JoshJ Posted August 27, 2010 Share Posted August 27, 2010 I have a camera frustum based volume, and I have it set up so that the camera can move through the scene while noise that's being multiplied in the volume stays static in world space. That's working very well. Now I want to scatter points onto the volume and I'd like those points to stay static in world space also. However, when I use the scatter sop to do this, the points are rescattered every frame since the volume changes, so it appears that the point positions change. Any ideas on how I could keep the points static throughout the animation, while having them be dynamically created and removed when outside of the volume bounds? I've already ruled out creating the points beforehand then culling them, as I need to create a massive amount of points, so this really needs to be some kind of dynamic process that updates itself throughout the animation. Quote Link to comment Share on other sites More sharing options...
ikarus Posted August 28, 2010 Share Posted August 28, 2010 probably would be best to generate them once for the whole thing then have it run through a ray sop to see whats inside the bounds of the volume which is animated Quote Link to comment Share on other sites More sharing options...
pclaes Posted August 29, 2010 Share Posted August 29, 2010 I think you can't really get around making the points first and culling them afterward. Rather than a ray sop I would use a lookup in a vopsop against an SDF/volume sample. What I tend to do in such a case is split your volume in subvolumes and do the culling in separate chunks on the farm. If you are using it to go straight to rendering, you could do it as a custom VRAYprocedural. Quote Link to comment Share on other sites More sharing options...
Macha Posted August 30, 2010 Share Posted August 30, 2010 (edited) Perhaps you could assign a point position by feeding a random generating function with an attribute. Something like setting the P attribute to rand($PT), rand($PT*2), rand($PT*3). That way each point would have a unique random position. BTW, how many points are we talking about? Grouping/deleting by sdf volumes is very fast. Edited August 30, 2010 by Macha Quote Link to comment Share on other sites More sharing options...
pclaes Posted August 30, 2010 Share Posted August 30, 2010 BTW, how many points are we talking about? Grouping/deleting by sdf volumes is very fast. It is, because it is as a vopsop, the part that is slow is the sdf generation though. And as he stated it is required at every frame. Still you could generate several highly detailed sequences of sdf in paralell if you have multiple cores you could run them in the background - once cached to disk they should be a lot faster. If it is just one sequence, better not compress them as that makes it slower. Quote Link to comment Share on other sites More sharing options...
JoshJ Posted August 30, 2010 Author Share Posted August 30, 2010 (edited) Thanks for all your suggestions guys... I ended up generating the points in cubes of like 25 to 50 square units, based on the distance from the camera's path... testing the distance from the camera and whether it will be in view, then writing them out to disk, then another set of nodes to read it back in and cull the area I want based on the camera's position. I wanted to generate them based on some particular vop sop noise that I already had in the frustum volume. Total amount of points would probably be 100 million or so, maybe more depending on the distance the camera has to travel (which was yet undetermined)... so I wanted to figure out some way to break it up and have a nice procedural system convenient for scaling up... I am really interested in the custom VRAY procedural approach... just haven't had time to look into it, if there's any tips/direction you can point me to, I'd highly appreciate it. Edited August 30, 2010 by joshjordan Quote Link to comment Share on other sites More sharing options...
pclaes Posted August 31, 2010 Share Posted August 31, 2010 I am really interested in the custom VRAY procedural approach... just haven't had time to look into it, if there's any tips/direction you can point me to, I'd highly appreciate it. keep an eye on this: http://www.peterclaes.be/blog/ I'm going to be updating it soon. At the moment just basic info, but I managed to get up to 160 million polygons rendering with my custom instancer. In other versions I have built in pointcloud and Level of Detail support. But don't expect anything complete very soon, I want to share information properly, step by step and thoroughly. That does take a bit of time. I will try and put the source code of my instancer online with the next posts. That way you can go through it yourself if you want to have a look at some of it. Quote Link to comment Share on other sites More sharing options...
JoshJ Posted August 31, 2010 Author Share Posted August 31, 2010 looks excellent, looking forward to seeing it in detail. Thanks! Quote Link to comment Share on other sites More sharing options...
cowboy71 Posted July 2, 2014 Share Posted July 2, 2014 (edited) hey there, So I came across this post while having a similar problem. Geometry into Vopsop, displace with noise then animate. Attached another vopsop for some color noise then used a scatter to copy some spheres onto the deforming geo. I'm noticing my scatter points jumping all around. I would like them to stay static so I could animate the noise within the vopsop to move them around. Not quite a newbie to Houdini, but not understanding the above solutions, if they are correct for my issue. Thanks Edited July 2, 2014 by cowboy71 Quote Link to comment Share on other sites More sharing options...
Recommended Posts
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.