Atom Posted December 16, 2014 Share Posted December 16, 2014 (edited) I have a small town setup as RDB static collision objects for a FLIP tank. The sim seems to be working ok but I am wondering about the large sphere size I am seeing in the viewport. I am following the flooded hallway tutorial on youtube. In the end he ends up using VDB to skin the fluid so I am wondering what the sphere size to VDB skin relation ship is? Do I need to make the spheres smaller? If so, what parameter controls the sphere size? Edited December 16, 2014 by Atom Quote Link to comment Share on other sites More sharing options...
kleer001 Posted December 16, 2014 Share Posted December 16, 2014 You'll want to scale up everything else but the fluid, by 10 or so. That looks like the default particle separation of 0.1 units. That number is in your Autodopnetwork|FlipFluidObject > particle separation. The reason you don't want to scale down your particles is that below 0.03 you get a bit of instability. Quote Link to comment Share on other sites More sharing options...
lukeiamyourfather Posted December 16, 2014 Share Posted December 16, 2014 (edited) The sphere size is based on the particle separation of the FLIP object. Based on the relative size of the particles there it looks like you need smaller particle separation (high resolution) to get reasonable results. Work at true to life scale when possible as it will achieve the most realistic motion and timing. Fudge the scale only when the scale itself is a problem for extremely large simulations and extremely small simulations. If you just want small points for the purpose of visualization only then change the "visprim" parameter on the FLIP object to particles instead of sprites. It's under Guides>Particles>Visualization and the default is sprites. Edited December 16, 2014 by lukeiamyourfather Quote Link to comment Share on other sites More sharing options...
Atom Posted December 16, 2014 Author Share Posted December 16, 2014 Thanks for the tips. I did scale every thing down by 100. This allows my system to actually function. Using a box emitter with a size of (0.5,75,150) produced too many particles even on frame #1. Is there a way to limit the number of particles or is that just a function of the volume of the emitter? Quote Link to comment Share on other sites More sharing options...
lukeiamyourfather Posted December 16, 2014 Share Posted December 16, 2014 One Houdini unit equals one meter. How big is your city supposed to be in life and how many Houdini units is it? The number of particles produced depends on the velocity of the emitter and the particle separation. If there are too many on the first frame it's probably because of the particle separation (too small). When you say too many, how many is that, why is it too many? Quote Link to comment Share on other sites More sharing options...
Atom Posted December 17, 2014 Author Share Posted December 17, 2014 (edited) I don't have an exact count of particles, I am not sure how to retireve/set that number? It is too many because my system quits responding and I have to simply abandon the attempt. I am a human, not a robot so I have limited time to devote to the experiment. I know patience is a virtue but using FLIP on my crappy old computer feels like a waste of time. The author of the Flooded Hallway Tutorial claims to have done the entire sim on his laptop, however. My computer is faster than most dual core laptops. If a unit is one of those little squares than my city, in it's default form, is quite big, several hundred square meters. That is one reason why I scaled everything down by 100. I may try just scaling by 10 and see if that offers any better results. I did get a single frame out of my computer today. Yeah, at this rate I'll be done in two short years. In this frame the fluid looks very viscous I am really shooting for a water like look. This is jsut a render of the particles. I am not sure what produces the surface though? Does that just happen auto-magically at render time by Mantra? I know you can take extra steps and create a surface, but if those steps are not needed why do I need to do them? Is that why I just have goop? Edited December 17, 2014 by Atom Quote Link to comment Share on other sites More sharing options...
Guest tar Posted December 17, 2014 Share Posted December 17, 2014 Pretty simple - just put more Ram into your computer Quote Link to comment Share on other sites More sharing options...
Atom Posted December 17, 2014 Author Share Posted December 17, 2014 (edited) Mo money, mo money I just looked at my cache files and they are 50MB per frame. Is that typical, low, high...? Can anyone explain the relationship between Division size and Particle Size? The help says that they should be somehow synchronized. I see Division Size in the Particle Source node and it is colored green indicating that it is controlled by an expression of something. If I change a green parameter am I really chaning it or wil it just jump back to a driven value when the frame changes? In the FlipTank of the AutoDOP network I see Particle Separation and Particle Radius Scale. Is that the particle size that the help file is mentioning or is there some other setting in another node that controls particle size? It is hard as a new user to read between the lines if the help files do not use the same terms as the labels in the nodes...uhg (this could be easier) Edited December 17, 2014 by Atom Quote Link to comment Share on other sites More sharing options...
Guest tar Posted December 17, 2014 Share Posted December 17, 2014 Sorry no time to go through all the points currently. In general you should run experiments on small test files, otherwise it'll take a long time to learn everything. 50MB files sizes for the Dop cache or .bgeo files? Seems normal size. If you only have 8GB of Ram it's going to be a great challenge to get detail need to stop the 'goopyness'. You will feel much happier with 24 or 32+ GB Ram. Also Linux and OsX seem to handle Ram fragmentation better in general. Quote Link to comment Share on other sites More sharing options...
Mandrake0 Posted December 17, 2014 Share Posted December 17, 2014 when it's slow check the static objects when they are to complex it takes also quiet some time in the sim. use the performace monitor to optimize and listen to that what marty said. :-) Quote Link to comment Share on other sites More sharing options...
Atom Posted December 17, 2014 Author Share Posted December 17, 2014 Thanks again, I realize my machine is under powered. I did discover the Cache setting in the root of the AutoDOP network. I set that to 0 instead of 5000 and that has helped a bit. I followed that up with a File node at the end of the AutoDOP in Write mode. So essentially I am instructing Houdini not to bother storing anything in memory, if possible, and use the disk as memory. My Windows box is running DDR2 @ 800Mhz. I tried buying memory on eBay, last year, three times. And each time was a bust, the memory I purchased would not work in my motherboard (totally dead and would not even turn on when installed) and I had to send it back. My computer is a frankenstein I assembled a few years back, but to obtain the higher memory modules I am typically looking at server style memory which does not seem to work in my motherboard. So I am really looking at a new motherboard, memory and CPU which sets the price at more than just starting over with a newer barebones system. Financially I need to work with the hardware I have, for the time being. I also have a MacMini with 16GB but it is running in Apprentice mode so I can't get final renders out of it without the watermark. I also noticed that there are new graphics cards requirements for release 14 and wonder if the mini is going to remain compatible at all? Quote Link to comment Share on other sites More sharing options...
lukeiamyourfather Posted December 17, 2014 Share Posted December 17, 2014 If a unit is one of those little squares than my city, in it's default form, is quite big, several hundred square meters. That is one reason why I scaled everything down by 100. I may try just scaling by 10 and see if that offers any better results. Scale does not determine the number of particles. The point separation on the FLIP object determines this. Ideally work with the objects at true to life scale and adjust the point separation on the FLIP object to give you the number of particles you want (or the hardware can handle). This is critical to get realistic results because if the scale is off the motion won't feel right. If the scale is too large then things will feel like they are moving super slow and if things are too small then the motion will be too fast. Our brains have a way of pointing out this kind of thing when we see it and it contributes to something looking "real" or "fake" on screen. If you're really interested in pursuing visual effects and computer graphics it helps a lot to have modern hardware. I'm not saying go out and buy a $10,000 workstation but spending $500 upgrading your existing machine could go a long way like new processor, motherboard and memory. Having said that the machine you're using now is significantly faster than what I used to learn... Quote Link to comment Share on other sites More sharing options...
TheDude Posted December 17, 2014 Share Posted December 17, 2014 Hi Atom, I read your earlier posts and I think the biggest problem here is optimization -- if you're emitter is (0.5,75,150), then this is a tidal wave. Think about a football pitch (soccer or american football), your emitter is 1.5x as long as that in Z, and 3/4 of that in y -- That's massive!! Looking at your screenshots, i'd estimate the first picture is probably about 40m long (that door on the right should be about 2m tall) -- You're machine is admittedly garbage by today's standards, but you'll still be able to get reasonable results with it. A lot of detail comes from simulations you make driven on the flip particles anyway, so no need to waste too much time on this part anyway From what I understand, Particle Seperation is the unit distance per particle in the FLIP. With this in mind, if your emitter is a box 1x1x1, and you have particle seperation at .2, 1 / .2 = 5 -- then you have 5 particles that can fit in each direction, so 5x5x5 is 125 particles. So applying this, your little path your emitting from looks wide enough for like 2 people to walk by but probably not a car. I'm guessing you could make the emitter there be about 2 metres wide, maybe .75 deep and 2 metres tall. Now you can figure out how many particles are going to be in this emitter alone on frame one. If you want 500 particles to be in your emission box then (.75/x)(2/x)(2/x) = 500 3/x^3 = 500 x^3 = .006 x = .1817 << so that's what your particle seperation needs to be IF particle seperation is the unit distance between the flip particles, which im pretty sure it is, but unfortunately it doesn't clearly say in the documentation. https://www.sidefx.com/docs/houdini13.0/nodes/dop/flipobject But anyway, make sure to focus on your scene size, it will affect everything in the simulation! Good luck! 1 Quote Link to comment Share on other sites More sharing options...
Guest tar Posted December 17, 2014 Share Posted December 17, 2014 If you do run into some money, possible a grab Houdini Engine license too, so you can sim on the MacMini as it has more Ram. Quote Link to comment Share on other sites More sharing options...
Atom Posted December 19, 2014 Author Share Posted December 19, 2014 Thanks again for the advice and explanations. I think I am going to give up on this simulation and focus on some more simple ones first. 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.