houdini python folder but could be precise regarding where I can find the packaged lib and what should I put into /bin /include /lib and /share ?
Thanks.
Thanks.
Technical Discussion » Houdini unable to load HFS OpenCL Platform Win10
- Edmond BG
- 45 posts
- Offline
Hello,
I have Houdini 18.0.391h currently installed and I wanted to use OpenCl to speed up vellum among other things.
The issue is that this message appears when I load a vellum project :
Meanwhile my GPU is learly listed as my OpenCL platform and appear in the houdini detail panel :
And when I run a vellum sim I can clearly see my CPU is used but not my GPU.
I should mention that I have the lastest Nvidia driver 442.19 installed with CUDA 10.2.
Any help is welcome.
Thanks.
I have Houdini 18.0.391h currently installed and I wanted to use OpenCl to speed up vellum among other things.
The issue is that this message appears when I load a vellum project :
Unable to load HFS OpenCL platform.
Meanwhile my GPU is learly listed as my OpenCL platform and appear in the houdini detail panel :
OpenCL Platform NVIDIA CUDA
Platform Vendor NVIDIA Corporation
Platform Version OpenCL 1.2 CUDA 10.2.131
OpenCL Device GeForce GTX 1060
OpenCL Type GPU
Device Version OpenCL 1.2 CUDA
Frequency 1670 MHz
Compute Units 10
Device Address Bits 64
Global Memory 6144 MB
Max Allocation 1536 MB
Global Cache 480 KB
Max Constant Args 9
Max Constant Size 64 KB
Local Mem Size 48 KB
2D Image Support 16384x32768
3D Image Support 16384x16384x16384
And when I run a vellum sim I can clearly see my CPU is used but not my GPU.
I should mention that I have the lastest Nvidia driver 442.19 installed with CUDA 10.2.
Any help is welcome.
Thanks.
Edited by Edmond BG - March 18, 2020 22:39:38
Houdini Indie and Apprentice » Replicate animation
- Edmond BG
- 45 posts
- Offline
Hello,
I need some help replicating an animation.
I have this animation in one object (Object Animation in .hipnc attached) that I want to trigger in multiple location and at different time.
I tried using another popnetwork and using an object merge>copy to points but the animation duplicated (object Multiple_Animation in the .hipnc) are synchronized.
I read about instancing but I also would like my animation to be played on different seed, with different color and of different length.
What are the correct steps to make this process fully procedural ?
Thanks.
I need some help replicating an animation.
I have this animation in one object (Object Animation in .hipnc attached) that I want to trigger in multiple location and at different time.
I tried using another popnetwork and using an object merge>copy to points but the animation duplicated (object Multiple_Animation in the .hipnc) are synchronized.
I read about instancing but I also would like my animation to be played on different seed, with different color and of different length.
What are the correct steps to make this process fully procedural ?
Thanks.
Houdini Indie and Apprentice » Need help for pop forces in bullet solver weird behavior
- Edmond BG
- 45 posts
- Offline
tamte
your sim is probably hitting sleeping threshold, you can verify that by tutning on sleeping visualizer
That's it ! Thanks a lot, I thought the sim was freezing but it was just falling a sleep !
Thanks tamte and vusta !
Houdini Indie and Apprentice » Need help for pop forces in bullet solver weird behavior
- Edmond BG
- 45 posts
- Offline
Thanks for helping.
I too experimented a bit and I reworked the chop network because it might have been linked to the way the node is activated but even here the simulation work or don't work if you simply activate or bypass a delay node.
For the moment I am concluding that it is probably a bug and report back if I find anything new.
Thanks again.
I too experimented a bit and I reworked the chop network because it might have been linked to the way the node is activated but even here the simulation work or don't work if you simply activate or bypass a delay node.
For the moment I am concluding that it is probably a bug and report back if I find anything new.
Thanks again.
-
- Quick Links