Leaderboard

Hello everyone! For the past 8 weeks I have been working on a Japanese Town House generator! This project was made for Breda University of Applied Sciences as a part of my Block A. This is my first Houdini project, so if anyone has feedback for me it would be greatly appreciated. The project was made from scratch as I did not find the built-in Labs building generator easy to build upon. The building generator allows you to change both the roof tiles, the lattices, the windows and the flags for different variances. The size of the building is determined by user placed bounding box, so the tool is able to generate an entire street from just a blockout! Here you can see a diagram of the basic logic used to generate the buildings, excluding the preview mesh generation. I have also made a short video using PDG to showcase different ways that the building generator can create distinct and unique outputs: WhiteOutSpedUp (1) (1).mp4 Here is the tool working in Houdini SpedUpHoudiniShowcase.mp4 And finally, the tool working in Unreal! ToolInUnreal.mp4 Thank you everyone for reading, if you have any questions please ask away!

@Ian10210123 We’re close to a solution! Plus, the first file (2D rotation) gives the same result as the simulation. I think we can now achieve this pretty easily using COP 2 in Houdini 20.5. Definitely worth a try! Just sharing some tests! I'm using something like UV coordinates in tangent space and mapping it like a texture—basically a spherical map with tiling. In DOPs, I managed to get some nice fluid-like movement going. There's always room to improve, and I just need more time to refine it, but it's getting close to the concept I'm aiming for! He also mentioned time-dependent movement in COPs, and I think this would be pretty doable in the new COPs. Just need to dial in a better smoke/fluid simulation to get the look right. Plus i forget to make gray scale and relief those or displacement but its manageable.

given two points on the long edge of the rectangle, let's say they're point 0 and 1: vector p0 = point(0, "P", 0); vector p1 = point(0, "P", 1); vector n = normalize(p0 - p1); // compute matrix that rotates the normalized edge onto +X matrix3 d = dihedral(n, {1,0,0}); // transform points with this rotation @P *= d;

Well, can't be sure but this might require more love to setup, bruteforce will not work well. Maybe something this way. liongrass_HairSim_OD_v02.hip

Hard to say without a hip. Something wrong on base level, behaviour is too weird. Might be collision geometry or so. Deformation can be done in many ways, point deform or wire deform with rotation, try to google tree setup examples. A a quick test. grass_collide_v01.hip

I'd first extrude the geometry and then fracture it. For a useful extrusion I changed the normals and deleted one face at the toes. fracture.hiplc

Seems like primnum is relative to source and impact points, you can transfer state by this. Imapct_RnD_v0003.hip

When ChatGPT, Gemini, and Bing Chat all rank Pragmatic VEX: Volume 1 as the most advanced Houdini VEX course, it’s rewarding to see it recognized for its depth and focus.

I think the author of the original post has truly mastered their role now. I enjoyed solving this simple challenge. This is for Houdini 20, but the same solution also works in Houdini 17 or 17.5. I've attached the file for anyone looking for an easy way to prepare assets for destruction or water effects collisions. This is a watertight geometry solution. vupham-version-vdb-resolution-issue-23Nov2024-v1.hipnc

Spent some time to make vellum snow simulation and Karma XPU render. Reflection volume pas Karma CPU.

@Ian10210123 Best to give you a File to investigate, Have Fun. // Find the closest point on the second input (input 1) based on the current point's position (@P) int near = nearpoint(1, @P); vector centor = point(1, "P", near); // The position of this closest point // Retrieve the 'dis_lif' attribute from the nearest point on input 1 // This attribute may represent a factor affecting influence or lifespan distance float dis_lif = point(1, "dis_lif", near); // Calculate a radian value based on the nearest point's "radians" attribute // This value is scaled by 'mdistance' and 'dis_lif' to control its effect over distance float radians = point(1, "radians", near) * chf("mdistance") * dis_lif; // Calculate a normalized distance factor (dis01) from the current point to the 'centor' // 'masdis' defines the max distance where influence fades out, fitting it into 0-1 range float dis01 = fit(distance(@P, centor), 0, chf("masdis"), 0, 1); // Map this distance factor through the 'ramp' parameter for smooth transitions // 'ramp' provides custom control over the falloff shape, helping blend and shape influences float distance = chramp("ramp", fit(distance(@P, centor), 0, radians, 0, 1)); // Calculate the direction from the current point to the 'centor' and normalize it vector dp = normalize(centor - @P); // Create a composite direction ('dir') that combines a rotational component and direct component // 'max_rot' controls rotation influence, while 'mas_dir' scales the direct movement toward 'centor' vector dir = (cross(set(-1, 0, 0), dp) * chf("max_rot") + dp * chf("mas_dir")) * distance; // Define 'divergence' for spreading or collapsing effect by using distance ramp control // 'div' ramp customizes the intensity across distances, and 'max_dif' scales the overall effect f@divergence = -chramp("div", dis01) * chf("max_dif"); // Calculate the bias for blending toward the desired direction // 'bias' is a control slider to determine how much of the influence should apply float bias = 1 - exp(-chf("bias")); // Blend the original velocity with the computed 'dir' vector based on the bias factor @vel = lerp(@vel, dir, bias); : Tuning Influence and Range with dis_lif and mdistance: dis_lif acts as a local intensity or "lifespan" modifier. Experiment with it to create different levels of influence per point. Think of it as an influence radius that you can control on a per-point basis. mdistance scales the overall effect range. Increasing it will make the effect spread further from each source point, which is helpful for soft transitions or distant influences. Customizing Falloff with Ramps: ramp and div ramps give you precise control over how the influence behaves over distance. These allow you to create linear, smooth, or even custom-curved falloffs to refine how forces dissipate or concentrate. Blending Directional Effects (max_rot and mas_dir): max_rot gives a twist effect, adding a rotational element. It’s particularly useful for creating swirl or spiral motions. mas_dir scales the direct influence toward each nearest point. You can use this to control the balance between pure rotation versus attraction to the nearest points. Controlling Divergence and Convergence: The divergence attribute is useful for controlling how particles spread or pull together. You can create effects that expand outward or collapse inward. Use max_dif to increase or decrease the overall strength of the divergence effect, helping shape behaviors like explosions, implosions, or subtle spreading. Blend with Bias: The bias parameter allows you to decide how much of the computed direction should affect the velocity, which is helpful for layering multiple effects without overwhelming the base movement. Suggested Exploration: Check Houdini Documentation on Gas Solvers: Gas solvers, such as Gas Vortex Confinement or Gas Blur, provide specialized controls for fluid and particle behavior, like adding swirl or turbulence. By combining these with custom VEX expressions, you can achieve sophisticated motion effects. Experiment with Fields: Try applying this code on different fields (density, temperature) in Houdini’s DOP context. By doing so, you can influence fluid simulations in creative ways. Explore Gas Field Wrangle Nodes: Using a Gas Field Wrangle within a solver gives you a powerful platform to apply these custom forces within smoke or fluid simulations, adding even more depth to particle or volume-based effects. Integrate with Houdini’s Attribute Noise or VOPs: Use Attribute Noise or VOP networks in combination with this VEX code to create more natural, randomized behaviors for particles and fluids, making the overall effect feel more dynamic and alive. Feel free to adjust parameters and experiment! Houdini’s flexibility and powerful VEX language mean you can customize these effects to a high degree, blending artistic intuition with technical control for unique motion patterns. ParcticleGasExOD.hipnc

@Ian10210123 COP 2

Wow! That’s it! It’s a big tip. Thank you so much

@fencer Thank you I get some ideas, but still wondering. Just want to clarify my question. tide.mp4

Maybe this should help.

Your setup is mostly already working. If you just make your 'vel' field a little bit higher res and skinnier using the Point Cloud Open in your Volume Vop and change the advection type to 'Update Velocity' in your Pop Advect by Volumes, you can get pretty close to your reference, especially if you increase the resolution of your grains a bit. You could try adding a velocity to your 'vel' that pulls towards the center of the curve if you want a slight suction effect.

Well, play with friction, attraction weight and point mass until you find nice look. And I think point mass play main role in all of this )

@Ian10210123 https://dl.acm.org/doi/10.1145/3072959.3073591

@Ian10210123 I posted that 3 years ago He use SphericalClebschMap you have files on Youtube Berlin MAth + advance MATH for Houdini to make those math Fileds , also used velocity field (FLOW DATA) , volume_musical_isomorphism1 , magneticLaplacian, AKA Clebsch_solver, therefor he gets nice transition , you can also make this with regular velocity field on 2d volume with 4 different curves , Just listen to Him on the end of that Video presentation when someone ask him HOW??? .. v@s = point(0,"s",vertexpoint(0,@vtxnum)); v@uv; #include "math.h" float theta = acos(@s.y); float phi = atan2(@s.z,@s.x); @uv.x = phi/(2*PI); @uv.y = theta/PI; Just One PIC,( imagine this with sequence Frames) just simple ex. Just search Clebsch_solver BERLIN MATH Houdini files exist Have FUN.

Looks great!

Thanks Ryew and fencer! Ryew, yes, the match size is exactly what I needed! What a silly boy I am! Thanks for the suggestion. I'll check the cookbook you linked, fencer. I'm sure there's lots of interesting stuff in there. It's been a while since I looked at it. In fact, it's been a while since I used Houdini at all! Thanks again!

Check cookbook "Equivalencies and differences" https://www.sidefx.com/docs/houdini/ref/expression_cookbook.html Try to use "-$YMAX" toP0_v01.hip

I developed this tool to make a faster and easy-to-read version of the combination of Pre-Flight Scene tool and labs archive project tool. Let's take a look of the advantage of my tool: about file viewer 1. faster loading speed than Pre-Flight Scene 2. allow to filter by node name, file name and extension 3. show absolute and relative path. 4. tell you which path is not under $HIP 5. double-click the node name and jump to the node 6. show the number of link status and the number of files with $HIP path about package 1. allow to copy Non-$HIP files to the folder where your .hip file is and replace the path to $HIP. 2. toggle of copy sequence files. Installation Create a new tool in shelf and copy the python codes. Don't be shy to give me any suggestion and feedback FileCollector.txt

There're many ways to control motion, fade over time, power, curvature, distance from a boat and so on. Ripple source is important first of all, not just one spot, but boat edge, leading area and so on, need more work here for a good result. Mask by curvature as an example. wave_v03.hip

You need a transform with a pivot to do that, there are couple ways however I just try to do what you already set up there. I store both attributes @transform matrix3 + @pivot from packed DOP IO. Then copy it over the "copy to point" result. Re-apply the matrix. File attached pig_copy_to_pts_transform_matrix_vp.hiplc

It looks like, that you have to use "restxform" and "rest" attribute. It seems, that you have to reset orient (but applying the rotation from restxform first) and the pivot should be modified by rest attribute and difference of the translation from restxform and the rest (but multplied with the inverted rotation of restxform). I'm pretty sure, that there must be a simpler way but I don't know ... pig_mod.hipnc

Hello I try to transfer a scene from houdini to blender (sadly for the moment i can't do what i want with karma). For the exemple I made a simple scene, in wich i try to transfer points information to use it in blender geo node (position, pscale and orient). For position and pscale, everything works fine, but for the orientation, i can't figure it out how to do it... I try this entagma tutorial, but it doesn't work properly on my scene for the orientation... Any ideas ? (here you can find the scene files : https://philippe_petitpas.fromsmash.com/H18ziH4BtS-ct

fix

This is very cool! After many years of Houdini, I'm ashamed to say I never knew about this! Thank you.

I found the problem At first, when i exported the peach model which i modeled in blender, i didn't "apply" the transformation... In case, if anybody need to transfer the orient to blender, that's how i did it :

Anim In Array Olivier Jeannel (Vimeo). animAr.hiplc

@Damn_daniel masking in Height values float value = @height % chf("spacing"); float min = chf("spacing") - (chf("line_width") / 2); float max = 0 + (chf("line_width") / 2); if(min <= value || value <= max) { @mask = 1; } SrBugArg.hiplc

without reseeding you can just delete points having a higher pointnumber than a certain value if (@ptnum > 1000){ removepoint(0, @ptnum); } i've attached a sample file with some different approaches and some sticky notes as well, i hope that is helpful you can get the velocity field out of a flip sim with an import dop fields, for one solution i fed it into a solver and for another one a setup a popsim with pop advect by volumes. both approaches use reseeding. i also tested a quick setup with reseeding turned off. hope this helps! flipvel.hipnc

Building on chapter 14 of the particle planet tutorial series posted on SideFX, I extended the scene to detect when the particles escape the surface. After they age two seconds, a popseek draws them back to the surface. White lines never leave the surface, lavender are free roaming, and yellow are free roaming aged particles heading back to the surface. There are lots of noise control to vary the speed and timing. ap_tut_particles_around_sphere_090323.hiplc

Check an idea. odForce_growthEffect_v02.hiplc

You can replicate it using point cloud, same process to Anto's explaination. Get close points and output averge position wiht pcfilter, higher max points higher smoothness. .

A simple mycelium growth simulation that works in 2D and 3D: It constantly adds points to the outer edge of point clouds (1) and favours doing so towards food (2). 1: Direction away from surrounding points: float radius = chf('radius'); int pts_max = chi('max_points'); int handle = pcopen(0, 'P', v@P, radius, pts_max); vector pos_avg = pcfilter(handle, 'P'); vector dir_avg = v@P - pos_avg; v@N = dir_avg; 2: Weighting probabilities and adding points: int pt_near = nearpoint(1, v@P); vector pos_near = point(1, 'P', pt_near); float prox = distance(v@P, pos_near); float weight_prox = fit(prox, prox_min, prox_max, 0.0, 1.0); float weight_rim = fit(length(v@N), rim_min, rim_max, 1.0, 0.0); float weight = dens * (weight_rim + weight_prox); if(nrandom('mersenne') > weight){ vector pos_add = v@P + normalize(v@N) * offset; int pt_add = addpoint(0, pos_add); addprim(0, 'polyline', i@ptnum, pt_add); } mycelium.hiplc

Hi Tom ! The issue you are having is because the vellum geometry passed to the solver is actually frozen at the creation frame, and does not update the animated attributes. What you need is to have access to that animated geo So in your popwrangle, you can specify the input 2 to be your animated geometry (it can be anything from your maskfromgeometry to the vellumstructs), and then refer to that geo instead of the dop geometry Hope that helps ! vellum_falloff.hiplc

easy peasy Note: you can exceed range 0-1...above 1 it will just proportionately overshoot....or undershoot too. vu_lerp.hiplc

A long time ago when working with 3d printing on Zcorp printers I use Materialise and Geomagic..Just one click.(In most cases )

Hi, I'm trying to recreate this very cool tyflow setup in Houdini. In the description he explain's how he is using dynamic PhysX rigidbodies and using attraction forces to move them, looks really cool. I've had a go with some success using the crowd system in Houdini with RBDs and a pop attract. Hip files here worms_03_GC.hip worms_05_GC.hip worms03.avi worms05.avi The idea works but I can't get a nice behaviour using the pop attract, has anyone got any ideas what might work better with the particle forces? Thanks worms_05_GC.hip

This seemed like an interesting VEX challenge and I wanted to give it a go Searched for a solution and found this https://stackoverflow.com/questions/14066933/direct-way-of-computing-clockwise-angle-between-2-vectors Used that in a wrangle node to determine if a corner is convex or concave and added some other stuff and so far it seems to work. For now, the solution works only for closed curves and treats them as 2D shapes on a XZ plane. compute_normals.hiplc

to automate this over x containers is outside the scope of this but look into using a wedge rop, it is not hard this doesn't support collisions as it is procedural, if your really tasked with collisions on 1000 containers with external anim that is a big effect and not a easy to avoid the cost. but it can be done with spring to glue constraints and point deform. A cheap sops collisions could be multiplying invertnormal you will find by distance to collision object. Damage.hip

@Yon Thanks alot! I stumbled upon this explanation and I think it is the most in-depth quick guide on the abc's of Pyro (That is, without getting too technical):

By default the pyro/smoke solvers will clear any density which enters the collision volume in a timestep. I would double check your collision volume is interpolating sub frame and also that you are feeding the velocity from the collision object into the sim. On top of that depending on the speed of your object you may need to increase the substeps on the solver to help account for it. Also on the smoke/pyro solver if you go to the 'Advanced/Collisions' tab you can toggle off 'Correct Collisions' which will stop the solver from deleting any density which gets inside the collision somehow. Hope this is helps Cheers Jake.

Look at the help file for the Scatter SOP: http://www.sidefx.com/docs/houdini15.0/nodes/sop/scatter The Tip section right at the top describes three different scenarios to scatter fixed point positions on deforming geometry including an example file. Very fast and you only evaluate the Scatter SOP from the frame you use in the Time Shift SOP. Btw rarely do you want to lock operators to get a rest frame in an input deforming archive. Always use the Time Shift SOP and 9 times out of 10 you are setting it to frame 1 or the start frame of the sequence which is $FSTART (in case you are dealing with simulations that start at frame 1001 for example). See the example file as to how to take in deforming unpacked geometry (from your alembic archive) and use an Ends SOP set to unroll then scatter and stick with the Attribute Transfer SOP using the recipe indicated in the help. See the attached example file for scattering on deforming edges. stick_scatter_points_on_deforming_geometry.hip