itriix Posted August 26, 2009 Share Posted August 26, 2009 Hi everyone, i'm just curious what the difference is between a couple different dop sims examples without sounding too "beginner"... i'm just curious... if my default scene setting scale is 1 (m) unit length and 1 (Kg) unit mass. If i create a default sphere, click "flames", and then begin simming my lo-res sim... without adjusting any values.. how does this compare to. creating a default sphere, and scaling it down to say .2 in size, clicking "flames" and doing a sim from there? the "voxel grid dimensions" will be smaller, and the "voxel divisions" will be the same. But with that smaller "dimension" grid size, and same "voxel divisions", that means MORE information can be captured than if you simmed with the default sphere of 1. The voxel grid dimensions for that larger sphere would be bigger, but the voxel divisions would be the same, therefore those voxels would cover a larger amount of area, and therefore "lose" more detail than the small one. Now i'm quite sure part of the answers will be that, *it depends* on what i'm trying to simulate... because a smaller sphere, would mean a "smaller" object... But my main point i'm getting at is, is this a common approach? would it be better to "scale" objects down, in order to get smaller dimension grids, with the same voxel counts in order to get "more information" essentially crammed in? And then just try and "tweak" parameters to make the dynamics "match" whatever size object your going for? I might just be confusing myself or all of you reading this. but really i'm just trying to figure out what most people DO when they begin a sim of any kind in dops. for really any kind of example. i'd like to know the approach, and workflow, for say a "real world" example... and also, just for simpler examples, such as, testing and playing around... any information to help me develop faster my workflow and ways of approaching this stuff would be awesome for me. I appreciate it, and hope that wasn't too vague. Just for some reason, I keep thinking, I should scale things down in order to "fit more"... this might be wrong though. so confirmation and explanation would be fantastic! thank you all so much for any responses. thanks cheers, Jonathan Quote Link to comment Share on other sites More sharing options...
Fomal Posted August 26, 2009 Share Posted August 26, 2009 Hi, The scale of your object or simulation does influence the behaviour a lot, and it's hard to guess the right size. Keeping the 1m per unit rule helps but isn't always applicable. Regarding the resolution. I think you're looking at it wrong. You can't say, the smaller the size of my container, the higher the resolution of my sim. By saying a resolution of 40 over the max axis, Houdini finds the largest scale of a bounding axis and computes the container from there (resulting in 40x40x40 with a regular cube). Scaling the box up or down does not change the voxel count, so the resolution stays the same. If what you're saying is true, you might as well move your camera back. A good trick to use is to calculate the voxel resolution for every unit. And scale this by the size of your actual container over a certain axis. This way it's easy to use the gas resize and still be able to slice your sim over the farm (keeping the max computed bounding box as the up res container). Say you have a box of 4x2x13. Specify x as you division method and multiply the number of uniform divisions (the actual resolution of voxels per unit) with 13 (your x axis size). Now, when you scale the box the resolution will be the same, as long as you multiply your unit resolution with the actual size of the specified axis of the container Quote Link to comment Share on other sites More sharing options...
itriix Posted August 28, 2009 Author Share Posted August 28, 2009 Hello and thanks for the suggestion. I have a couple questions. First off, yes I was aware that scaling the box up and down does not change the voxel count, but I was referring to the "area" that the voxel would make up. Here is my example: Say container A is 10, 20, 10 with 50 uniform voxel divisions... Say container B is 1, 2, 1 with 50 uniform voxel divisions... Thererfore, the "area" of a voxel in container A is a lot larger than the "area" of a voxel in container B. So then, that means that a lot more "data" can be lost inside that "larger area" voxel... in container A... Whereas, since the area of the voxel in container B is a lot smaller, that means it would be able to record "finer details"....? Does that make sense? Once again, I admit my understanding of all this is limited but I just keep thinking about this instead. NEXT: Say you have a box of 4x2x13. Specify x as you division method and multiply the number of uniform divisions (the actual resolution of voxels per unit) with 13 (your x axis size). Now, when you scale the box the resolution will be the same, as long as you multiply your unit resolution with the actual size of the specified axis of the container Let me see if I am following this correctly... if there is a container 4x2x13... and say uniform divisions are at 50... are you saying to maybe put in an expression in the uniform divisions, that just says: 50 x 13...? which would give me 650 for the uniform divisions? that seems HUGE... i must be misinterpreting this i think could you please clarify for me more? thanks so much again! Quote Link to comment Share on other sites More sharing options...
essencevfx Posted August 28, 2009 Share Posted August 28, 2009 (edited) the 'area' of the voxel that you're speaking of is relative to the scale of the container , so best just to work in real world space (more or less), don't leave it up to houdini to decide that for you. If you want a ball to light on fire, figure out how big your ball is in units, and build your sim/scene/container around that. This way the voxel scale will be accurate to the effect your trying to create. I think a default sphere ( that's what you mentioned you were working with ?) is like 5 units, so 5 meters ...thats a damn big sphere Edited August 28, 2009 by essencevfx Quote Link to comment Share on other sites More sharing options...
Marc Posted August 30, 2009 Share Posted August 30, 2009 Simming things at real world scale is definitely the way to go since all forces will be working at that level. Scaling it down to .2 of it's original size will throw off all the force calculations and make it difficult to predict what was going to happen. Having said that however, simming things at large scales in Houdini is a pain in the ass. So for example, if you have an airplane falling from the sky trailing smoke and fire then simming at real world scale is going to be difficult at best and completely fail at worst. So.. 'it depends on what you're simming' I suppose . As with most things in Houdini, there's no right or wrong answer (...er.. that I know of ), but some will give better results than others. So the good rule of thumb is stick to real world scales to start with. M 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.