I was not able to find the agent catalog, but for some reason I was able to find a expression in the documentations for getting the length of clips:
f@statelen_1= agentcliplength(geoself(), 0, “idle_5”);
http://www.sidefx.com/docs/houdini15.0/vex/functions/agentcliplength [sidefx.com]
I was looking in the documentations for the agent catalog and for some reason I got to this page, so it kinda works, thanks!
Found 231 posts.
Search results Show results as topic list.
Houdini Learning Materials » Trigger based on crowd state end
- vleermeneer
- 245 posts
- Offline
Houdini Learning Materials » Trigger based on crowd state end
- vleermeneer
- 245 posts
- Offline
Thanks I will try this when I get home, I will let you know what happens.
Also, do you happen to know if the Slug Fest Crowd Simulation training on cmiVFX also includes information about fuzzy logic?
Thanks for sharing this tutorial, somehow I was not aware of this one from Cmivfx I've got 60% off so I might buy this one when I get home..
Also, do you happen to know if the Slug Fest Crowd Simulation training on cmiVFX also includes information about fuzzy logic?
Thanks for sharing this tutorial, somehow I was not aware of this one from Cmivfx I've got 60% off so I might buy this one when I get home..
Houdini Learning Materials » Trigger based on crowd state end
- vleermeneer
- 245 posts
- Offline
Is it possible to procedurally call out when a current state has ended?
Because now I am defining when the state duration reaches a certain time, the agent will switch to another animation.
Or am I missing something?
Because now I am defining when the state duration reaches a certain time, the agent will switch to another animation.
Or am I missing something?
Houdini Indie and Apprentice » Offset crowd uv's procedurally
- vleermeneer
- 245 posts
- Offline
Oh damn!
I wish I knew this earlier because I allready created all my geometry, my current setup has four agents instead now of just one, so now I have 84 in total instead of 21. I will have to try out the material stylesheet for my next crowds.
I wish I knew this earlier because I allready created all my geometry, my current setup has four agents instead now of just one, so now I have 84 in total instead of 21. I will have to try out the material stylesheet for my next crowds.
Houdini Indie and Apprentice » Offset crowd uv's procedurally
- vleermeneer
- 245 posts
- Offline
Hi guys,
I would like to offset my uv's for my crowd procedurally, is there a efficient way to do this?
My uv's are set up stacked on top of eachother (see attachment), so I just need to offset the uv's using uv transform. Is this ever possible to randomize per crowd procedurally? Or should I just create four agents for something like this?
I currently have 21 models with 4 textures per file, adjusting the uv's would give me 84 different characters using just 21 maps instead of 84 maps, thats why I stacked them..
I would like to offset my uv's for my crowd procedurally, is there a efficient way to do this?
My uv's are set up stacked on top of eachother (see attachment), so I just need to offset the uv's using uv transform. Is this ever possible to randomize per crowd procedurally? Or should I just create four agents for something like this?
I currently have 21 models with 4 textures per file, adjusting the uv's would give me 84 different characters using just 21 maps instead of 84 maps, thats why I stacked them..
Technical Discussion » htoa in render view
- vleermeneer
- 245 posts
- Offline
This is actually my first time using Arnold in Houdini!
At school we have many Arnold nodes at our renderfarm, and I just found out htoa use the same license as mtoa so I just installed htoa.
I really like arnold! It is extremely fast, GI is always turned on so you instantly have realistic results. Compared to Mantra the workflow feels the same actually, the shaders have the same settings so its easy to switch to Arnold.
You can try Arnold for free by the way!
On their page you can download the plugin, if you dont have a license you will get some watermarks but that should be ok for trying out..
At school we have many Arnold nodes at our renderfarm, and I just found out htoa use the same license as mtoa so I just installed htoa.
I really like arnold! It is extremely fast, GI is always turned on so you instantly have realistic results. Compared to Mantra the workflow feels the same actually, the shaders have the same settings so its easy to switch to Arnold.
You can try Arnold for free by the way!
On their page you can download the plugin, if you dont have a license you will get some watermarks but that should be ok for trying out..
Technical Discussion » htoa in render view
- vleermeneer
- 245 posts
- Offline
I found a solution:
Delete the cam.py node in the arnold scripts/obj folder..
C:\solidangle\htoa\htoa-1.6.1_r1459_houdini-14.0.395\scripts\obj
Then add the attributes manually by using the “add arnold property node” and everything should work fine now.. im now asking arnold developer why this is the case.
Delete the cam.py node in the arnold scripts/obj folder..
C:\solidangle\htoa\htoa-1.6.1_r1459_houdini-14.0.395\scripts\obj
Then add the attributes manually by using the “add arnold property node” and everything should work fine now.. im now asking arnold developer why this is the case.
Technical Discussion » htoa in render view
- vleermeneer
- 245 posts
- Offline
Hi guys,
I'm having some problems using htoa in houdini.
For some reason I am not able to render arnold in the render view.
But when I press the render node my lefthand side of the screen (inside my viewport) arnold will render succesfully to mplay.. but when I am looking through the camera and press the same button Arnold will not render anymore.
This is the error I receive:
00:00:00 319MB ERROR | Unsupported camera projection: None
00:00:00 24MB | Received: log_verbosity: 7942
00:00:00 24MB | Received: log_console_enable: 1
00:00:00 35MB ERROR | there is no camera!
I'm having some problems using htoa in houdini.
For some reason I am not able to render arnold in the render view.
But when I press the render node my lefthand side of the screen (inside my viewport) arnold will render succesfully to mplay.. but when I am looking through the camera and press the same button Arnold will not render anymore.
This is the error I receive:
00:00:00 319MB ERROR | Unsupported camera projection: None
00:00:00 24MB | Received: log_verbosity: 7942
00:00:00 24MB | Received: log_console_enable: 1
00:00:00 35MB ERROR | there is no camera!
Houdini Indie and Apprentice » Getting positive and negative angles between two vectors
- vleermeneer
- 245 posts
- Offline
Thanks this is working, I wasnt aware that the vectors pointing downwards and upwards actually meant something.
Do you also happen to know why/how this is happening? I would like to understand what Houdini actually is doing to get this result.
By the way; I used the cross product between forward and backward curve direction instead, because these were already calculated and basically the same result.
Do you also happen to know why/how this is happening? I would like to understand what Houdini actually is doing to get this result.
By the way; I used the cross product between forward and backward curve direction instead, because these were already calculated and basically the same result.
Houdini Indie and Apprentice » Getting positive and negative angles between two vectors
- vleermeneer
- 245 posts
- Offline
I have two vectors, a red one and a blue one as you can see in my screenshot.
When I use a dot product and arc cosine I am getting the angles, but they're all positive values. In the image you can obviously see the red normal is sometimes on the left side and sometimes on the right side of the blue normal (which is correct in my case).
My final goal is I want to check when corners are convex and when they are concave. So by checking the angles concave should be -45 degrees and concave should be 45 degrees, but all I get is the difference between the two vectors (which is always 45 degrees).
Am I missing something?
When I use a dot product and arc cosine I am getting the angles, but they're all positive values. In the image you can obviously see the red normal is sometimes on the left side and sometimes on the right side of the blue normal (which is correct in my case).
My final goal is I want to check when corners are convex and when they are concave. So by checking the angles concave should be -45 degrees and concave should be 45 degrees, but all I get is the difference between the two vectors (which is always 45 degrees).
Am I missing something?
Technical Discussion » More efficient way of time offsetting copy's
- vleermeneer
- 245 posts
- Offline
Somehow the caching actually seems to be working!
I'm going to try and expand this technique and test it out over a render farm perhaps. I will get back to you on this, thanks!
I'm going to try and expand this technique and test it out over a render farm perhaps. I will get back to you on this, thanks!
Technical Discussion » (solved)when FLIP collide with RBD, it explosion
- vleermeneer
- 245 posts
- Offline
Try change the feedback scale, by default its set to 0 meaning it cannot push other objects.
flipsolver node: volume motion –> solver –> feedback scale
Dont overdo this setting, 1 means 100%, 0 means 0%
Ofcourse you can enter 5000% as you like
flipsolver node: volume motion –> solver –> feedback scale
Dont overdo this setting, 1 means 100%, 0 means 0%
Ofcourse you can enter 5000% as you like
Technical Discussion » More efficient way of time offsetting copy's
- vleermeneer
- 245 posts
- Offline
Your file instantly crashes somehow (multiple times), I will try again tomorrow after work!
Technical Discussion » More efficient way of time offsetting copy's
- vleermeneer
- 245 posts
- Offline
Thanks for your response mzigaib.
It's actually not about performance in the viewport, I'm more bother about the cooking time. This is a low demanding scene, image I have 1.000.000 animated objects. Copy stamping will fail at huge values with huge geometry.
The only way for me to get this to run stable is to export some objects with different time offsets and copy them seperately in seperate copy sops.
Time offsettings would be too slow with copy stamping, I've also tried this with the instance node. First export different variations of a model, create an attribute to switch the file paths at render time. But when I reach extreme amounts mantra will eventually choke on a 24GB system.
Even when I export mantra archive with ifd material and use delayed load procedural.
I think the only fast way is to have seperate instance nodes or seperate copy nodes, I was thinking about writing a simple python script to do this.
It's actually not about performance in the viewport, I'm more bother about the cooking time. This is a low demanding scene, image I have 1.000.000 animated objects. Copy stamping will fail at huge values with huge geometry.
The only way for me to get this to run stable is to export some objects with different time offsets and copy them seperately in seperate copy sops.
Time offsettings would be too slow with copy stamping, I've also tried this with the instance node. First export different variations of a model, create an attribute to switch the file paths at render time. But when I reach extreme amounts mantra will eventually choke on a 24GB system.
Even when I export mantra archive with ifd material and use delayed load procedural.
I think the only fast way is to have seperate instance nodes or seperate copy nodes, I was thinking about writing a simple python script to do this.
Technical Discussion » More efficient way of time offsetting copy's
- vleermeneer
- 245 posts
- Offline
I always use pack before copying but this is not exactly what I meant.
I made a quick file to show you what I mean.
In this file you can see three different scenarios:
- first one is regular copy stamping for random size and rotation
- second one is the same size and rotation but this is done differently (look at huge performance increase)
- the third one is I only stamped the timeoffset (look at decrease in performance again)
I made a quick file to show you what I mean.
In this file you can see three different scenarios:
- first one is regular copy stamping for random size and rotation
- second one is the same size and rotation but this is done differently (look at huge performance increase)
- the third one is I only stamped the timeoffset (look at decrease in performance again)
Technical Discussion » More efficient way of time offsetting copy's
- vleermeneer
- 245 posts
- Offline
First I wish you all a merry Christmas!
I have one loopable model instanced across a surface and I want to offset the animation without loss in performance.
When using copy stamp my performace decreased drastically.
For full point instancing I cannot use instancepoint() because it only works on object level.
I wonder if someone has an efficient technique this.
Cheers
I have one loopable model instanced across a surface and I want to offset the animation without loss in performance.
When using copy stamp my performace decreased drastically.
For full point instancing I cannot use instancepoint() because it only works on object level.
I wonder if someone has an efficient technique this.
Cheers
Houdini Indie and Apprentice » Clean Intersecting Paths
- vleermeneer
- 245 posts
- Offline
Houdini Indie and Apprentice » Clean Intersecting Paths
- vleermeneer
- 245 posts
- Offline
Hey man, sorry for my extremely late response!
I had been very busy lately..
Anyways, I had a few days off so I decided to tackle this issue once more and it is a but more rock solid now! Still not perfect, dont do funny weird stuff.. :wink:
I've also supplied some texxtures to this system.
The file is a bit messy and very complex so beware..
I had been very busy lately..
Anyways, I had a few days off so I decided to tackle this issue once more and it is a but more rock solid now! Still not perfect, dont do funny weird stuff.. :wink:
I've also supplied some texxtures to this system.
The file is a bit messy and very complex so beware..
Houdini Lounge » Particles are flipping direction (just like gimbal lock)
- vleermeneer
- 245 posts
- Offline
I have created a flocking simulation using POPs in DOPs and when I instance my models on my particles, at some point they suddenly flip 180 degrees..
I've tried alot to fix this.
Using vopsop comparing directions a few frames apart and trying to average these out. (might be incorrect the way I did this)
Tried using Slerp in vopsop, also no succes.. (also might be incorrect the way I did this)
Assigning up vector to be always up also does not work.
Anyone else have an idea how to prevent the flipping effect?
Maybe a simple vopsop solution?
Too bad I cant share the scene file because of NDA
I've tried alot to fix this.
Using vopsop comparing directions a few frames apart and trying to average these out. (might be incorrect the way I did this)
Tried using Slerp in vopsop, also no succes.. (also might be incorrect the way I did this)
Assigning up vector to be always up also does not work.
Anyone else have an idea how to prevent the flipping effect?
Maybe a simple vopsop solution?
Too bad I cant share the scene file because of NDA
Houdini Indie and Apprentice » Merge Mantra Nodes To EXR
- vleermeneer
- 245 posts
- Offline
Oh wow thats really easy!
Thanks!
Now I only have one problem in nuke, it hides the alpha channel somehow.. I cant see it with a shuffle node, however I can edit my shuffle and add it manually.
Thanks!
Now I only have one problem in nuke, it hides the alpha channel somehow.. I cant see it with a shuffle node, however I can edit my shuffle and add it manually.
-
- Quick Links